k

.github/pull_request_template.md

@@ -1,15 +1,11 @@
 ## Description
-
 [Provide a brief description of the changes in this PR]
 
-## How Has This Been Tested?
 
+## How Has This Been Tested?
 [Describe the tests you ran to verify your changes]
 
+
 ## Backporting (check the box to trigger backport action)
-
 Note: You have to check that the action passes, otherwise resolve the conflicts manually and tag the patches.
-
 - [ ] This PR should be backported (make sure to check that the backport attempt succeeds)
-- [ ] I have included a link to a Linear ticket in my description.
-- [ ] [Optional] Override Linear Check

.github/workflows/docker-build-push-model-server-container-on-tag.yml

@@ -118,6 +118,6 @@ jobs:
           TRIVY_DB_REPOSITORY: "public.ecr.aws/aquasecurity/trivy-db:2"
           TRIVY_JAVA_DB_REPOSITORY: "public.ecr.aws/aquasecurity/trivy-java-db:1"
         with:
-          image-ref: docker.io/${{ env.REGISTRY_IMAGE }}:${{ github.ref_name }}
+          image-ref: docker.io/onyxdotapp/onyx-model-server:${{ github.ref_name }}
           severity: "CRITICAL,HIGH"
           timeout: "10m"

.github/workflows/pr-linear-check.yml

@@ -1,29 +0,0 @@
-name: Ensure PR references Linear
-
-on:
-  pull_request:
-    types: [opened, edited, reopened, synchronize]
-
-jobs:
-  linear-check:
-    runs-on: ubuntu-latest
-    steps:
-      - name: Check PR body for Linear link or override
-        run: |
-          PR_BODY="${{ github.event.pull_request.body }}"
-
-          # Looking for "https://linear.app" in the body
-          if echo "$PR_BODY" | grep -qE "https://linear\.app"; then
-            echo "Found a Linear link. Check passed."
-            exit 0
-          fi
-
-          # Looking for a checked override: "[x] Override Linear Check"
-          if echo "$PR_BODY" | grep -q "\[x\].*Override Linear Check"; then
-            echo "Override box is checked. Check passed."
-            exit 0
-          fi
-
-          # Otherwise, fail the run
-          echo "No Linear link or override found in the PR description."
-          exit 1

.vscode/env_template.txt

@@ -5,8 +5,6 @@
 # For local dev, often user Authentication is not needed
 AUTH_TYPE=disabled
 
-# Skip warm up for dev
-SKIP_WARM_UP=True
 
 # Always keep these on for Dev
 # Logs all model prompts to stdout
@@ -29,7 +27,6 @@ REQUIRE_EMAIL_VERIFICATION=False
 
 # Set these so if you wipe the DB, you don't end up having to go through the UI every time
 GEN_AI_API_KEY=<REPLACE THIS>
-OPENAI_API_KEY=<REPLACE THIS>
 # If answer quality isn't important for dev, use gpt-4o-mini since it's cheaper
 GEN_AI_MODEL_VERSION=gpt-4o
 FAST_GEN_AI_MODEL_VERSION=gpt-4o

.vscode/launch.template.jsonc

@@ -28,7 +28,6 @@
 		  		"Celery heavy", 
 		  		"Celery indexing", 
 		  		"Celery beat",
-                "Celery monitoring",
             ],
 			"presentation": {
 				 "group": "1",
@@ -52,8 +51,7 @@
 		  		"Celery light", 
 		  		"Celery heavy", 
 		  		"Celery indexing", 
-		  		"Celery beat",
-                "Celery monitoring",
+		  		"Celery beat"
 		  	],
 			"presentation": {
 				 "group": "1",
@@ -271,31 +269,6 @@
 			 },
             "consoleTitle": "Celery indexing Console"
         },
-        {
-            "name": "Celery monitoring",
-            "type": "debugpy",
-            "request": "launch",
-            "module": "celery",
-            "cwd": "${workspaceFolder}/backend",
-            "envFile": "${workspaceFolder}/.vscode/.env",
-            "env": {},
-            "args": [
-                "-A",
-                "onyx.background.celery.versioned_apps.monitoring",
-                "worker",
-                "--pool=solo",
-                "--concurrency=1",
-                "--prefetch-multiplier=1",
-                "--loglevel=INFO",
-                "--hostname=monitoring@%n",
-                "-Q",
-                "monitoring",
-            ],
-            "presentation": {
-				 "group": "2",
-			 },
-            "consoleTitle": "Celery monitoring Console"
-        },
         {
             "name": "Celery beat",
             "type": "debugpy",
@@ -382,20 +355,5 @@
                 "PYTHONPATH": "."
             },
         },
-        {
-            "name": "Install Python Requirements",
-            "type": "node",
-            "request": "launch",
-            "runtimeExecutable": "bash",
-            "runtimeArgs": [
-                "-c",
-                "pip install -r backend/requirements/default.txt && pip install -r backend/requirements/dev.txt && pip install -r backend/requirements/ee.txt && pip install -r backend/requirements/model_server.txt"
-            ],
-            "cwd": "${workspaceFolder}",
-            "console": "integratedTerminal",
-            "presentation": {
-                 "group": "3"
-            }
-        },
     ]
 }

CONTRIBUTING.md

@@ -12,10 +12,6 @@ As an open source project in a rapidly changing space, we welcome all contributi
 
 The [GitHub Issues](https://github.com/onyx-dot-app/onyx/issues) page is a great place to start for contribution ideas.
 
-To ensure that your contribution is aligned with the project's direction, please reach out to Hagen (or any other maintainer) on the Onyx team
-via [Slack](https://join.slack.com/t/onyx-dot-app/shared_invite/zt-2twesxdr6-5iQitKZQpgq~hYIZ~dv3KA) /
-[Discord](https://discord.gg/TDJ59cGV2X) or [email](mailto:founders@onyx.app).
-
 Issues that have been explicitly approved by the maintainers (aligned with the direction of the project)
 will be marked with the `approved by maintainers` label.
 Issues marked `good first issue` are an especially great place to start.
@@ -27,8 +23,8 @@ If you have a new/different contribution in mind, we'd love to hear about it!
 Your input is vital to making sure that Onyx moves in the right direction.
 Before starting on implementation, please raise a GitHub issue.
 
-Also, always feel free to message the founders (Chris Weaver / Yuhong Sun) on
-[Slack](https://join.slack.com/t/onyx-dot-app/shared_invite/zt-2twesxdr6-5iQitKZQpgq~hYIZ~dv3KA) /
+And always feel free to message us (Chris Weaver / Yuhong Sun) on
+[Slack](https://join.slack.com/t/danswer/shared_invite/zt-1w76msxmd-HJHLe3KNFIAIzk_0dSOKaQ) /
 [Discord](https://discord.gg/TDJ59cGV2X) directly about anything at all.
 
 ### Contributing Code
@@ -46,7 +42,7 @@ Our goal is to make contributing as easy as possible. If you run into any issues
 That way we can help future contributors and users can avoid the same issue.
 
 We also have support channels and generally interesting discussions on our
-[Slack](https://join.slack.com/t/onyx-dot-app/shared_invite/zt-2twesxdr6-5iQitKZQpgq~hYIZ~dv3KA)
+[Slack](https://join.slack.com/t/danswer/shared_invite/zt-1w76msxmd-HJHLe3KNFIAIzk_0dSOKaQ)
 and
 [Discord](https://discord.gg/TDJ59cGV2X).
 
@@ -127,47 +123,7 @@ Once the above is done, navigate to `onyx/web` run:
 npm i
 ```
 
-## Formatting and Linting
-
-### Backend
-
-For the backend, you'll need to setup pre-commit hooks (black / reorder-python-imports).
-First, install pre-commit (if you don't have it already) following the instructions
-[here](https://pre-commit.com/#installation).
-
-With the virtual environment active, install the pre-commit library with:
-
-```bash
-pip install pre-commit
-```
-
-Then, from the `onyx/backend` directory, run:
-
-```bash
-pre-commit install
-```
-
-Additionally, we use `mypy` for static type checking.
-Onyx is fully type-annotated, and we want to keep it that way!
-To run the mypy checks manually, run `python -m mypy .` from the `onyx/backend` directory.
-
-### Web
-
-We use `prettier` for formatting. The desired version (2.8.8) will be installed via a `npm i` from the `onyx/web` directory.
-To run the formatter, use `npx prettier --write .` from the `onyx/web` directory.
-Please double check that prettier passes before creating a pull request.
-
-# Running the application for development
-
-## Developing using VSCode Debugger (recommended)
-
-We highly recommend using VSCode debugger for development.
-See [CONTRIBUTING_VSCODE.md](./CONTRIBUTING_VSCODE.md) for more details.
-
-Otherwise, you can follow the instructions below to run the application for development.
-
-## Manually running the application for development
-### Docker containers for external software
+#### Docker containers for external software
 
 You will need Docker installed to run these containers.
 
@@ -179,7 +135,7 @@ docker compose -f docker-compose.dev.yml -p onyx-stack up -d index relational_db
 
 (index refers to Vespa, relational_db refers to Postgres, and cache refers to Redis)
 
-### Running Onyx locally
+#### Running Onyx locally
 
 To start the frontend, navigate to `onyx/web` and run:
 
@@ -267,6 +223,35 @@ If you want to make changes to Onyx and run those changes in Docker, you can als
 docker compose -f docker-compose.dev.yml -p onyx-stack up -d --build
 ```
 
+### Formatting and Linting
+
+#### Backend
+
+For the backend, you'll need to setup pre-commit hooks (black / reorder-python-imports).
+First, install pre-commit (if you don't have it already) following the instructions
+[here](https://pre-commit.com/#installation).
+
+With the virtual environment active, install the pre-commit library with:
+
+```bash
+pip install pre-commit
+```
+
+Then, from the `onyx/backend` directory, run:
+
+```bash
+pre-commit install
+```
+
+Additionally, we use `mypy` for static type checking.
+Onyx is fully type-annotated, and we want to keep it that way!
+To run the mypy checks manually, run `python -m mypy .` from the `onyx/backend` directory.
+
+#### Web
+
+We use `prettier` for formatting. The desired version (2.8.8) will be installed via a `npm i` from the `onyx/web` directory.
+To run the formatter, use `npx prettier --write .` from the `onyx/web` directory.
+Please double check that prettier passes before creating a pull request.
 
 ### Release Process
 

CONTRIBUTING_VSCODE.md

@@ -1,30 +0,0 @@
-# VSCode Debugging Setup
-
-This guide explains how to set up and use VSCode's debugging capabilities with this project.
-
-## Initial Setup
-
-1. **Environment Setup**:
-   - Copy `.vscode/.env.template` to `.vscode/.env`
-   - Fill in the necessary environment variables in `.vscode/.env`
-2. **launch.json**:
-   - Copy `.vscode/launch.template.jsonc` to `.vscode/launch.json`
-
-## Using the Debugger
-
-Before starting, make sure the Docker Daemon is running.
-
-1. Open the Debug view in VSCode (Cmd+Shift+D on macOS)
-2. From the dropdown at the top, select "Clear and Restart External Volumes and Containers" and press the green play button
-3. From the dropdown at the top, select "Run All Onyx Services" and press the green play button
-4. CD into web, run "npm i" followed by npm run dev.
-5. Now, you can navigate to onyx in your browser (default is http://localhost:3000) and start using the app
-6. You can set breakpoints by clicking to the left of line numbers to help debug while the app is running
-7. Use the debug toolbar to step through code, inspect variables, etc.
-
-## Features
-
-- Hot reload is enabled for the web server and API servers
-- Python debugging is configured with debugpy
-- Environment variables are loaded from `.vscode/.env`
-- Console output is organized in the integrated terminal with labeled tabs

backend/alembic/versions/027381bce97c_add_shortcut_option_for_users.py

@@ -1,29 +0,0 @@
-"""add shortcut option for users
-
-Revision ID: 027381bce97c
-Revises: 6fc7886d665d
-Create Date: 2025-01-14 12:14:00.814390
-
-"""
-from alembic import op
-import sqlalchemy as sa
-
-
-# revision identifiers, used by Alembic.
-revision = "027381bce97c"
-down_revision = "6fc7886d665d"
-branch_labels = None
-depends_on = None
-
-
-def upgrade() -> None:
-    op.add_column(
-        "user",
-        sa.Column(
-            "shortcut_enabled", sa.Boolean(), nullable=False, server_default="false"
-        ),
-    )
-
-
-def downgrade() -> None:
-    op.drop_column("user", "shortcut_enabled")

backend/alembic/versions/0f7ff6d75b57_add_index_to_index_attempt_time_created.py

@@ -1,36 +0,0 @@
-"""add index to index_attempt.time_created
-
-Revision ID: 0f7ff6d75b57
-Revises: 369644546676
-Create Date: 2025-01-10 14:01:14.067144
-
-"""
-from alembic import op
-
-# revision identifiers, used by Alembic.
-revision = "0f7ff6d75b57"
-down_revision = "fec3db967bf7"
-branch_labels: None = None
-depends_on: None = None
-
-
-def upgrade() -> None:
-    op.create_index(
-        op.f("ix_index_attempt_status"),
-        "index_attempt",
-        ["status"],
-        unique=False,
-    )
-
-    op.create_index(
-        op.f("ix_index_attempt_time_created"),
-        "index_attempt",
-        ["time_created"],
-        unique=False,
-    )
-
-
-def downgrade() -> None:
-    op.drop_index(op.f("ix_index_attempt_time_created"), table_name="index_attempt")
-
-    op.drop_index(op.f("ix_index_attempt_status"), table_name="index_attempt")

backend/alembic/versions/369644546676_add_composite_index_for_index_attempt_.py

@@ -1,35 +0,0 @@
-"""add composite index for index attempt time updated
-
-Revision ID: 369644546676
-Revises: 2955778aa44c
-Create Date: 2025-01-08 15:38:17.224380
-
-"""
-from alembic import op
-from sqlalchemy import text
-
-# revision identifiers, used by Alembic.
-revision = "369644546676"
-down_revision = "2955778aa44c"
-branch_labels: None = None
-depends_on: None = None
-
-
-def upgrade() -> None:
-    op.create_index(
-        "ix_index_attempt_ccpair_search_settings_time_updated",
-        "index_attempt",
-        [
-            "connector_credential_pair_id",
-            "search_settings_id",
-            text("time_updated DESC"),
-        ],
-        unique=False,
-    )
-
-
-def downgrade() -> None:
-    op.drop_index(
-        "ix_index_attempt_ccpair_search_settings_time_updated",
-        table_name="index_attempt",
-    )

backend/alembic/versions/3c6531f32351_add_back_input_prompts.py

@@ -1,59 +0,0 @@
-"""add back input prompts
-
-Revision ID: 3c6531f32351
-Revises: aeda5f2df4f6
-Create Date: 2025-01-13 12:49:51.705235
-
-"""
-from alembic import op
-import sqlalchemy as sa
-import fastapi_users_db_sqlalchemy
-
-# revision identifiers, used by Alembic.
-revision = "3c6531f32351"
-down_revision = "aeda5f2df4f6"
-branch_labels = None
-depends_on = None
-
-
-def upgrade() -> None:
-    op.create_table(
-        "inputprompt",
-        sa.Column("id", sa.Integer(), autoincrement=True, nullable=False),
-        sa.Column("prompt", sa.String(), nullable=False),
-        sa.Column("content", sa.String(), nullable=False),
-        sa.Column("active", sa.Boolean(), nullable=False),
-        sa.Column("is_public", sa.Boolean(), nullable=False),
-        sa.Column(
-            "user_id",
-            fastapi_users_db_sqlalchemy.generics.GUID(),
-            nullable=True,
-        ),
-        sa.ForeignKeyConstraint(
-            ["user_id"],
-            ["user.id"],
-        ),
-        sa.PrimaryKeyConstraint("id"),
-    )
-    op.create_table(
-        "inputprompt__user",
-        sa.Column("input_prompt_id", sa.Integer(), nullable=False),
-        sa.Column(
-            "user_id", fastapi_users_db_sqlalchemy.generics.GUID(), nullable=False
-        ),
-        sa.Column("disabled", sa.Boolean(), nullable=False, default=False),
-        sa.ForeignKeyConstraint(
-            ["input_prompt_id"],
-            ["inputprompt.id"],
-        ),
-        sa.ForeignKeyConstraint(
-            ["user_id"],
-            ["user.id"],
-        ),
-        sa.PrimaryKeyConstraint("input_prompt_id", "user_id"),
-    )
-
-
-def downgrade() -> None:
-    op.drop_table("inputprompt__user")
-    op.drop_table("inputprompt")

backend/alembic/versions/47e5bef3a1d7_add_persona_categories.py

@@ -40,6 +40,6 @@ def upgrade() -> None:
 
 
 def downgrade() -> None:
-    op.drop_constraint("persona_category_id_fkey", "persona", type_="foreignkey")
+    op.drop_constraint("fk_persona_category", "persona", type_="foreignkey")
     op.drop_column("persona", "category_id")
     op.drop_table("persona_category")

backend/alembic/versions/6fc7886d665d_make_categories_labels_and_many_to_many.py

@@ -1,80 +0,0 @@
-"""make categories labels and many to many
-
-Revision ID: 6fc7886d665d
-Revises: 3c6531f32351
-Create Date: 2025-01-13 18:12:18.029112
-
-"""
-from alembic import op
-import sqlalchemy as sa
-
-
-# revision identifiers, used by Alembic.
-revision = "6fc7886d665d"
-down_revision = "3c6531f32351"
-branch_labels = None
-depends_on = None
-
-
-def upgrade() -> None:
-    # Rename persona_category table to persona_label
-    op.rename_table("persona_category", "persona_label")
-
-    # Create the new association table
-    op.create_table(
-        "persona__persona_label",
-        sa.Column("persona_id", sa.Integer(), nullable=False),
-        sa.Column("persona_label_id", sa.Integer(), nullable=False),
-        sa.ForeignKeyConstraint(
-            ["persona_id"],
-            ["persona.id"],
-        ),
-        sa.ForeignKeyConstraint(
-            ["persona_label_id"],
-            ["persona_label.id"],
-            ondelete="CASCADE",
-        ),
-        sa.PrimaryKeyConstraint("persona_id", "persona_label_id"),
-    )
-
-    # Copy existing relationships to the new table
-    op.execute(
-        """
-        INSERT INTO persona__persona_label (persona_id, persona_label_id)
-        SELECT id, category_id FROM persona WHERE category_id IS NOT NULL
-    """
-    )
-
-    # Remove the old category_id column from persona table
-    op.drop_column("persona", "category_id")
-
-
-def downgrade() -> None:
-    # Rename persona_label table back to persona_category
-    op.rename_table("persona_label", "persona_category")
-
-    # Add back the category_id column to persona table
-    op.add_column("persona", sa.Column("category_id", sa.Integer(), nullable=True))
-    op.create_foreign_key(
-        "persona_category_id_fkey",
-        "persona",
-        "persona_category",
-        ["category_id"],
-        ["id"],
-    )
-
-    # Copy the first label relationship back to the persona table
-    op.execute(
-        """
-        UPDATE persona
-        SET category_id = (
-            SELECT persona_label_id
-            FROM persona__persona_label
-            WHERE persona__persona_label.persona_id = persona.id
-            LIMIT 1
-        )
-    """
-    )
-
-    # Drop the association table
-    op.drop_table("persona__persona_label")

backend/alembic/versions/97dbb53fa8c8_add_syncrecord.py

@@ -1,72 +0,0 @@
-"""Add SyncRecord
-
-Revision ID: 97dbb53fa8c8
-Revises: 369644546676
-Create Date: 2025-01-11 19:39:50.426302
-
-"""
-from alembic import op
-import sqlalchemy as sa
-
-# revision identifiers, used by Alembic.
-revision = "97dbb53fa8c8"
-down_revision = "be2ab2aa50ee"
-branch_labels = None
-depends_on = None
-
-
-def upgrade() -> None:
-    op.create_table(
-        "sync_record",
-        sa.Column("id", sa.Integer(), nullable=False),
-        sa.Column("entity_id", sa.Integer(), nullable=False),
-        sa.Column(
-            "sync_type",
-            sa.Enum(
-                "DOCUMENT_SET",
-                "USER_GROUP",
-                "CONNECTOR_DELETION",
-                name="synctype",
-                native_enum=False,
-                length=40,
-            ),
-            nullable=False,
-        ),
-        sa.Column(
-            "sync_status",
-            sa.Enum(
-                "IN_PROGRESS",
-                "SUCCESS",
-                "FAILED",
-                "CANCELED",
-                name="syncstatus",
-                native_enum=False,
-                length=40,
-            ),
-            nullable=False,
-        ),
-        sa.Column("num_docs_synced", sa.Integer(), nullable=False),
-        sa.Column("sync_start_time", sa.DateTime(timezone=True), nullable=False),
-        sa.Column("sync_end_time", sa.DateTime(timezone=True), nullable=True),
-        sa.PrimaryKeyConstraint("id"),
-    )
-
-    # Add index for fetch_latest_sync_record query
-    op.create_index(
-        "ix_sync_record_entity_id_sync_type_sync_start_time",
-        "sync_record",
-        ["entity_id", "sync_type", "sync_start_time"],
-    )
-
-    # Add index for cleanup_sync_records query
-    op.create_index(
-        "ix_sync_record_entity_id_sync_type_sync_status",
-        "sync_record",
-        ["entity_id", "sync_type", "sync_status"],
-    )
-
-
-def downgrade() -> None:
-    op.drop_index("ix_sync_record_entity_id_sync_type_sync_status")
-    op.drop_index("ix_sync_record_entity_id_sync_type_sync_start_time")
-    op.drop_table("sync_record")

backend/alembic/versions/aeda5f2df4f6_add_pinned_assistants.py

@@ -1,27 +0,0 @@
-"""add pinned assistants
-
-Revision ID: aeda5f2df4f6
-Revises: c5eae4a75a1b
-Create Date: 2025-01-09 16:04:10.770636
-
-"""
-from alembic import op
-import sqlalchemy as sa
-from sqlalchemy.dialects import postgresql
-
-# revision identifiers, used by Alembic.
-revision = "aeda5f2df4f6"
-down_revision = "c5eae4a75a1b"
-branch_labels = None
-depends_on = None
-
-
-def upgrade() -> None:
-    op.add_column(
-        "user", sa.Column("pinned_assistants", postgresql.JSONB(), nullable=True)
-    )
-    op.execute('UPDATE "user" SET pinned_assistants = chosen_assistants')
-
-
-def downgrade() -> None:
-    op.drop_column("user", "pinned_assistants")

backend/alembic/versions/be2ab2aa50ee_fix_capitalization.py

@@ -1,38 +0,0 @@
-"""fix_capitalization
-
-Revision ID: be2ab2aa50ee
-Revises: 369644546676
-Create Date: 2025-01-10 13:13:26.228960
-
-"""
-from alembic import op
-
-# revision identifiers, used by Alembic.
-revision = "be2ab2aa50ee"
-down_revision = "369644546676"
-branch_labels = None
-depends_on = None
-
-
-def upgrade() -> None:
-    op.execute(
-        """
-        UPDATE document
-        SET
-            external_user_group_ids = ARRAY(
-                SELECT LOWER(unnest(external_user_group_ids))
-            ),
-            last_modified = NOW()
-        WHERE
-            external_user_group_ids IS NOT NULL
-            AND external_user_group_ids::text[] <> ARRAY(
-                SELECT LOWER(unnest(external_user_group_ids))
-            )::text[]
-    """
-    )
-
-
-def downgrade() -> None:
-    # No way to cleanly persist the bad state through an upgrade/downgrade
-    # cycle, so we just pass
-    pass

backend/alembic/versions/c5eae4a75a1b_add_chat_message__standard_answer_table.py

@@ -1,36 +0,0 @@
-"""Add chat_message__standard_answer table
-
-Revision ID: c5eae4a75a1b
-Revises: 0f7ff6d75b57
-Create Date: 2025-01-15 14:08:49.688998
-
-"""
-from alembic import op
-import sqlalchemy as sa
-
-# revision identifiers, used by Alembic.
-revision = "c5eae4a75a1b"
-down_revision = "0f7ff6d75b57"
-branch_labels = None
-depends_on = None
-
-
-def upgrade() -> None:
-    op.create_table(
-        "chat_message__standard_answer",
-        sa.Column("chat_message_id", sa.Integer(), nullable=False),
-        sa.Column("standard_answer_id", sa.Integer(), nullable=False),
-        sa.ForeignKeyConstraint(
-            ["chat_message_id"],
-            ["chat_message.id"],
-        ),
-        sa.ForeignKeyConstraint(
-            ["standard_answer_id"],
-            ["standard_answer.id"],
-        ),
-        sa.PrimaryKeyConstraint("chat_message_id", "standard_answer_id"),
-    )
-
-
-def downgrade() -> None:
-    op.drop_table("chat_message__standard_answer")

backend/alembic/versions/c7bf5721733e_add_has_been_indexed_to_.py

@@ -1,48 +0,0 @@
-"""Add has_been_indexed to DocumentByConnectorCredentialPair
-
-Revision ID: c7bf5721733e
-Revises: fec3db967bf7
-Create Date: 2025-01-13 12:39:05.831693
-
-"""
-from alembic import op
-import sqlalchemy as sa
-
-# revision identifiers, used by Alembic.
-revision = "c7bf5721733e"
-down_revision = "027381bce97c"
-branch_labels = None
-depends_on = None
-
-
-def upgrade() -> None:
-    # assume all existing rows have been indexed, no better approach
-    op.add_column(
-        "document_by_connector_credential_pair",
-        sa.Column("has_been_indexed", sa.Boolean(), nullable=True),
-    )
-    op.execute(
-        "UPDATE document_by_connector_credential_pair SET has_been_indexed = TRUE"
-    )
-    op.alter_column(
-        "document_by_connector_credential_pair",
-        "has_been_indexed",
-        nullable=False,
-    )
-
-    # Add index to optimize get_document_counts_for_cc_pairs query pattern
-    op.create_index(
-        "idx_document_cc_pair_counts",
-        "document_by_connector_credential_pair",
-        ["connector_id", "credential_id", "has_been_indexed"],
-        unique=False,
-    )
-
-
-def downgrade() -> None:
-    # Remove the index first before removing the column
-    op.drop_index(
-        "idx_document_cc_pair_counts",
-        table_name="document_by_connector_credential_pair",
-    )
-    op.drop_column("document_by_connector_credential_pair", "has_been_indexed")

backend/alembic/versions/fec3db967bf7_add_time_updated_to_usergroup_and_.py

@@ -1,41 +0,0 @@
-"""Add time_updated to UserGroup and DocumentSet
-
-Revision ID: fec3db967bf7
-Revises: 97dbb53fa8c8
-Create Date: 2025-01-12 15:49:02.289100
-
-"""
-from alembic import op
-import sqlalchemy as sa
-
-# revision identifiers, used by Alembic.
-revision = "fec3db967bf7"
-down_revision = "97dbb53fa8c8"
-branch_labels = None
-depends_on = None
-
-
-def upgrade() -> None:
-    op.add_column(
-        "document_set",
-        sa.Column(
-            "time_last_modified_by_user",
-            sa.DateTime(timezone=True),
-            nullable=False,
-            server_default=sa.func.now(),
-        ),
-    )
-    op.add_column(
-        "user_group",
-        sa.Column(
-            "time_last_modified_by_user",
-            sa.DateTime(timezone=True),
-            nullable=False,
-            server_default=sa.func.now(),
-        ),
-    )
-
-
-def downgrade() -> None:
-    op.drop_column("user_group", "time_last_modified_by_user")
-    op.drop_column("document_set", "time_last_modified_by_user")

backend/alembic_tenants/versions/a4f6ee863c47_mapping_for_anonymous_user_path.py

@@ -1,31 +0,0 @@
-"""mapping for anonymous user path
-
-Revision ID: a4f6ee863c47
-Revises: 14a83a331951
-Create Date: 2025-01-04 14:16:58.697451
-
-"""
-import sqlalchemy as sa
-
-from alembic import op
-
-
-# revision identifiers, used by Alembic.
-revision = "a4f6ee863c47"
-down_revision = "14a83a331951"
-branch_labels = None
-depends_on = None
-
-
-def upgrade() -> None:
-    op.create_table(
-        "tenant_anonymous_user_path",
-        sa.Column("tenant_id", sa.String(), primary_key=True, nullable=False),
-        sa.Column("anonymous_user_path", sa.String(), nullable=False),
-        sa.PrimaryKeyConstraint("tenant_id"),
-        sa.UniqueConstraint("anonymous_user_path"),
-    )
-
-
-def downgrade() -> None:
-    op.drop_table("tenant_anonymous_user_path")

backend/ee/onyx/access/access.py

@@ -3,10 +3,6 @@ from sqlalchemy.orm import Session
 from ee.onyx.db.external_perm import fetch_external_groups_for_user
 from ee.onyx.db.user_group import fetch_user_groups_for_documents
 from ee.onyx.db.user_group import fetch_user_groups_for_user
-from ee.onyx.external_permissions.post_query_censoring import (
-    DOC_SOURCE_TO_CHUNK_CENSORING_FUNCTION,
-)
-from ee.onyx.external_permissions.sync_params import DOC_PERMISSIONS_FUNC_MAP
 from onyx.access.access import (
     _get_access_for_documents as get_access_for_documents_without_groups,
 )
@@ -14,7 +10,6 @@ from onyx.access.access import _get_acl_for_user as get_acl_for_user_without_gro
 from onyx.access.models import DocumentAccess
 from onyx.access.utils import prefix_external_group
 from onyx.access.utils import prefix_user_group
-from onyx.db.document import get_document_sources
 from onyx.db.document import get_documents_by_ids
 from onyx.db.models import User
 
@@ -57,20 +52,9 @@ def _get_access_for_documents(
     )
     doc_id_map = {doc.id: doc for doc in documents}
 
-    # Get all sources in one batch
-    doc_id_to_source_map = get_document_sources(
-        db_session=db_session,
-        document_ids=document_ids,
-    )
-
     access_map = {}
     for document_id, non_ee_access in non_ee_access_dict.items():
         document = doc_id_map[document_id]
-        source = doc_id_to_source_map.get(document_id)
-        is_only_censored = (
-            source in DOC_SOURCE_TO_CHUNK_CENSORING_FUNCTION
-            and source not in DOC_PERMISSIONS_FUNC_MAP
-        )
 
         ext_u_emails = (
             set(document.external_user_emails)
@@ -86,11 +70,7 @@ def _get_access_for_documents(
 
         # If the document is determined to be "public" externally (through a SYNC connector)
         # then it's given the same access level as if it were marked public within Onyx
-        # If its censored, then it's public anywhere during the search and then permissions are
-        # applied after the search
-        is_public_anywhere = (
-            document.is_public or non_ee_access.is_public or is_only_censored
-        )
+        is_public_anywhere = document.is_public or non_ee_access.is_public
 
         # To avoid collisions of group namings between connectors, they need to be prefixed
         access_map[document_id] = DocumentAccess(

backend/ee/onyx/auth/users.py

@@ -1,7 +1,5 @@
-from datetime import datetime
 from functools import lru_cache
 
-import jwt
 import requests
 from fastapi import Depends
 from fastapi import HTTPException
@@ -22,7 +20,6 @@ from ee.onyx.server.seeding import get_seed_config
 from ee.onyx.utils.secrets import extract_hashed_cookie
 from onyx.auth.users import current_admin_user
 from onyx.configs.app_configs import AUTH_TYPE
-from onyx.configs.app_configs import USER_AUTH_SECRET
 from onyx.configs.constants import AuthType
 from onyx.db.models import User
 from onyx.utils.logger import setup_logger
@@ -121,17 +118,3 @@ async def current_cloud_superuser(
             detail="Access denied. User must be a cloud superuser to perform this action.",
         )
     return user
-
-
-def generate_anonymous_user_jwt_token(tenant_id: str) -> str:
-    payload = {
-        "tenant_id": tenant_id,
-        # Token does not expire
-        "iat": datetime.utcnow(),  # Issued at time
-    }
-
-    return jwt.encode(payload, USER_AUTH_SECRET, algorithm="HS256")
-
-
-def decode_anonymous_user_jwt_token(token: str) -> dict:
-    return jwt.decode(token, USER_AUTH_SECRET, algorithms=["HS256"])

backend/ee/onyx/background/celery/tasks/beat_schedule.py

@@ -1,9 +1,6 @@
 from datetime import timedelta
 from typing import Any
 
-from onyx.background.celery.tasks.beat_schedule import (
-    cloud_tasks_to_schedule as base_cloud_tasks_to_schedule,
-)
 from onyx.background.celery.tasks.beat_schedule import (
     tasks_to_schedule as base_tasks_to_schedule,
 )
@@ -11,7 +8,7 @@ from onyx.configs.constants import OnyxCeleryTask
 
 ee_tasks_to_schedule = [
     {
-        "name": "autogenerate-usage-report",
+        "name": "autogenerate_usage_report",
         "task": OnyxCeleryTask.AUTOGENERATE_USAGE_REPORT_TASK,
         "schedule": timedelta(days=30),  # TODO: change this to config flag
     },
@@ -23,9 +20,5 @@ ee_tasks_to_schedule = [
 ]
 
 
-def get_cloud_tasks_to_schedule() -> list[dict[str, Any]]:
-    return base_cloud_tasks_to_schedule
-
-
 def get_tasks_to_schedule() -> list[dict[str, Any]]:
     return ee_tasks_to_schedule + base_tasks_to_schedule

backend/ee/onyx/background/celery/tasks/vespa/tasks.py

@@ -8,9 +8,6 @@ from ee.onyx.db.user_group import fetch_user_group
 from ee.onyx.db.user_group import mark_user_group_as_synced
 from ee.onyx.db.user_group import prepare_user_group_for_deletion
 from onyx.background.celery.apps.app_base import task_logger
-from onyx.db.enums import SyncStatus
-from onyx.db.enums import SyncType
-from onyx.db.sync_record import update_sync_record_status
 from onyx.redis.redis_usergroup import RedisUserGroup
 from onyx.utils.logger import setup_logger
 
@@ -46,59 +43,24 @@ def monitor_usergroup_taskset(
         f"User group sync progress: usergroup_id={usergroup_id} remaining={count} initial={initial_count}"
     )
     if count > 0:
-        update_sync_record_status(
-            db_session=db_session,
-            entity_id=usergroup_id,
-            sync_type=SyncType.USER_GROUP,
-            sync_status=SyncStatus.IN_PROGRESS,
-            num_docs_synced=count,
-        )
         return
 
     user_group = fetch_user_group(db_session=db_session, user_group_id=usergroup_id)
     if user_group:
         usergroup_name = user_group.name
-        try:
-            if user_group.is_up_for_deletion:
-                # this prepare should have been run when the deletion was scheduled,
-                # but run it again to be sure we're ready to go
-                mark_user_group_as_synced(db_session, user_group)
-                prepare_user_group_for_deletion(db_session, usergroup_id)
-                delete_user_group(db_session=db_session, user_group=user_group)
-
-                update_sync_record_status(
-                    db_session=db_session,
-                    entity_id=usergroup_id,
-                    sync_type=SyncType.USER_GROUP,
-                    sync_status=SyncStatus.SUCCESS,
-                    num_docs_synced=initial_count,
-                )
-
-                task_logger.info(
-                    f"Deleted usergroup: name={usergroup_name} id={usergroup_id}"
-                )
-            else:
-                mark_user_group_as_synced(db_session=db_session, user_group=user_group)
-
-                update_sync_record_status(
-                    db_session=db_session,
-                    entity_id=usergroup_id,
-                    sync_type=SyncType.USER_GROUP,
-                    sync_status=SyncStatus.SUCCESS,
-                    num_docs_synced=initial_count,
-                )
-
-                task_logger.info(
-                    f"Synced usergroup. name={usergroup_name} id={usergroup_id}"
-                )
-        except Exception as e:
-            update_sync_record_status(
-                db_session=db_session,
-                entity_id=usergroup_id,
-                sync_type=SyncType.USER_GROUP,
-                sync_status=SyncStatus.FAILED,
-                num_docs_synced=initial_count,
+        if user_group.is_up_for_deletion:
+            # this prepare should have been run when the deletion was scheduled,
+            # but run it again to be sure we're ready to go
+            mark_user_group_as_synced(db_session, user_group)
+            prepare_user_group_for_deletion(db_session, usergroup_id)
+            delete_user_group(db_session=db_session, user_group=user_group)
+            task_logger.info(
+                f"Deleted usergroup: name={usergroup_name} id={usergroup_id}"
+            )
+        else:
+            mark_user_group_as_synced(db_session=db_session, user_group=user_group)
+            task_logger.info(
+                f"Synced usergroup. name={usergroup_name} id={usergroup_id}"
             )
-            raise e
 
     rug.reset()

backend/ee/onyx/configs/app_configs.py

@@ -61,5 +61,3 @@ POSTHOG_API_KEY = os.environ.get("POSTHOG_API_KEY") or "FooBar"
 POSTHOG_HOST = os.environ.get("POSTHOG_HOST") or "https://us.i.posthog.com"
 
 HUBSPOT_TRACKING_URL = os.environ.get("HUBSPOT_TRACKING_URL")
-
-ANONYMOUS_USER_COOKIE_NAME = "onyx_anonymous_user"

backend/ee/onyx/db/analytics.py

@@ -345,8 +345,7 @@ def fetch_assistant_unique_users_total(
 def user_can_view_assistant_stats(
     db_session: Session, user: User | None, assistant_id: int
 ) -> bool:
-    # If user is None and auth is disabled, assume the user is an admin
-
+    # If user is None, assume the user is an admin or auth is disabled
     if user is None or user.role == UserRole.ADMIN:
         return True
 

backend/ee/onyx/db/document.py

@@ -5,7 +5,7 @@ from sqlalchemy import select
 from sqlalchemy.orm import Session
 
 from onyx.access.models import ExternalAccess
-from onyx.access.utils import build_ext_group_name_for_onyx
+from onyx.access.utils import prefix_group_w_source
 from onyx.configs.constants import DocumentSource
 from onyx.db.models import Document as DbDocument
 
@@ -25,7 +25,7 @@ def upsert_document_external_perms__no_commit(
     ).first()
 
     prefixed_external_groups = [
-        build_ext_group_name_for_onyx(
+        prefix_group_w_source(
             ext_group_name=group_id,
             source=source_type,
         )
@@ -66,7 +66,7 @@ def upsert_document_external_perms(
     ).first()
 
     prefixed_external_groups: set[str] = {
-        build_ext_group_name_for_onyx(
+        prefix_group_w_source(
             ext_group_name=group_id,
             source=source_type,
         )

backend/ee/onyx/db/external_perm.py

@@ -6,12 +6,10 @@ from sqlalchemy import delete
 from sqlalchemy import select
 from sqlalchemy.orm import Session
 
-from onyx.access.utils import build_ext_group_name_for_onyx
+from onyx.access.utils import prefix_group_w_source
 from onyx.configs.constants import DocumentSource
-from onyx.db.models import User
 from onyx.db.models import User__ExternalUserGroupId
 from onyx.db.users import batch_add_ext_perm_user_if_not_exists
-from onyx.db.users import get_user_by_email
 from onyx.utils.logger import setup_logger
 
 logger = setup_logger()
@@ -62,10 +60,8 @@ def replace_user__ext_group_for_cc_pair(
             all_group_member_emails.add(user_email)
 
     # batch add users if they don't exist and get their ids
-    all_group_members: list[User] = batch_add_ext_perm_user_if_not_exists(
-        db_session=db_session,
-        # NOTE: this function handles case sensitivity for emails
-        emails=list(all_group_member_emails),
+    all_group_members = batch_add_ext_perm_user_if_not_exists(
+        db_session=db_session, emails=list(all_group_member_emails)
     )
 
     delete_user__ext_group_for_cc_pair__no_commit(
@@ -87,14 +83,12 @@ def replace_user__ext_group_for_cc_pair(
                     f" with email {user_email} not found"
                 )
                 continue
-            external_group_id = build_ext_group_name_for_onyx(
-                ext_group_name=external_group.id,
-                source=source,
-            )
             new_external_permissions.append(
                 User__ExternalUserGroupId(
                     user_id=user_id,
-                    external_user_group_id=external_group_id,
+                    external_user_group_id=prefix_group_w_source(
+                        external_group.id, source
+                    ),
                     cc_pair_id=cc_pair_id,
                 )
             )
@@ -112,21 +106,3 @@ def fetch_external_groups_for_user(
             User__ExternalUserGroupId.user_id == user_id
         )
     ).all()
-
-
-def fetch_external_groups_for_user_email_and_group_ids(
-    db_session: Session,
-    user_email: str,
-    group_ids: list[str],
-) -> list[User__ExternalUserGroupId]:
-    user = get_user_by_email(db_session=db_session, email=user_email)
-    if user is None:
-        return []
-    user_id = user.id
-    user_ext_groups = db_session.scalars(
-        select(User__ExternalUserGroupId).where(
-            User__ExternalUserGroupId.user_id == user_id,
-            User__ExternalUserGroupId.external_user_group_id.in_(group_ids),
-        )
-    ).all()
-    return list(user_ext_groups)

backend/ee/onyx/db/query_history.py

@@ -1,135 +1,27 @@
-from collections.abc import Sequence
-from datetime import datetime
+import datetime
+from typing import Literal
 
 from sqlalchemy import asc
 from sqlalchemy import BinaryExpression
 from sqlalchemy import ColumnElement
 from sqlalchemy import desc
-from sqlalchemy import distinct
 from sqlalchemy.orm import contains_eager
 from sqlalchemy.orm import joinedload
 from sqlalchemy.orm import Session
-from sqlalchemy.sql import case
-from sqlalchemy.sql import func
-from sqlalchemy.sql import select
-from sqlalchemy.sql.expression import literal
 from sqlalchemy.sql.expression import UnaryExpression
 
-from onyx.configs.constants import QAFeedbackType
 from onyx.db.models import ChatMessage
-from onyx.db.models import ChatMessageFeedback
 from onyx.db.models import ChatSession
 
-
-def _build_filter_conditions(
-    start_time: datetime | None,
-    end_time: datetime | None,
-    feedback_filter: QAFeedbackType | None,
-) -> list[ColumnElement]:
-    """
-    Helper function to build all filter conditions for chat sessions.
-    Filters by start and end time, feedback type, and any sessions without messages.
-    start_time: Date from which to filter
-    end_time: Date to which to filter
-    feedback_filter: Feedback type to filter by
-    Returns: List of filter conditions
-    """
-    conditions = []
-
-    if start_time is not None:
-        conditions.append(ChatSession.time_created >= start_time)
-    if end_time is not None:
-        conditions.append(ChatSession.time_created <= end_time)
-
-    if feedback_filter is not None:
-        feedback_subq = (
-            select(ChatMessage.chat_session_id)
-            .join(ChatMessageFeedback)
-            .group_by(ChatMessage.chat_session_id)
-            .having(
-                case(
-                    (
-                        case(
-                            {literal(feedback_filter == QAFeedbackType.LIKE): True},
-                            else_=False,
-                        ),
-                        func.bool_and(ChatMessageFeedback.is_positive),
-                    ),
-                    (
-                        case(
-                            {literal(feedback_filter == QAFeedbackType.DISLIKE): True},
-                            else_=False,
-                        ),
-                        func.bool_and(func.not_(ChatMessageFeedback.is_positive)),
-                    ),
-                    else_=func.bool_or(ChatMessageFeedback.is_positive)
-                    & func.bool_or(func.not_(ChatMessageFeedback.is_positive)),
-                )
-            )
-        )
-        conditions.append(ChatSession.id.in_(feedback_subq))
-
-    return conditions
-
-
-def get_total_filtered_chat_sessions_count(
-    db_session: Session,
-    start_time: datetime | None,
-    end_time: datetime | None,
-    feedback_filter: QAFeedbackType | None,
-) -> int:
-    conditions = _build_filter_conditions(start_time, end_time, feedback_filter)
-    stmt = (
-        select(func.count(distinct(ChatSession.id)))
-        .select_from(ChatSession)
-        .filter(*conditions)
-    )
-    return db_session.scalar(stmt) or 0
-
-
-def get_page_of_chat_sessions(
-    start_time: datetime | None,
-    end_time: datetime | None,
-    db_session: Session,
-    page_num: int,
-    page_size: int,
-    feedback_filter: QAFeedbackType | None = None,
-) -> Sequence[ChatSession]:
-    conditions = _build_filter_conditions(start_time, end_time, feedback_filter)
-
-    subquery = (
-        select(ChatSession.id, ChatSession.time_created)
-        .filter(*conditions)
-        .order_by(ChatSession.id, desc(ChatSession.time_created))
-        .distinct(ChatSession.id)
-        .limit(page_size)
-        .offset(page_num * page_size)
-        .subquery()
-    )
-
-    stmt = (
-        select(ChatSession)
-        .join(subquery, ChatSession.id == subquery.c.id)
-        .outerjoin(ChatMessage, ChatSession.id == ChatMessage.chat_session_id)
-        .options(
-            joinedload(ChatSession.user),
-            joinedload(ChatSession.persona),
-            contains_eager(ChatSession.messages).joinedload(
-                ChatMessage.chat_message_feedbacks
-            ),
-        )
-        .order_by(desc(ChatSession.time_created), asc(ChatMessage.id))
-    )
-
-    return db_session.scalars(stmt).unique().all()
+SortByOptions = Literal["time_sent"]
 
 
 def fetch_chat_sessions_eagerly_by_time(
-    start: datetime,
-    end: datetime,
+    start: datetime.datetime,
+    end: datetime.datetime,
     db_session: Session,
     limit: int | None = 500,
-    initial_time: datetime | None = None,
+    initial_time: datetime.datetime | None = None,
 ) -> list[ChatSession]:
     time_order: UnaryExpression = desc(ChatSession.time_created)
     message_order: UnaryExpression = asc(ChatMessage.id)

backend/ee/onyx/db/token_limit.py

@@ -7,7 +7,6 @@ from sqlalchemy import select
 from sqlalchemy.orm import aliased
 from sqlalchemy.orm import Session
 
-from onyx.configs.app_configs import DISABLE_AUTH
 from onyx.configs.constants import TokenRateLimitScope
 from onyx.db.models import TokenRateLimit
 from onyx.db.models import TokenRateLimit__UserGroup
@@ -21,11 +20,10 @@ from onyx.server.token_rate_limits.models import TokenRateLimitArgs
 def _add_user_filters(
     stmt: Select, user: User | None, get_editable: bool = True
 ) -> Select:
-    # If user is None and auth is disabled, assume the user is an admin
-    if (user is None and DISABLE_AUTH) or (user and user.role == UserRole.ADMIN):
+    # If user is None, assume the user is an admin or auth is disabled
+    if user is None or user.role == UserRole.ADMIN:
         return stmt
 
-    stmt = stmt.distinct()
     TRLimit_UG = aliased(TokenRateLimit__UserGroup)
     User__UG = aliased(User__UserGroup)
 
@@ -48,12 +46,6 @@ def _add_user_filters(
     that the user isn't a curator for
     - if we are not editing, we show all token_rate_limits in the groups the user curates
     """
-
-    # If user is None, this is an anonymous user and we should only show public token_rate_limits
-    if user is None:
-        where_clause = TokenRateLimit.scope == TokenRateLimitScope.GLOBAL
-        return stmt.where(where_clause)
-
     where_clause = User__UG.user_id == user.id
     if user.role == UserRole.CURATOR and get_editable:
         where_clause &= User__UG.is_curator == True  # noqa: E712
@@ -111,10 +103,10 @@ def insert_user_group_token_rate_limit(
     return token_limit
 
 
-def fetch_user_group_token_rate_limits_for_user(
+def fetch_user_group_token_rate_limits(
     db_session: Session,
     group_id: int,
-    user: User | None,
+    user: User | None = None,
     enabled_only: bool = False,
     ordered: bool = True,
     get_editable: bool = True,

backend/ee/onyx/db/user_group.py

@@ -374,9 +374,7 @@ def _add_user_group__cc_pair_relationships__no_commit(
 
 
 def insert_user_group(db_session: Session, user_group: UserGroupCreate) -> UserGroup:
-    db_user_group = UserGroup(
-        name=user_group.name, time_last_modified_by_user=func.now()
-    )
+    db_user_group = UserGroup(name=user_group.name)
     db_session.add(db_user_group)
     db_session.flush()  # give the group an ID
 
@@ -632,10 +630,6 @@ def update_user_group(
         select(User).where(User.id.in_(removed_user_ids))  # type: ignore
     ).unique()
     _validate_curator_status__no_commit(db_session, list(removed_users))
-
-    # update "time_updated" to now
-    db_user_group.time_last_modified_by_user = func.now()
-
     db_session.commit()
     return db_user_group
 
@@ -705,10 +699,7 @@ def delete_user_group_cc_pair_relationship__no_commit(
     connector_credential_pair_id matches the given cc_pair_id.
 
     Should be used very carefully (only for connectors that are being deleted)."""
-    cc_pair = get_connector_credential_pair_from_id(
-        db_session=db_session,
-        cc_pair_id=cc_pair_id,
-    )
+    cc_pair = get_connector_credential_pair_from_id(cc_pair_id, db_session)
     if not cc_pair:
         raise ValueError(f"Connector Credential Pair '{cc_pair_id}' does not exist")
 

backend/ee/onyx/external_permissions/confluence/doc_sync.py

@@ -24,9 +24,7 @@ _REQUEST_PAGINATION_LIMIT = 5000
 def _get_server_space_permissions(
     confluence_client: OnyxConfluence, space_key: str
 ) -> ExternalAccess:
-    space_permissions = confluence_client.get_all_space_permissions_server(
-        space_key=space_key
-    )
+    space_permissions = confluence_client.get_space_permissions(space_key=space_key)
 
     viewspace_permissions = []
     for permission_category in space_permissions:
@@ -69,13 +67,6 @@ def _get_server_space_permissions(
         else:
             logger.warning(f"Email for user {user_name} not found in Confluence")
 
-    if not user_emails and not group_names:
-        logger.warning(
-            "No user emails or group names found in Confluence space permissions"
-            f"\nSpace key: {space_key}"
-            f"\nSpace permissions: {space_permissions}"
-        )
-
     return ExternalAccess(
         external_user_emails=user_emails,
         external_user_group_ids=group_names,

backend/ee/onyx/external_permissions/confluence/group_sync.py

@@ -30,7 +30,6 @@ def _build_group_member_email_map(
                 )
         if not email:
             # If we still don't have an email, skip this user
-            logger.warning(f"user result missing email field: {user_result}")
             continue
 
         for group in confluence_client.paginated_groups_by_user_retrieval(user):

backend/ee/onyx/external_permissions/google_drive/doc_sync.py

@@ -120,12 +120,9 @@ def _get_permissions_from_slim_doc(
         elif permission_type == "anyone":
             public = True
 
-    drive_id = permission_info.get("drive_id")
-    group_ids = group_emails | ({drive_id} if drive_id is not None else set())
-
     return ExternalAccess(
         external_user_emails=user_emails,
-        external_user_group_ids=group_ids,
+        external_user_group_ids=group_emails,
         is_public=public,
     )
 

backend/ee/onyx/external_permissions/google_drive/group_sync.py

@@ -1,127 +1,16 @@
 from ee.onyx.db.external_perm import ExternalUserGroup
 from onyx.connectors.google_drive.connector import GoogleDriveConnector
 from onyx.connectors.google_utils.google_utils import execute_paginated_retrieval
-from onyx.connectors.google_utils.resources import AdminService
 from onyx.connectors.google_utils.resources import get_admin_service
-from onyx.connectors.google_utils.resources import get_drive_service
 from onyx.db.models import ConnectorCredentialPair
 from onyx.utils.logger import setup_logger
 
 logger = setup_logger()
 
 
-def _get_drive_members(
-    google_drive_connector: GoogleDriveConnector,
-) -> dict[str, tuple[set[str], set[str]]]:
-    """
-    This builds a map of drive ids to their members (group and user emails).
-    E.g. {
-        "drive_id_1": ({"group_email_1"}, {"user_email_1", "user_email_2"}),
-        "drive_id_2": ({"group_email_3"}, {"user_email_3"}),
-    }
-    """
-    drive_ids = google_drive_connector.get_all_drive_ids()
-
-    drive_id_to_members_map: dict[str, tuple[set[str], set[str]]] = {}
-    drive_service = get_drive_service(
-        google_drive_connector.creds,
-        google_drive_connector.primary_admin_email,
-    )
-
-    for drive_id in drive_ids:
-        group_emails: set[str] = set()
-        user_emails: set[str] = set()
-        for permission in execute_paginated_retrieval(
-            drive_service.permissions().list,
-            list_key="permissions",
-            fileId=drive_id,
-            fields="permissions(emailAddress, type)",
-            supportsAllDrives=True,
-        ):
-            if permission["type"] == "group":
-                group_emails.add(permission["emailAddress"])
-            elif permission["type"] == "user":
-                user_emails.add(permission["emailAddress"])
-        drive_id_to_members_map[drive_id] = (group_emails, user_emails)
-    return drive_id_to_members_map
-
-
-def _get_all_groups(
-    admin_service: AdminService,
-    google_domain: str,
-) -> set[str]:
-    """
-    This gets all the group emails.
-    """
-    group_emails: set[str] = set()
-    for group in execute_paginated_retrieval(
-        admin_service.groups().list,
-        list_key="groups",
-        domain=google_domain,
-        fields="groups(email)",
-    ):
-        group_emails.add(group["email"])
-    return group_emails
-
-
-def _map_group_email_to_member_emails(
-    admin_service: AdminService,
-    group_emails: set[str],
-) -> dict[str, set[str]]:
-    """
-    This maps group emails to their member emails.
-    """
-    group_to_member_map: dict[str, set[str]] = {}
-    for group_email in group_emails:
-        group_member_emails: set[str] = set()
-        for member in execute_paginated_retrieval(
-            admin_service.members().list,
-            list_key="members",
-            groupKey=group_email,
-            fields="members(email)",
-        ):
-            group_member_emails.add(member["email"])
-
-        group_to_member_map[group_email] = group_member_emails
-    return group_to_member_map
-
-
-def _build_onyx_groups(
-    drive_id_to_members_map: dict[str, tuple[set[str], set[str]]],
-    group_email_to_member_emails_map: dict[str, set[str]],
-) -> list[ExternalUserGroup]:
-    onyx_groups: list[ExternalUserGroup] = []
-
-    # Convert all drive member definitions to onyx groups
-    # This is because having drive level access means you have
-    # irrevocable access to all the files in the drive.
-    for drive_id, (group_emails, user_emails) in drive_id_to_members_map.items():
-        all_member_emails: set[str] = user_emails
-        for group_email in group_emails:
-            all_member_emails.update(group_email_to_member_emails_map[group_email])
-        onyx_groups.append(
-            ExternalUserGroup(
-                id=drive_id,
-                user_emails=list(all_member_emails),
-            )
-        )
-
-    # Convert all group member definitions to onyx groups
-    for group_email, member_emails in group_email_to_member_emails_map.items():
-        onyx_groups.append(
-            ExternalUserGroup(
-                id=group_email,
-                user_emails=list(member_emails),
-            )
-        )
-
-    return onyx_groups
-
-
 def gdrive_group_sync(
     cc_pair: ConnectorCredentialPair,
 ) -> list[ExternalUserGroup]:
-    # Initialize connector and build credential/service objects
     google_drive_connector = GoogleDriveConnector(
         **cc_pair.connector.connector_specific_config
     )
@@ -130,23 +19,34 @@ def gdrive_group_sync(
         google_drive_connector.creds, google_drive_connector.primary_admin_email
     )
 
-    # Get all drive members
-    drive_id_to_members_map = _get_drive_members(google_drive_connector)
+    onyx_groups: list[ExternalUserGroup] = []
+    for group in execute_paginated_retrieval(
+        admin_service.groups().list,
+        list_key="groups",
+        domain=google_drive_connector.google_domain,
+        fields="groups(email)",
+    ):
+        # The id is the group email
+        group_email = group["email"]
 
-    # Get all group emails
-    all_group_emails = _get_all_groups(
-        admin_service, google_drive_connector.google_domain
-    )
+        # Gather group member emails
+        group_member_emails: list[str] = []
+        for member in execute_paginated_retrieval(
+            admin_service.members().list,
+            list_key="members",
+            groupKey=group_email,
+            fields="members(email)",
+        ):
+            group_member_emails.append(member["email"])
 
-    # Map group emails to their members
-    group_email_to_member_emails_map = _map_group_email_to_member_emails(
-        admin_service, all_group_emails
-    )
+        if not group_member_emails:
+            continue
 
-    # Convert the maps to onyx groups
-    onyx_groups = _build_onyx_groups(
-        drive_id_to_members_map=drive_id_to_members_map,
-        group_email_to_member_emails_map=group_email_to_member_emails_map,
-    )
+        onyx_groups.append(
+            ExternalUserGroup(
+                id=group_email,
+                user_emails=list(group_member_emails),
+            )
+        )
 
     return onyx_groups

backend/ee/onyx/external_permissions/post_query_censoring.py

@@ -1,84 +0,0 @@
-from collections.abc import Callable
-
-from ee.onyx.db.connector_credential_pair import get_all_auto_sync_cc_pairs
-from ee.onyx.external_permissions.salesforce.postprocessing import (
-    censor_salesforce_chunks,
-)
-from onyx.configs.constants import DocumentSource
-from onyx.context.search.pipeline import InferenceChunk
-from onyx.db.engine import get_session_context_manager
-from onyx.db.models import User
-from onyx.utils.logger import setup_logger
-
-logger = setup_logger()
-
-DOC_SOURCE_TO_CHUNK_CENSORING_FUNCTION: dict[
-    DocumentSource,
-    # list of chunks to be censored and the user email. returns censored chunks
-    Callable[[list[InferenceChunk], str], list[InferenceChunk]],
-] = {
-    DocumentSource.SALESFORCE: censor_salesforce_chunks,
-}
-
-
-def _get_all_censoring_enabled_sources() -> set[DocumentSource]:
-    """
-    Returns the set of sources that have censoring enabled.
-    This is based on if the access_type is set to sync and the connector
-    source is included in DOC_SOURCE_TO_CHUNK_CENSORING_FUNCTION.
-
-    NOTE: This means if there is a source has a single cc_pair that is sync,
-    all chunks for that source will be censored, even if the connector that
-    indexed that chunk is not sync. This was done to avoid getting the cc_pair
-    for every single chunk.
-    """
-    with get_session_context_manager() as db_session:
-        enabled_sync_connectors = get_all_auto_sync_cc_pairs(db_session)
-        return {
-            cc_pair.connector.source
-            for cc_pair in enabled_sync_connectors
-            if cc_pair.connector.source in DOC_SOURCE_TO_CHUNK_CENSORING_FUNCTION
-        }
-
-
-# NOTE: This is only called if ee is enabled.
-def _post_query_chunk_censoring(
-    chunks: list[InferenceChunk],
-    user: User | None,
-) -> list[InferenceChunk]:
-    """
-    This function checks all chunks to see if they need to be sent to a censoring
-    function. If they do, it sends them to the censoring function and returns the
-    censored chunks. If they don't, it returns the original chunks.
-    """
-    if user is None:
-        # if user is None, permissions are not enforced
-        return chunks
-
-    chunks_to_keep = []
-    chunks_to_process: dict[DocumentSource, list[InferenceChunk]] = {}
-
-    sources_to_censor = _get_all_censoring_enabled_sources()
-    for chunk in chunks:
-        # Separate out chunks that require permission post-processing by source
-        if chunk.source_type in sources_to_censor:
-            chunks_to_process.setdefault(chunk.source_type, []).append(chunk)
-        else:
-            chunks_to_keep.append(chunk)
-
-    # For each source, filter out the chunks using the permission
-    # check function for that source
-    # TODO: Use a threadpool/multiprocessing to process the sources in parallel
-    for source, chunks_for_source in chunks_to_process.items():
-        censor_chunks_for_source = DOC_SOURCE_TO_CHUNK_CENSORING_FUNCTION[source]
-        try:
-            censored_chunks = censor_chunks_for_source(chunks_for_source, user.email)
-        except Exception as e:
-            logger.exception(
-                f"Failed to censor chunks for source {source} so throwing out all"
-                f" chunks for this source and continuing: {e}"
-            )
-            continue
-        chunks_to_keep.extend(censored_chunks)
-
-    return chunks_to_keep

backend/ee/onyx/external_permissions/salesforce/postprocessing.py

@@ -1,226 +0,0 @@
-import time
-
-from ee.onyx.db.external_perm import fetch_external_groups_for_user_email_and_group_ids
-from ee.onyx.external_permissions.salesforce.utils import (
-    get_any_salesforce_client_for_doc_id,
-)
-from ee.onyx.external_permissions.salesforce.utils import get_objects_access_for_user_id
-from ee.onyx.external_permissions.salesforce.utils import (
-    get_salesforce_user_id_from_email,
-)
-from onyx.configs.app_configs import BLURB_SIZE
-from onyx.context.search.models import InferenceChunk
-from onyx.db.engine import get_session_context_manager
-from onyx.utils.logger import setup_logger
-
-logger = setup_logger()
-
-
-# Types
-ChunkKey = tuple[str, int]  # (doc_id, chunk_id)
-ContentRange = tuple[int, int | None]  # (start_index, end_index) None means to the end
-
-
-# NOTE: Used for testing timing
-def _get_dummy_object_access_map(
-    object_ids: set[str], user_email: str, chunks: list[InferenceChunk]
-) -> dict[str, bool]:
-    time.sleep(0.15)
-    # return {object_id: True for object_id in object_ids}
-    import random
-
-    return {object_id: random.choice([True, False]) for object_id in object_ids}
-
-
-def _get_objects_access_for_user_email_from_salesforce(
-    object_ids: set[str],
-    user_email: str,
-    chunks: list[InferenceChunk],
-) -> dict[str, bool] | None:
-    """
-    This function wraps the salesforce call as we may want to change how this
-    is done in the future. (E.g. replace it with the above function)
-    """
-    # This is cached in the function so the first query takes an extra 0.1-0.3 seconds
-    # but subsequent queries for this source are essentially instant
-    first_doc_id = chunks[0].document_id
-    with get_session_context_manager() as db_session:
-        salesforce_client = get_any_salesforce_client_for_doc_id(
-            db_session, first_doc_id
-        )
-
-    # This is cached in the function so the first query takes an extra 0.1-0.3 seconds
-    # but subsequent queries by the same user are essentially instant
-    start_time = time.time()
-    user_id = get_salesforce_user_id_from_email(salesforce_client, user_email)
-    end_time = time.time()
-    logger.info(
-        f"Time taken to get Salesforce user ID: {end_time - start_time} seconds"
-    )
-    if user_id is None:
-        return None
-
-    # This is the only query that is not cached in the function
-    # so it takes 0.1-0.2 seconds total
-    object_id_to_access = get_objects_access_for_user_id(
-        salesforce_client, user_id, list(object_ids)
-    )
-    return object_id_to_access
-
-
-def _extract_salesforce_object_id_from_url(url: str) -> str:
-    return url.split("/")[-1]
-
-
-def _get_object_ranges_for_chunk(
-    chunk: InferenceChunk,
-) -> dict[str, list[ContentRange]]:
-    """
-    Given a chunk, return a dictionary of salesforce object ids and the content ranges
-    for that object id in the current chunk
-    """
-    if chunk.source_links is None:
-        return {}
-
-    object_ranges: dict[str, list[ContentRange]] = {}
-    end_index = None
-    descending_source_links = sorted(
-        chunk.source_links.items(), key=lambda x: x[0], reverse=True
-    )
-    for start_index, url in descending_source_links:
-        object_id = _extract_salesforce_object_id_from_url(url)
-        if object_id not in object_ranges:
-            object_ranges[object_id] = []
-        object_ranges[object_id].append((start_index, end_index))
-        end_index = start_index
-    return object_ranges
-
-
-def _create_empty_censored_chunk(uncensored_chunk: InferenceChunk) -> InferenceChunk:
-    """
-    Create a copy of the unfiltered chunk where potentially sensitive content is removed
-    to be added later if the user has access to each of the sub-objects
-    """
-    empty_censored_chunk = InferenceChunk(
-        **uncensored_chunk.model_dump(),
-    )
-    empty_censored_chunk.content = ""
-    empty_censored_chunk.blurb = ""
-    empty_censored_chunk.source_links = {}
-    return empty_censored_chunk
-
-
-def _update_censored_chunk(
-    censored_chunk: InferenceChunk,
-    uncensored_chunk: InferenceChunk,
-    content_range: ContentRange,
-) -> InferenceChunk:
-    """
-    Update the filtered chunk with the content and source links from the unfiltered chunk using the content ranges
-    """
-    start_index, end_index = content_range
-
-    # Update the content of the filtered chunk
-    permitted_content = uncensored_chunk.content[start_index:end_index]
-    permitted_section_start_index = len(censored_chunk.content)
-    censored_chunk.content = permitted_content + censored_chunk.content
-
-    # Update the source links of the filtered chunk
-    if uncensored_chunk.source_links is not None:
-        if censored_chunk.source_links is None:
-            censored_chunk.source_links = {}
-        link_content = uncensored_chunk.source_links[start_index]
-        censored_chunk.source_links[permitted_section_start_index] = link_content
-
-    # Update the blurb of the filtered chunk
-    censored_chunk.blurb = censored_chunk.content[:BLURB_SIZE]
-
-    return censored_chunk
-
-
-# TODO: Generalize this to other sources
-def censor_salesforce_chunks(
-    chunks: list[InferenceChunk],
-    user_email: str,
-    # This is so we can provide a mock access map for testing
-    access_map: dict[str, bool] | None = None,
-) -> list[InferenceChunk]:
-    # object_id -> list[((doc_id, chunk_id), (start_index, end_index))]
-    object_to_content_map: dict[str, list[tuple[ChunkKey, ContentRange]]] = {}
-
-    # (doc_id, chunk_id) -> chunk
-    uncensored_chunks: dict[ChunkKey, InferenceChunk] = {}
-
-    # keep track of all object ids that we have seen to make it easier to get
-    # the access for these object ids
-    object_ids: set[str] = set()
-
-    for chunk in chunks:
-        chunk_key = (chunk.document_id, chunk.chunk_id)
-        # create a dictionary to quickly look up the unfiltered chunk
-        uncensored_chunks[chunk_key] = chunk
-
-        # for each chunk, get a dictionary of object ids and the content ranges
-        # for that object id in the current chunk
-        object_ranges_for_chunk = _get_object_ranges_for_chunk(chunk)
-        for object_id, ranges in object_ranges_for_chunk.items():
-            object_ids.add(object_id)
-            for start_index, end_index in ranges:
-                object_to_content_map.setdefault(object_id, []).append(
-                    (chunk_key, (start_index, end_index))
-                )
-
-    # This is so we can provide a mock access map for testing
-    if access_map is None:
-        access_map = _get_objects_access_for_user_email_from_salesforce(
-            object_ids=object_ids,
-            user_email=user_email,
-            chunks=chunks,
-        )
-        if access_map is None:
-            # If the user is not found in Salesforce, access_map will be None
-            # so we should just return an empty list because no chunks will be
-            # censored
-            return []
-
-    censored_chunks: dict[ChunkKey, InferenceChunk] = {}
-    for object_id, content_list in object_to_content_map.items():
-        # if the user does not have access to the object, or the object is not in the
-        # access_map, do not include its content in the filtered chunks
-        if not access_map.get(object_id, False):
-            continue
-
-        # if we got this far, the user has access to the object so we can create or update
-        # the filtered chunk(s) for this object
-        # NOTE: we only create a censored chunk if the user has access to some
-        # part of the chunk
-        for chunk_key, content_range in content_list:
-            if chunk_key not in censored_chunks:
-                censored_chunks[chunk_key] = _create_empty_censored_chunk(
-                    uncensored_chunks[chunk_key]
-                )
-
-            uncensored_chunk = uncensored_chunks[chunk_key]
-            censored_chunk = _update_censored_chunk(
-                censored_chunk=censored_chunks[chunk_key],
-                uncensored_chunk=uncensored_chunk,
-                content_range=content_range,
-            )
-            censored_chunks[chunk_key] = censored_chunk
-
-    return list(censored_chunks.values())
-
-
-# NOTE: This is not used anywhere.
-def _get_objects_access_for_user_email(
-    object_ids: set[str], user_email: str
-) -> dict[str, bool]:
-    with get_session_context_manager() as db_session:
-        external_groups = fetch_external_groups_for_user_email_and_group_ids(
-            db_session=db_session,
-            user_email=user_email,
-            # Maybe make a function that adds a salesforce prefix to the group ids
-            group_ids=list(object_ids),
-        )
-        external_group_ids = {group.external_user_group_id for group in external_groups}
-        return {group_id: group_id in external_group_ids for group_id in object_ids}

backend/ee/onyx/external_permissions/salesforce/utils.py

@@ -1,177 +0,0 @@
-from simple_salesforce import Salesforce
-from sqlalchemy.orm import Session
-
-from onyx.connectors.salesforce.sqlite_functions import get_user_id_by_email
-from onyx.connectors.salesforce.sqlite_functions import init_db
-from onyx.connectors.salesforce.sqlite_functions import NULL_ID_STRING
-from onyx.connectors.salesforce.sqlite_functions import update_email_to_id_table
-from onyx.db.connector_credential_pair import get_connector_credential_pair_from_id
-from onyx.db.document import get_cc_pairs_for_document
-from onyx.utils.logger import setup_logger
-
-logger = setup_logger()
-
-_ANY_SALESFORCE_CLIENT: Salesforce | None = None
-
-
-def get_any_salesforce_client_for_doc_id(
-    db_session: Session, doc_id: str
-) -> Salesforce:
-    """
-    We create a salesforce client for the first cc_pair for the first doc_id where
-    salesforce censoring is enabled. After that we just cache and reuse the same
-    client for all queries.
-
-    We do this to reduce the number of postgres queries we make at query time.
-
-    This may be problematic if they are using multiple cc_pairs for salesforce.
-    E.g. there are 2 different credential sets for 2 different salesforce cc_pairs
-    but only one has the permissions to access the permissions needed for the query.
-    """
-    global _ANY_SALESFORCE_CLIENT
-    if _ANY_SALESFORCE_CLIENT is None:
-        cc_pairs = get_cc_pairs_for_document(db_session, doc_id)
-        first_cc_pair = cc_pairs[0]
-        credential_json = first_cc_pair.credential.credential_json
-        _ANY_SALESFORCE_CLIENT = Salesforce(
-            username=credential_json["sf_username"],
-            password=credential_json["sf_password"],
-            security_token=credential_json["sf_security_token"],
-        )
-    return _ANY_SALESFORCE_CLIENT
-
-
-def _query_salesforce_user_id(sf_client: Salesforce, user_email: str) -> str | None:
-    query = f"SELECT Id FROM User WHERE Email = '{user_email}'"
-    result = sf_client.query(query)
-    if len(result["records"]) == 0:
-        return None
-    return result["records"][0]["Id"]
-
-
-# This contains only the user_ids that we have found in Salesforce.
-# If we don't know their user_id, we don't store anything in the cache.
-_CACHED_SF_EMAIL_TO_ID_MAP: dict[str, str] = {}
-
-
-def get_salesforce_user_id_from_email(
-    sf_client: Salesforce,
-    user_email: str,
-) -> str | None:
-    """
-    We cache this so we don't have to query Salesforce for every query and salesforce
-    user IDs never change.
-    Memory usage is fine because we just store 2 small strings per user.
-
-    If the email is not in the cache, we check the local salesforce database for the info.
-    If the user is not found in the local salesforce database, we query Salesforce.
-    Whatever we get back from Salesforce is added to the database.
-    If no user_id is found, we add a NULL_ID_STRING to the database for that email so
-    we don't query Salesforce again (which is slow) but we still check the local salesforce
-    database every query until a user id is found. This is acceptable because the query time
-    is quite fast.
-    If a user_id is created in Salesforce, it will be added to the local salesforce database
-    next time the connector is run. Then that value will be found in this function and cached.
-
-    NOTE: First time this runs, it may be slow if it hasn't already been updated in the local
-    salesforce database. (Around 0.1-0.3 seconds)
-    If it's cached or stored in the local salesforce database, it's fast (<0.001 seconds).
-    """
-    global _CACHED_SF_EMAIL_TO_ID_MAP
-    if user_email in _CACHED_SF_EMAIL_TO_ID_MAP:
-        if _CACHED_SF_EMAIL_TO_ID_MAP[user_email] is not None:
-            return _CACHED_SF_EMAIL_TO_ID_MAP[user_email]
-
-    db_exists = True
-    try:
-        # Check if the user is already in the database
-        user_id = get_user_id_by_email(user_email)
-    except Exception:
-        init_db()
-        try:
-            user_id = get_user_id_by_email(user_email)
-        except Exception as e:
-            logger.error(f"Error checking if user is in database: {e}")
-            user_id = None
-            db_exists = False
-
-    # If no entry is found in the database (indicated by user_id being None)...
-    if user_id is None:
-        # ...query Salesforce and store the result in the database
-        user_id = _query_salesforce_user_id(sf_client, user_email)
-        if db_exists:
-            update_email_to_id_table(user_email, user_id)
-            return user_id
-        elif user_id is None:
-            return None
-    elif user_id == NULL_ID_STRING:
-        return None
-    # If the found user_id is real, cache it
-    _CACHED_SF_EMAIL_TO_ID_MAP[user_email] = user_id
-    return user_id
-
-
-_MAX_RECORD_IDS_PER_QUERY = 200
-
-
-def get_objects_access_for_user_id(
-    salesforce_client: Salesforce,
-    user_id: str,
-    record_ids: list[str],
-) -> dict[str, bool]:
-    """
-    Salesforce has a limit of 200 record ids per query. So we just truncate
-    the list of record ids to 200. We only ever retrieve 50 chunks at a time
-    so this should be fine (unlikely that we retrieve all 50 chunks contain
-    4 unique objects).
-    If we decide this isn't acceptable we can use multiple queries but they
-    should be in parallel so query time doesn't get too long.
-    """
-    truncated_record_ids = record_ids[:_MAX_RECORD_IDS_PER_QUERY]
-    record_ids_str = "'" + "','".join(truncated_record_ids) + "'"
-    access_query = f"""
-    SELECT RecordId, HasReadAccess
-    FROM UserRecordAccess
-    WHERE RecordId IN ({record_ids_str})
-    AND UserId = '{user_id}'
-    """
-    result = salesforce_client.query_all(access_query)
-    return {record["RecordId"]: record["HasReadAccess"] for record in result["records"]}
-
-
-_CC_PAIR_ID_SALESFORCE_CLIENT_MAP: dict[int, Salesforce] = {}
-_DOC_ID_TO_CC_PAIR_ID_MAP: dict[str, int] = {}
-
-
-# NOTE: This is not used anywhere.
-def _get_salesforce_client_for_doc_id(db_session: Session, doc_id: str) -> Salesforce:
-    """
-    Uses a document id to get the cc_pair that indexed that document and uses the credentials
-    for that cc_pair to create a Salesforce client.
-    Problems:
-    - There may be multiple cc_pairs for a document, and we don't know which one to use.
-        - right now we just use the first one
-    - Building a new Salesforce client for each document is slow.
-    - Memory usage could be an issue as we build these dictionaries.
-    """
-    if doc_id not in _DOC_ID_TO_CC_PAIR_ID_MAP:
-        cc_pairs = get_cc_pairs_for_document(db_session, doc_id)
-        first_cc_pair = cc_pairs[0]
-        _DOC_ID_TO_CC_PAIR_ID_MAP[doc_id] = first_cc_pair.id
-
-    cc_pair_id = _DOC_ID_TO_CC_PAIR_ID_MAP[doc_id]
-    if cc_pair_id not in _CC_PAIR_ID_SALESFORCE_CLIENT_MAP:
-        cc_pair = get_connector_credential_pair_from_id(
-            db_session=db_session,
-            cc_pair_id=cc_pair_id,
-        )
-        if cc_pair is None:
-            raise ValueError(f"CC pair {cc_pair_id} not found")
-        credential_json = cc_pair.credential.credential_json
-        _CC_PAIR_ID_SALESFORCE_CLIENT_MAP[cc_pair_id] = Salesforce(
-            username=credential_json["sf_username"],
-            password=credential_json["sf_password"],
-            security_token=credential_json["sf_security_token"],
-        )
-
-    return _CC_PAIR_ID_SALESFORCE_CLIENT_MAP[cc_pair_id]

backend/ee/onyx/external_permissions/sync_params.py

@@ -8,9 +8,6 @@ from ee.onyx.external_permissions.confluence.group_sync import confluence_group_
 from ee.onyx.external_permissions.gmail.doc_sync import gmail_doc_sync
 from ee.onyx.external_permissions.google_drive.doc_sync import gdrive_doc_sync
 from ee.onyx.external_permissions.google_drive.group_sync import gdrive_group_sync
-from ee.onyx.external_permissions.post_query_censoring import (
-    DOC_SOURCE_TO_CHUNK_CENSORING_FUNCTION,
-)
 from ee.onyx.external_permissions.slack.doc_sync import slack_doc_sync
 from onyx.access.models import DocExternalAccess
 from onyx.configs.constants import DocumentSource
@@ -74,7 +71,4 @@ EXTERNAL_GROUP_SYNC_PERIODS: dict[DocumentSource, int] = {
 
 
 def check_if_valid_sync_source(source_type: DocumentSource) -> bool:
-    return (
-        source_type in DOC_PERMISSIONS_FUNC_MAP
-        or source_type in DOC_SOURCE_TO_CHUNK_CENSORING_FUNCTION
-    )
+    return source_type in DOC_PERMISSIONS_FUNC_MAP

backend/ee/onyx/onyxbot/slack/handlers/handle_standard_answers.py

@@ -150,9 +150,9 @@ def _handle_standard_answers(
             db_session=db_session,
             description="",
             user_id=None,
-            persona_id=(
-                slack_channel_config.persona.id if slack_channel_config.persona else 0
-            ),
+            persona_id=slack_channel_config.persona.id
+            if slack_channel_config.persona
+            else 0,
             onyxbot_flow=True,
             slack_thread_id=slack_thread_id,
         )
@@ -182,7 +182,7 @@ def _handle_standard_answers(
             formatted_answers.append(formatted_answer)
         answer_message = "\n\n".join(formatted_answers)
 
-        chat_message = create_new_chat_message(
+        _ = create_new_chat_message(
             chat_session_id=chat_session.id,
             parent_message=new_user_message,
             prompt_id=prompt.id if prompt else None,
@@ -191,13 +191,8 @@ def _handle_standard_answers(
             message_type=MessageType.ASSISTANT,
             error=None,
             db_session=db_session,
-            commit=False,
+            commit=True,
         )
-        # attach the standard answers to the chat message
-        chat_message.standard_answers = [
-            standard_answer for standard_answer, _ in matching_standard_answers
-        ]
-        db_session.commit()
 
         update_emote_react(
             emoji=DANSWER_REACT_EMOJI,

backend/ee/onyx/server/middleware/tenant_tracking.py

@@ -7,8 +7,6 @@ from fastapi import HTTPException
 from fastapi import Request
 from fastapi import Response
 
-from ee.onyx.auth.users import decode_anonymous_user_jwt_token
-from ee.onyx.configs.app_configs import ANONYMOUS_USER_COOKIE_NAME
 from onyx.auth.api_key import extract_tenant_from_api_key_header
 from onyx.db.engine import is_valid_schema_name
 from onyx.redis.redis_pool import retrieve_auth_token_data_from_redis
@@ -50,16 +48,6 @@ async def _get_tenant_id_from_request(
     if tenant_id:
         return tenant_id
 
-    # Check for anonymous user cookie
-    anonymous_user_cookie = request.cookies.get(ANONYMOUS_USER_COOKIE_NAME)
-    if anonymous_user_cookie:
-        try:
-            anonymous_user_data = decode_anonymous_user_jwt_token(anonymous_user_cookie)
-            return anonymous_user_data.get("tenant_id", POSTGRES_DEFAULT_SCHEMA)
-        except Exception as e:
-            logger.error(f"Error decoding anonymous user cookie: {str(e)}")
-            # Continue and attempt to authenticate
-
     try:
         # Look up token data in Redis
         token_data = await retrieve_auth_token_data_from_redis(request)

backend/ee/onyx/server/query_history/api.py

@@ -1,23 +1,19 @@
 import csv
 import io
 from datetime import datetime
+from datetime import timedelta
 from datetime import timezone
+from typing import Literal
 from uuid import UUID
 
 from fastapi import APIRouter
 from fastapi import Depends
 from fastapi import HTTPException
-from fastapi import Query
 from fastapi.responses import StreamingResponse
+from pydantic import BaseModel
 from sqlalchemy.orm import Session
 
 from ee.onyx.db.query_history import fetch_chat_sessions_eagerly_by_time
-from ee.onyx.db.query_history import get_page_of_chat_sessions
-from ee.onyx.db.query_history import get_total_filtered_chat_sessions_count
-from ee.onyx.server.query_history.models import ChatSessionMinimal
-from ee.onyx.server.query_history.models import ChatSessionSnapshot
-from ee.onyx.server.query_history.models import MessageSnapshot
-from ee.onyx.server.query_history.models import QuestionAnswerPairSnapshot
 from onyx.auth.users import current_admin_user
 from onyx.auth.users import get_display_email
 from onyx.chat.chat_utils import create_chat_chain
@@ -27,15 +23,257 @@ from onyx.configs.constants import SessionType
 from onyx.db.chat import get_chat_session_by_id
 from onyx.db.chat import get_chat_sessions_by_user
 from onyx.db.engine import get_session
+from onyx.db.models import ChatMessage
 from onyx.db.models import ChatSession
 from onyx.db.models import User
-from onyx.server.documents.models import PaginatedReturn
 from onyx.server.query_and_chat.models import ChatSessionDetails
 from onyx.server.query_and_chat.models import ChatSessionsResponse
 
 router = APIRouter()
 
 
+class AbridgedSearchDoc(BaseModel):
+    """A subset of the info present in `SearchDoc`"""
+
+    document_id: str
+    semantic_identifier: str
+    link: str | None
+
+
+class MessageSnapshot(BaseModel):
+    message: str
+    message_type: MessageType
+    documents: list[AbridgedSearchDoc]
+    feedback_type: QAFeedbackType | None
+    feedback_text: str | None
+    time_created: datetime
+
+    @classmethod
+    def build(cls, message: ChatMessage) -> "MessageSnapshot":
+        latest_messages_feedback_obj = (
+            message.chat_message_feedbacks[-1]
+            if len(message.chat_message_feedbacks) > 0
+            else None
+        )
+        feedback_type = (
+            (
+                QAFeedbackType.LIKE
+                if latest_messages_feedback_obj.is_positive
+                else QAFeedbackType.DISLIKE
+            )
+            if latest_messages_feedback_obj
+            else None
+        )
+        feedback_text = (
+            latest_messages_feedback_obj.feedback_text
+            if latest_messages_feedback_obj
+            else None
+        )
+        return cls(
+            message=message.message,
+            message_type=message.message_type,
+            documents=[
+                AbridgedSearchDoc(
+                    document_id=document.document_id,
+                    semantic_identifier=document.semantic_id,
+                    link=document.link,
+                )
+                for document in message.search_docs
+            ],
+            feedback_type=feedback_type,
+            feedback_text=feedback_text,
+            time_created=message.time_sent,
+        )
+
+
+class ChatSessionMinimal(BaseModel):
+    id: UUID
+    user_email: str
+    name: str | None
+    first_user_message: str
+    first_ai_message: str
+    assistant_id: int | None
+    assistant_name: str | None
+    time_created: datetime
+    feedback_type: QAFeedbackType | Literal["mixed"] | None
+    flow_type: SessionType
+    conversation_length: int
+
+
+class ChatSessionSnapshot(BaseModel):
+    id: UUID
+    user_email: str
+    name: str | None
+    messages: list[MessageSnapshot]
+    assistant_id: int | None
+    assistant_name: str | None
+    time_created: datetime
+    flow_type: SessionType
+
+
+class QuestionAnswerPairSnapshot(BaseModel):
+    chat_session_id: UUID
+    # 1-indexed message number in the chat_session
+    # e.g. the first message pair in the chat_session is 1, the second is 2, etc.
+    message_pair_num: int
+    user_message: str
+    ai_response: str
+    retrieved_documents: list[AbridgedSearchDoc]
+    feedback_type: QAFeedbackType | None
+    feedback_text: str | None
+    persona_name: str | None
+    user_email: str
+    time_created: datetime
+    flow_type: SessionType
+
+    @classmethod
+    def from_chat_session_snapshot(
+        cls,
+        chat_session_snapshot: ChatSessionSnapshot,
+    ) -> list["QuestionAnswerPairSnapshot"]:
+        message_pairs: list[tuple[MessageSnapshot, MessageSnapshot]] = []
+        for ind in range(1, len(chat_session_snapshot.messages), 2):
+            message_pairs.append(
+                (
+                    chat_session_snapshot.messages[ind - 1],
+                    chat_session_snapshot.messages[ind],
+                )
+            )
+
+        return [
+            cls(
+                chat_session_id=chat_session_snapshot.id,
+                message_pair_num=ind + 1,
+                user_message=user_message.message,
+                ai_response=ai_message.message,
+                retrieved_documents=ai_message.documents,
+                feedback_type=ai_message.feedback_type,
+                feedback_text=ai_message.feedback_text,
+                persona_name=chat_session_snapshot.assistant_name,
+                user_email=get_display_email(chat_session_snapshot.user_email),
+                time_created=user_message.time_created,
+                flow_type=chat_session_snapshot.flow_type,
+            )
+            for ind, (user_message, ai_message) in enumerate(message_pairs)
+        ]
+
+    def to_json(self) -> dict[str, str | None]:
+        return {
+            "chat_session_id": str(self.chat_session_id),
+            "message_pair_num": str(self.message_pair_num),
+            "user_message": self.user_message,
+            "ai_response": self.ai_response,
+            "retrieved_documents": "|".join(
+                [
+                    doc.link or doc.semantic_identifier
+                    for doc in self.retrieved_documents
+                ]
+            ),
+            "feedback_type": self.feedback_type.value if self.feedback_type else "",
+            "feedback_text": self.feedback_text or "",
+            "persona_name": self.persona_name,
+            "user_email": self.user_email,
+            "time_created": str(self.time_created),
+            "flow_type": self.flow_type,
+        }
+
+
+def determine_flow_type(chat_session: ChatSession) -> SessionType:
+    return SessionType.SLACK if chat_session.onyxbot_flow else SessionType.CHAT
+
+
+def fetch_and_process_chat_session_history_minimal(
+    db_session: Session,
+    start: datetime,
+    end: datetime,
+    feedback_filter: QAFeedbackType | None = None,
+    limit: int | None = 500,
+) -> list[ChatSessionMinimal]:
+    chat_sessions = fetch_chat_sessions_eagerly_by_time(
+        start=start, end=end, db_session=db_session, limit=limit
+    )
+
+    minimal_sessions = []
+    for chat_session in chat_sessions:
+        if not chat_session.messages:
+            continue
+
+        first_user_message = next(
+            (
+                message.message
+                for message in chat_session.messages
+                if message.message_type == MessageType.USER
+            ),
+            "",
+        )
+        first_ai_message = next(
+            (
+                message.message
+                for message in chat_session.messages
+                if message.message_type == MessageType.ASSISTANT
+            ),
+            "",
+        )
+
+        has_positive_feedback = any(
+            feedback.is_positive
+            for message in chat_session.messages
+            for feedback in message.chat_message_feedbacks
+        )
+
+        has_negative_feedback = any(
+            not feedback.is_positive
+            for message in chat_session.messages
+            for feedback in message.chat_message_feedbacks
+        )
+
+        feedback_type: QAFeedbackType | Literal["mixed"] | None = (
+            "mixed"
+            if has_positive_feedback and has_negative_feedback
+            else QAFeedbackType.LIKE
+            if has_positive_feedback
+            else QAFeedbackType.DISLIKE
+            if has_negative_feedback
+            else None
+        )
+
+        if feedback_filter:
+            if feedback_filter == QAFeedbackType.LIKE and not has_positive_feedback:
+                continue
+            if feedback_filter == QAFeedbackType.DISLIKE and not has_negative_feedback:
+                continue
+
+        flow_type = determine_flow_type(chat_session)
+
+        minimal_sessions.append(
+            ChatSessionMinimal(
+                id=chat_session.id,
+                user_email=get_display_email(
+                    chat_session.user.email if chat_session.user else None
+                ),
+                name=chat_session.description,
+                first_user_message=first_user_message,
+                first_ai_message=first_ai_message,
+                assistant_id=chat_session.persona_id,
+                assistant_name=(
+                    chat_session.persona.name if chat_session.persona else None
+                ),
+                time_created=chat_session.time_created,
+                feedback_type=feedback_type,
+                flow_type=flow_type,
+                conversation_length=len(
+                    [
+                        m
+                        for m in chat_session.messages
+                        if m.message_type != MessageType.SYSTEM
+                    ]
+                ),
+            )
+        )
+
+    return minimal_sessions
+
+
 def fetch_and_process_chat_session_history(
     db_session: Session,
     start: datetime,
@@ -81,7 +319,7 @@ def snapshot_from_chat_session(
     except RuntimeError:
         return None
 
-    flow_type = SessionType.SLACK if chat_session.onyxbot_flow else SessionType.CHAT
+    flow_type = determine_flow_type(chat_session)
 
     return ChatSessionSnapshot(
         id=chat_session.id,
@@ -133,38 +371,22 @@ def get_user_chat_sessions(
 
 @router.get("/admin/chat-session-history")
 def get_chat_session_history(
-    page_num: int = Query(0, ge=0),
-    page_size: int = Query(10, ge=1),
     feedback_type: QAFeedbackType | None = None,
-    start_time: datetime | None = None,
-    end_time: datetime | None = None,
+    start: datetime | None = None,
+    end: datetime | None = None,
     _: User | None = Depends(current_admin_user),
     db_session: Session = Depends(get_session),
-) -> PaginatedReturn[ChatSessionMinimal]:
-    page_of_chat_sessions = get_page_of_chat_sessions(
-        page_num=page_num,
-        page_size=page_size,
+) -> list[ChatSessionMinimal]:
+    return fetch_and_process_chat_session_history_minimal(
         db_session=db_session,
-        start_time=start_time,
-        end_time=end_time,
+        start=start
+        or (
+            datetime.now(tz=timezone.utc) - timedelta(days=30)
+        ),  # default is 30d lookback
+        end=end or datetime.now(tz=timezone.utc),
         feedback_filter=feedback_type,
     )
 
-    total_filtered_chat_sessions_count = get_total_filtered_chat_sessions_count(
-        db_session=db_session,
-        start_time=start_time,
-        end_time=end_time,
-        feedback_filter=feedback_type,
-    )
-
-    return PaginatedReturn(
-        items=[
-            ChatSessionMinimal.from_chat_session(chat_session)
-            for chat_session in page_of_chat_sessions
-        ],
-        total_items=total_filtered_chat_sessions_count,
-    )
-
 
 @router.get("/admin/chat-session-history/{chat_session_id}")
 def get_chat_session_admin(

backend/ee/onyx/server/query_history/models.py

@@ -1,218 +0,0 @@
-from datetime import datetime
-from uuid import UUID
-
-from pydantic import BaseModel
-
-from onyx.auth.users import get_display_email
-from onyx.configs.constants import MessageType
-from onyx.configs.constants import QAFeedbackType
-from onyx.configs.constants import SessionType
-from onyx.db.models import ChatMessage
-from onyx.db.models import ChatSession
-
-
-class AbridgedSearchDoc(BaseModel):
-    """A subset of the info present in `SearchDoc`"""
-
-    document_id: str
-    semantic_identifier: str
-    link: str | None
-
-
-class MessageSnapshot(BaseModel):
-    id: int
-    message: str
-    message_type: MessageType
-    documents: list[AbridgedSearchDoc]
-    feedback_type: QAFeedbackType | None
-    feedback_text: str | None
-    time_created: datetime
-
-    @classmethod
-    def build(cls, message: ChatMessage) -> "MessageSnapshot":
-        latest_messages_feedback_obj = (
-            message.chat_message_feedbacks[-1]
-            if len(message.chat_message_feedbacks) > 0
-            else None
-        )
-        feedback_type = (
-            (
-                QAFeedbackType.LIKE
-                if latest_messages_feedback_obj.is_positive
-                else QAFeedbackType.DISLIKE
-            )
-            if latest_messages_feedback_obj
-            else None
-        )
-        feedback_text = (
-            latest_messages_feedback_obj.feedback_text
-            if latest_messages_feedback_obj
-            else None
-        )
-        return cls(
-            id=message.id,
-            message=message.message,
-            message_type=message.message_type,
-            documents=[
-                AbridgedSearchDoc(
-                    document_id=document.document_id,
-                    semantic_identifier=document.semantic_id,
-                    link=document.link,
-                )
-                for document in message.search_docs
-            ],
-            feedback_type=feedback_type,
-            feedback_text=feedback_text,
-            time_created=message.time_sent,
-        )
-
-
-class ChatSessionMinimal(BaseModel):
-    id: UUID
-    user_email: str
-    name: str | None
-    first_user_message: str
-    first_ai_message: str
-    assistant_id: int | None
-    assistant_name: str | None
-    time_created: datetime
-    feedback_type: QAFeedbackType | None
-    flow_type: SessionType
-    conversation_length: int
-
-    @classmethod
-    def from_chat_session(cls, chat_session: ChatSession) -> "ChatSessionMinimal":
-        first_user_message = next(
-            (
-                message.message
-                for message in chat_session.messages
-                if message.message_type == MessageType.USER
-            ),
-            "",
-        )
-        first_ai_message = next(
-            (
-                message.message
-                for message in chat_session.messages
-                if message.message_type == MessageType.ASSISTANT
-            ),
-            "",
-        )
-
-        list_of_message_feedbacks = [
-            feedback.is_positive
-            for message in chat_session.messages
-            for feedback in message.chat_message_feedbacks
-        ]
-        session_feedback_type = None
-        if list_of_message_feedbacks:
-            if all(list_of_message_feedbacks):
-                session_feedback_type = QAFeedbackType.LIKE
-            elif not any(list_of_message_feedbacks):
-                session_feedback_type = QAFeedbackType.DISLIKE
-            else:
-                session_feedback_type = QAFeedbackType.MIXED
-
-        return cls(
-            id=chat_session.id,
-            user_email=get_display_email(
-                chat_session.user.email if chat_session.user else None
-            ),
-            name=chat_session.description,
-            first_user_message=first_user_message,
-            first_ai_message=first_ai_message,
-            assistant_id=chat_session.persona_id,
-            assistant_name=(
-                chat_session.persona.name if chat_session.persona else None
-            ),
-            time_created=chat_session.time_created,
-            feedback_type=session_feedback_type,
-            flow_type=SessionType.SLACK
-            if chat_session.onyxbot_flow
-            else SessionType.CHAT,
-            conversation_length=len(
-                [
-                    message
-                    for message in chat_session.messages
-                    if message.message_type != MessageType.SYSTEM
-                ]
-            ),
-        )
-
-
-class ChatSessionSnapshot(BaseModel):
-    id: UUID
-    user_email: str
-    name: str | None
-    messages: list[MessageSnapshot]
-    assistant_id: int | None
-    assistant_name: str | None
-    time_created: datetime
-    flow_type: SessionType
-
-
-class QuestionAnswerPairSnapshot(BaseModel):
-    chat_session_id: UUID
-    # 1-indexed message number in the chat_session
-    # e.g. the first message pair in the chat_session is 1, the second is 2, etc.
-    message_pair_num: int
-    user_message: str
-    ai_response: str
-    retrieved_documents: list[AbridgedSearchDoc]
-    feedback_type: QAFeedbackType | None
-    feedback_text: str | None
-    persona_name: str | None
-    user_email: str
-    time_created: datetime
-    flow_type: SessionType
-
-    @classmethod
-    def from_chat_session_snapshot(
-        cls,
-        chat_session_snapshot: ChatSessionSnapshot,
-    ) -> list["QuestionAnswerPairSnapshot"]:
-        message_pairs: list[tuple[MessageSnapshot, MessageSnapshot]] = []
-        for ind in range(1, len(chat_session_snapshot.messages), 2):
-            message_pairs.append(
-                (
-                    chat_session_snapshot.messages[ind - 1],
-                    chat_session_snapshot.messages[ind],
-                )
-            )
-
-        return [
-            cls(
-                chat_session_id=chat_session_snapshot.id,
-                message_pair_num=ind + 1,
-                user_message=user_message.message,
-                ai_response=ai_message.message,
-                retrieved_documents=ai_message.documents,
-                feedback_type=ai_message.feedback_type,
-                feedback_text=ai_message.feedback_text,
-                persona_name=chat_session_snapshot.assistant_name,
-                user_email=get_display_email(chat_session_snapshot.user_email),
-                time_created=user_message.time_created,
-                flow_type=chat_session_snapshot.flow_type,
-            )
-            for ind, (user_message, ai_message) in enumerate(message_pairs)
-        ]
-
-    def to_json(self) -> dict[str, str | None]:
-        return {
-            "chat_session_id": str(self.chat_session_id),
-            "message_pair_num": str(self.message_pair_num),
-            "user_message": self.user_message,
-            "ai_response": self.ai_response,
-            "retrieved_documents": "|".join(
-                [
-                    doc.link or doc.semantic_identifier
-                    for doc in self.retrieved_documents
-                ]
-            ),
-            "feedback_type": self.feedback_type.value if self.feedback_type else "",
-            "feedback_text": self.feedback_text or "",
-            "persona_name": self.persona_name,
-            "user_email": self.user_email,
-            "time_created": str(self.time_created),
-            "flow_type": self.flow_type,
-        }

backend/ee/onyx/server/seeding.py

@@ -24,7 +24,7 @@ from onyx.db.llm import update_default_provider
 from onyx.db.llm import upsert_llm_provider
 from onyx.db.models import Tool
 from onyx.db.persona import upsert_persona
-from onyx.server.features.persona.models import PersonaUpsertRequest
+from onyx.server.features.persona.models import CreatePersonaRequest
 from onyx.server.manage.llm.models import LLMProviderUpsertRequest
 from onyx.server.settings.models import Settings
 from onyx.server.settings.store import store_settings as store_base_settings
@@ -57,7 +57,7 @@ class SeedConfiguration(BaseModel):
     llms: list[LLMProviderUpsertRequest] | None = None
     admin_user_emails: list[str] | None = None
     seeded_logo_path: str | None = None
-    personas: list[PersonaUpsertRequest] | None = None
+    personas: list[CreatePersonaRequest] | None = None
     settings: Settings | None = None
     enterprise_settings: EnterpriseSettings | None = None
 
@@ -128,7 +128,7 @@ def _seed_llms(
         )
 
 
-def _seed_personas(db_session: Session, personas: list[PersonaUpsertRequest]) -> None:
+def _seed_personas(db_session: Session, personas: list[CreatePersonaRequest]) -> None:
     if personas:
         logger.notice("Seeding Personas")
         for persona in personas:

backend/ee/onyx/server/tenants/anonymous_user_path.py

@@ -1,59 +0,0 @@
-from sqlalchemy import select
-from sqlalchemy.orm import Session
-
-from onyx.db.models import TenantAnonymousUserPath
-
-
-def get_anonymous_user_path(tenant_id: str, db_session: Session) -> str | None:
-    result = db_session.execute(
-        select(TenantAnonymousUserPath).where(
-            TenantAnonymousUserPath.tenant_id == tenant_id
-        )
-    )
-    result_scalar = result.scalar_one_or_none()
-    if result_scalar:
-        return result_scalar.anonymous_user_path
-    else:
-        return None
-
-
-def modify_anonymous_user_path(
-    tenant_id: str, anonymous_user_path: str, db_session: Session
-) -> None:
-    # Enforce lowercase path at DB operation level
-    anonymous_user_path = anonymous_user_path.lower()
-
-    existing_entry = (
-        db_session.query(TenantAnonymousUserPath).filter_by(tenant_id=tenant_id).first()
-    )
-
-    if existing_entry:
-        existing_entry.anonymous_user_path = anonymous_user_path
-
-    else:
-        new_entry = TenantAnonymousUserPath(
-            tenant_id=tenant_id, anonymous_user_path=anonymous_user_path
-        )
-        db_session.add(new_entry)
-
-    db_session.commit()
-
-
-def get_tenant_id_for_anonymous_user_path(
-    anonymous_user_path: str, db_session: Session
-) -> str | None:
-    result = db_session.execute(
-        select(TenantAnonymousUserPath).where(
-            TenantAnonymousUserPath.anonymous_user_path == anonymous_user_path
-        )
-    )
-    result_scalar = result.scalar_one_or_none()
-    if result_scalar:
-        return result_scalar.tenant_id
-    else:
-        return None
-
-
-def validate_anonymous_user_path(path: str) -> None:
-    if not path or "/" in path or not path.replace("-", "").isalnum():
-        raise ValueError("Invalid path. Use only letters, numbers, and hyphens.")

backend/ee/onyx/server/tenants/api.py

@@ -3,23 +3,13 @@ from fastapi import APIRouter
 from fastapi import Depends
 from fastapi import HTTPException
 from fastapi import Response
-from sqlalchemy.exc import IntegrityError
 from sqlalchemy.orm import Session
 
 from ee.onyx.auth.users import current_cloud_superuser
-from ee.onyx.auth.users import generate_anonymous_user_jwt_token
-from ee.onyx.configs.app_configs import ANONYMOUS_USER_COOKIE_NAME
 from ee.onyx.configs.app_configs import STRIPE_SECRET_KEY
 from ee.onyx.server.tenants.access import control_plane_dep
-from ee.onyx.server.tenants.anonymous_user_path import get_anonymous_user_path
-from ee.onyx.server.tenants.anonymous_user_path import (
-    get_tenant_id_for_anonymous_user_path,
-)
-from ee.onyx.server.tenants.anonymous_user_path import modify_anonymous_user_path
-from ee.onyx.server.tenants.anonymous_user_path import validate_anonymous_user_path
 from ee.onyx.server.tenants.billing import fetch_billing_information
 from ee.onyx.server.tenants.billing import fetch_tenant_stripe_information
-from ee.onyx.server.tenants.models import AnonymousUserPath
 from ee.onyx.server.tenants.models import BillingInformation
 from ee.onyx.server.tenants.models import ImpersonateRequest
 from ee.onyx.server.tenants.models import ProductGatingRequest
@@ -27,11 +17,9 @@ from ee.onyx.server.tenants.provisioning import delete_user_from_control_plane
 from ee.onyx.server.tenants.user_mapping import get_tenant_id_for_email
 from ee.onyx.server.tenants.user_mapping import remove_all_users_from_tenant
 from ee.onyx.server.tenants.user_mapping import remove_users_from_tenant
-from onyx.auth.users import anonymous_user_enabled
 from onyx.auth.users import auth_backend
 from onyx.auth.users import current_admin_user
 from onyx.auth.users import get_redis_strategy
-from onyx.auth.users import optional_user
 from onyx.auth.users import User
 from onyx.configs.app_configs import WEB_DOMAIN
 from onyx.db.auth import get_user_count
@@ -48,79 +36,11 @@ from onyx.utils.logger import setup_logger
 from shared_configs.contextvars import CURRENT_TENANT_ID_CONTEXTVAR
 
 stripe.api_key = STRIPE_SECRET_KEY
+
 logger = setup_logger()
 router = APIRouter(prefix="/tenants")
 
 
-@router.get("/anonymous-user-path")
-async def get_anonymous_user_path_api(
-    tenant_id: str | None = Depends(get_current_tenant_id),
-    _: User | None = Depends(current_admin_user),
-) -> AnonymousUserPath:
-    if tenant_id is None:
-        raise HTTPException(status_code=404, detail="Tenant not found")
-
-    with get_session_with_tenant(tenant_id=None) as db_session:
-        current_path = get_anonymous_user_path(tenant_id, db_session)
-
-    return AnonymousUserPath(anonymous_user_path=current_path)
-
-
-@router.post("/anonymous-user-path")
-async def set_anonymous_user_path_api(
-    anonymous_user_path: str,
-    tenant_id: str = Depends(get_current_tenant_id),
-    _: User | None = Depends(current_admin_user),
-) -> None:
-    try:
-        validate_anonymous_user_path(anonymous_user_path)
-    except ValueError as e:
-        raise HTTPException(status_code=400, detail=str(e))
-
-    with get_session_with_tenant(tenant_id=None) as db_session:
-        try:
-            modify_anonymous_user_path(tenant_id, anonymous_user_path, db_session)
-        except IntegrityError:
-            raise HTTPException(
-                status_code=409,
-                detail="The anonymous user path is already in use. Please choose a different path.",
-            )
-        except Exception as e:
-            logger.exception(f"Failed to modify anonymous user path: {str(e)}")
-            raise HTTPException(
-                status_code=500,
-                detail="An unexpected error occurred while modifying the anonymous user path",
-            )
-
-
-@router.post("/anonymous-user")
-async def login_as_anonymous_user(
-    anonymous_user_path: str,
-    _: User | None = Depends(optional_user),
-) -> Response:
-    with get_session_with_tenant(tenant_id=None) as db_session:
-        tenant_id = get_tenant_id_for_anonymous_user_path(
-            anonymous_user_path, db_session
-        )
-        if not tenant_id:
-            raise HTTPException(status_code=404, detail="Tenant not found")
-
-    if not anonymous_user_enabled(tenant_id=tenant_id):
-        raise HTTPException(status_code=403, detail="Anonymous user is not enabled")
-
-    token = generate_anonymous_user_jwt_token(tenant_id)
-
-    response = Response()
-    response.set_cookie(
-        key=ANONYMOUS_USER_COOKIE_NAME,
-        value=token,
-        httponly=True,
-        secure=True,
-        samesite="strict",
-    )
-    return response
-
-
 @router.post("/product-gating")
 def gate_product(
     product_gating_request: ProductGatingRequest, _: None = Depends(control_plane_dep)

backend/ee/onyx/server/tenants/models.py

@@ -44,7 +44,3 @@ class TenantCreationPayload(BaseModel):
 class TenantDeletionPayload(BaseModel):
     tenant_id: str
     email: str
-
-
-class AnonymousUserPath(BaseModel):
-    anonymous_user_path: str | None

backend/ee/onyx/server/token_rate_limits/api.py

@@ -5,7 +5,7 @@ from fastapi import Depends
 from sqlalchemy.orm import Session
 
 from ee.onyx.db.token_limit import fetch_all_user_group_token_rate_limits_by_group
-from ee.onyx.db.token_limit import fetch_user_group_token_rate_limits_for_user
+from ee.onyx.db.token_limit import fetch_user_group_token_rate_limits
 from ee.onyx.db.token_limit import insert_user_group_token_rate_limit
 from onyx.auth.users import current_admin_user
 from onyx.auth.users import current_curator_or_admin_user
@@ -51,10 +51,8 @@ def get_group_token_limit_settings(
 ) -> list[TokenRateLimitDisplay]:
     return [
         TokenRateLimitDisplay.from_db(token_rate_limit)
-        for token_rate_limit in fetch_user_group_token_rate_limits_for_user(
-            db_session=db_session,
-            group_id=group_id,
-            user=user,
+        for token_rate_limit in fetch_user_group_token_rate_limits(
+            db_session, group_id, user
         )
     ]
 

backend/onyx/access/utils.py

@@ -19,9 +19,6 @@ def prefix_external_group(ext_group_name: str) -> str:
     return f"external_group:{ext_group_name}"
 
 
-def build_ext_group_name_for_onyx(ext_group_name: str, source: DocumentSource) -> str:
-    """
-    External groups may collide across sources, every source needs its own prefix.
-    NOTE: the name is lowercased to handle case sensitivity for group names
-    """
-    return f"{source.value}_{ext_group_name}".lower()
+def prefix_group_w_source(ext_group_name: str, source: DocumentSource) -> str:
+    """External groups may collide across sources, every source needs its own prefix."""
+    return f"{source.value.upper()}_{ext_group_name}"

backend/onyx/auth/noauth_user.py

@@ -23,7 +23,6 @@ def load_no_auth_user_preferences(store: KeyValueStore) -> UserPreferences:
         preferences_data = cast(
             Mapping[str, Any], store.load(KV_NO_AUTH_USER_PREFERENCES_KEY)
         )
-        print("preferences_data", preferences_data)
         return UserPreferences(**preferences_data)
     except KvKeyNotFoundError:
         return UserPreferences(

backend/onyx/auth/users.py

@@ -80,7 +80,6 @@ from onyx.db.auth import get_user_db
 from onyx.db.auth import SQLAlchemyUserAdminDB
 from onyx.db.engine import get_async_session
 from onyx.db.engine import get_async_session_with_tenant
-from onyx.db.engine import get_current_tenant_id
 from onyx.db.engine import get_session_with_tenant
 from onyx.db.models import OAuthAccount
 from onyx.db.models import User
@@ -145,8 +144,11 @@ def user_needs_to_be_verified() -> bool:
     return False
 
 
-def anonymous_user_enabled(*, tenant_id: str | None = None) -> bool:
-    redis_client = get_redis_client(tenant_id=tenant_id)
+def anonymous_user_enabled() -> bool:
+    if MULTI_TENANT:
+        return False
+
+    redis_client = get_redis_client(tenant_id=None)
     value = redis_client.get(OnyxRedisLocks.ANONYMOUS_USER_ENABLED)
 
     if value is None:
@@ -771,10 +773,9 @@ async def current_limited_user(
 
 async def current_chat_accesssible_user(
     user: User | None = Depends(optional_user),
-    tenant_id: str | None = Depends(get_current_tenant_id),
 ) -> User | None:
     return await double_check_user(
-        user, allow_anonymous_access=anonymous_user_enabled(tenant_id=tenant_id)
+        user, allow_anonymous_access=anonymous_user_enabled()
     )
 
 

backend/onyx/background/celery/apps/app_base.py

@@ -20,7 +20,6 @@ from sqlalchemy.orm import Session
 from onyx.background.celery.apps.task_formatters import CeleryTaskColoredFormatter
 from onyx.background.celery.apps.task_formatters import CeleryTaskPlainFormatter
 from onyx.background.celery.celery_utils import celery_is_worker_primary
-from onyx.configs.constants import ONYX_CLOUD_CELERY_TASK_PREFIX
 from onyx.configs.constants import OnyxRedisLocks
 from onyx.db.engine import get_sqlalchemy_engine
 from onyx.document_index.vespa.shared_utils.utils import get_vespa_http_client
@@ -101,10 +100,6 @@ def on_task_postrun(
     if not task_id:
         return
 
-    if task.name.startswith(ONYX_CLOUD_CELERY_TASK_PREFIX):
-        # this is a cloud / all tenant task ... no postrun is needed
-        return
-
     # Get tenant_id directly from kwargs- each celery task has a tenant_id kwarg
     if not kwargs:
         logger.error(f"Task {task.name} (ID: {task_id}) is missing kwargs")
@@ -166,34 +161,9 @@ def on_task_postrun(
         return
 
 
-def on_celeryd_init(sender: str, conf: Any = None, **kwargs: Any) -> None:
+def on_celeryd_init(sender: Any = None, conf: Any = None, **kwargs: Any) -> None:
     """The first signal sent on celery worker startup"""
-
-    # NOTE(rkuo): start method "fork" is unsafe and we really need it to be "spawn"
-    # But something is blocking set_start_method from working in the cloud unless
-    # force=True. so we use force=True as a fallback.
-
-    all_start_methods: list[str] = multiprocessing.get_all_start_methods()
-    logger.info(f"Multiprocessing all start methods: {all_start_methods}")
-
-    try:
-        multiprocessing.set_start_method("spawn")  # fork is unsafe, set to spawn
-    except Exception:
-        logger.info(
-            "Multiprocessing set_start_method exceptioned. Trying force=True..."
-        )
-        try:
-            multiprocessing.set_start_method(
-                "spawn", force=True
-            )  # fork is unsafe, set to spawn
-        except Exception:
-            logger.info(
-                "Multiprocessing set_start_method force=True exceptioned even with force=True."
-            )
-
-    logger.info(
-        f"Multiprocessing selected start method: {multiprocessing.get_start_method()}"
-    )
+    multiprocessing.set_start_method("spawn")  # fork is unsafe, set to spawn
 
 
 def wait_for_redis(sender: Any, **kwargs: Any) -> None:

backend/onyx/background/celery/apps/beat.py

@@ -1,6 +1,5 @@
 from datetime import timedelta
 from typing import Any
-from typing import cast
 
 from celery import Celery
 from celery import signals
@@ -8,14 +7,12 @@ from celery.beat import PersistentScheduler  # type: ignore
 from celery.signals import beat_init
 
 import onyx.background.celery.apps.app_base as app_base
-from onyx.configs.constants import ONYX_CLOUD_CELERY_TASK_PREFIX
 from onyx.configs.constants import POSTGRES_CELERY_BEAT_APP_NAME
 from onyx.db.engine import get_all_tenant_ids
 from onyx.db.engine import SqlEngine
 from onyx.utils.logger import setup_logger
 from onyx.utils.variable_functionality import fetch_versioned_implementation
 from shared_configs.configs import IGNORED_SYNCING_TENANT_LIST
-from shared_configs.configs import MULTI_TENANT
 
 logger = setup_logger(__name__)
 
@@ -31,7 +28,7 @@ class DynamicTenantScheduler(PersistentScheduler):
         self._last_reload = self.app.now() - self._reload_interval
         # Let the parent class handle store initialization
         self.setup_schedule()
-        self._try_updating_schedule()
+        self._update_tenant_tasks()
         logger.info(f"Set reload interval to {self._reload_interval}")
 
     def setup_schedule(self) -> None:
@@ -47,154 +44,105 @@ class DynamicTenantScheduler(PersistentScheduler):
             or (now - self._last_reload) > self._reload_interval
         ):
             logger.info("Reload interval reached, initiating task update")
-            try:
-                self._try_updating_schedule()
-            except (AttributeError, KeyError) as e:
-                logger.exception(f"Failed to process task configuration: {str(e)}")
-            except Exception as e:
-                logger.exception(f"Unexpected error updating tasks: {str(e)}")
-
+            self._update_tenant_tasks()
             self._last_reload = now
             logger.info("Task update completed, reset reload timer")
         return retval
 
-    def _generate_schedule(
-        self, tenant_ids: list[str] | list[None]
-    ) -> dict[str, dict[str, Any]]:
-        """Given a list of tenant id's, generates a new beat schedule for celery."""
-        logger.info("Fetching tasks to schedule")
+    def _update_tenant_tasks(self) -> None:
+        logger.info("Starting task update process")
+        try:
+            logger.info("Fetching all IDs")
+            tenant_ids = get_all_tenant_ids()
+            logger.info(f"Found {len(tenant_ids)} IDs")
 
-        new_schedule: dict[str, dict[str, Any]] = {}
-
-        if MULTI_TENANT:
-            # cloud tasks only need the single task beat across all tenants
-            get_cloud_tasks_to_schedule = fetch_versioned_implementation(
-                "onyx.background.celery.tasks.beat_schedule",
-                "get_cloud_tasks_to_schedule",
+            logger.info("Fetching tasks to schedule")
+            tasks_to_schedule = fetch_versioned_implementation(
+                "onyx.background.celery.tasks.beat_schedule", "get_tasks_to_schedule"
             )
 
-            cloud_tasks_to_schedule: list[
-                dict[str, Any]
-            ] = get_cloud_tasks_to_schedule()
-            for task in cloud_tasks_to_schedule:
-                task_name = task["name"]
-                cloud_task = {
-                    "task": task["task"],
-                    "schedule": task["schedule"],
-                    "kwargs": {},
-                }
-                if options := task.get("options"):
-                    logger.debug(f"Adding options to task {task_name}: {options}")
-                    cloud_task["options"] = options
-                new_schedule[task_name] = cloud_task
+            new_beat_schedule: dict[str, dict[str, Any]] = {}
 
-        # regular task beats are multiplied across all tenants
-        get_tasks_to_schedule = fetch_versioned_implementation(
-            "onyx.background.celery.tasks.beat_schedule", "get_tasks_to_schedule"
-        )
+            current_schedule = self.schedule.items()
 
-        tasks_to_schedule: list[dict[str, Any]] = get_tasks_to_schedule()
+            existing_tenants = set()
+            for task_name, _ in current_schedule:
+                if "-" in task_name:
+                    existing_tenants.add(task_name.split("-")[-1])
+            logger.info(f"Found {len(existing_tenants)} existing items in schedule")
 
-        for tenant_id in tenant_ids:
-            if IGNORED_SYNCING_TENANT_LIST and tenant_id in IGNORED_SYNCING_TENANT_LIST:
-                logger.info(
-                    f"Skipping tenant {tenant_id} as it is in the ignored syncing list"
-                )
-                continue
-
-            for task in tasks_to_schedule:
-                task_name = task["name"]
-                tenant_task_name = f"{task['name']}-{tenant_id}"
-
-                logger.debug(f"Creating task configuration for {tenant_task_name}")
-                tenant_task = {
-                    "task": task["task"],
-                    "schedule": task["schedule"],
-                    "kwargs": {"tenant_id": tenant_id},
-                }
-                if options := task.get("options"):
-                    logger.debug(
-                        f"Adding options to task {tenant_task_name}: {options}"
+            for tenant_id in tenant_ids:
+                if (
+                    IGNORED_SYNCING_TENANT_LIST
+                    and tenant_id in IGNORED_SYNCING_TENANT_LIST
+                ):
+                    logger.info(
+                        f"Skipping tenant {tenant_id} as it is in the ignored syncing list"
                     )
-                    tenant_task["options"] = options
-                new_schedule[tenant_task_name] = tenant_task
+                    continue
 
-        return new_schedule
+                if tenant_id not in existing_tenants:
+                    logger.info(f"Processing new item: {tenant_id}")
 
-    def _try_updating_schedule(self) -> None:
-        """Only updates the actual beat schedule on the celery app when it changes"""
+                for task in tasks_to_schedule():
+                    task_name = f"{task['name']}-{tenant_id}"
+                    logger.debug(f"Creating task configuration for {task_name}")
+                    new_task = {
+                        "task": task["task"],
+                        "schedule": task["schedule"],
+                        "kwargs": {"tenant_id": tenant_id},
+                    }
+                    if options := task.get("options"):
+                        logger.debug(f"Adding options to task {task_name}: {options}")
+                        new_task["options"] = options
+                    new_beat_schedule[task_name] = new_task
 
-        logger.info("_try_updating_schedule starting")
+            if self._should_update_schedule(current_schedule, new_beat_schedule):
+                logger.info(
+                    "Schedule update required",
+                    extra={
+                        "new_tasks": len(new_beat_schedule),
+                        "current_tasks": len(current_schedule),
+                    },
+                )
 
-        tenant_ids = get_all_tenant_ids()
-        logger.info(f"Found {len(tenant_ids)} IDs")
+                # Create schedule entries
+                entries = {}
+                for name, entry in new_beat_schedule.items():
+                    entries[name] = self.Entry(
+                        name=name,
+                        app=self.app,
+                        task=entry["task"],
+                        schedule=entry["schedule"],
+                        options=entry.get("options", {}),
+                        kwargs=entry.get("kwargs", {}),
+                    )
 
-        # get current schedule and extract current tenants
-        current_schedule = self.schedule.items()
+                # Update the schedule using the scheduler's methods
+                self.schedule.clear()
+                self.schedule.update(entries)
 
-        current_tenants = set()
-        for task_name, _ in current_schedule:
-            task_name = cast(str, task_name)
-            if task_name.startswith(ONYX_CLOUD_CELERY_TASK_PREFIX):
-                continue
+                # Ensure changes are persisted
+                self.sync()
 
-            if "_" in task_name:
-                # example: "check-for-condition-tenant_12345678-abcd-efgh-ijkl-12345678"
-                # -> "12345678-abcd-efgh-ijkl-12345678"
-                current_tenants.add(task_name.split("_")[-1])
-        logger.info(f"Found {len(current_tenants)} existing items in schedule")
+                logger.info("Schedule update completed successfully")
+            else:
+                logger.info("Schedule is up to date, no changes needed")
+        except (AttributeError, KeyError) as e:
+            logger.exception(f"Failed to process task configuration: {str(e)}")
+        except Exception as e:
+            logger.exception(f"Unexpected error updating tasks: {str(e)}")
 
-        for tenant_id in tenant_ids:
-            if tenant_id not in current_tenants:
-                logger.info(f"Processing new tenant: {tenant_id}")
-
-        new_schedule = self._generate_schedule(tenant_ids)
-
-        if DynamicTenantScheduler._compare_schedules(current_schedule, new_schedule):
-            logger.info(
-                "_try_updating_schedule: Current schedule is up to date, no changes needed"
-            )
-            return
-
-        logger.info(
-            "Schedule update required",
-            extra={
-                "new_tasks": len(new_schedule),
-                "current_tasks": len(current_schedule),
-            },
-        )
-
-        # Create schedule entries
-        entries = {}
-        for name, entry in new_schedule.items():
-            entries[name] = self.Entry(
-                name=name,
-                app=self.app,
-                task=entry["task"],
-                schedule=entry["schedule"],
-                options=entry.get("options", {}),
-                kwargs=entry.get("kwargs", {}),
-            )
-
-        # Update the schedule using the scheduler's methods
-        self.schedule.clear()
-        self.schedule.update(entries)
-
-        # Ensure changes are persisted
-        self.sync()
-
-        logger.info("_try_updating_schedule: Schedule updated successfully")
-
-    @staticmethod
-    def _compare_schedules(schedule1: dict, schedule2: dict) -> bool:
-        """Compare schedules to determine if an update is needed.
-        True if equivalent, False if not."""
-        current_tasks = set(name for name, _ in schedule1)
-        new_tasks = set(schedule2.keys())
-        if current_tasks != new_tasks:
-            return False
-
-        return True
+    def _should_update_schedule(
+        self, current_schedule: dict, new_schedule: dict
+    ) -> bool:
+        """Compare schedules to determine if an update is needed."""
+        logger.debug("Comparing current and new schedules")
+        current_tasks = set(name for name, _ in current_schedule)
+        new_tasks = set(new_schedule.keys())
+        needs_update = current_tasks != new_tasks
+        logger.debug(f"Schedule update needed: {needs_update}")
+        return needs_update
 
 
 @beat_init.connect

backend/onyx/background/celery/apps/heavy.py

@@ -1,9 +1,9 @@
+import multiprocessing
 from typing import Any
 
 from celery import Celery
 from celery import signals
 from celery import Task
-from celery.apps.worker import Worker
 from celery.signals import celeryd_init
 from celery.signals import worker_init
 from celery.signals import worker_ready
@@ -49,16 +49,17 @@ def on_task_postrun(
 
 
 @celeryd_init.connect
-def on_celeryd_init(sender: str, conf: Any = None, **kwargs: Any) -> None:
+def on_celeryd_init(sender: Any = None, conf: Any = None, **kwargs: Any) -> None:
     app_base.on_celeryd_init(sender, conf, **kwargs)
 
 
 @worker_init.connect
-def on_worker_init(sender: Worker, **kwargs: Any) -> None:
+def on_worker_init(sender: Any, **kwargs: Any) -> None:
     logger.info("worker_init signal received.")
+    logger.info(f"Multiprocessing start method: {multiprocessing.get_start_method()}")
 
     SqlEngine.set_app_name(POSTGRES_CELERY_WORKER_HEAVY_APP_NAME)
-    SqlEngine.init_engine(pool_size=sender.concurrency, max_overflow=8)  # type: ignore
+    SqlEngine.init_engine(pool_size=4, max_overflow=12)
 
     app_base.wait_for_redis(sender, **kwargs)
     app_base.wait_for_db(sender, **kwargs)

backend/onyx/background/celery/apps/indexing.py

@@ -1,9 +1,9 @@
+import multiprocessing
 from typing import Any
 
 from celery import Celery
 from celery import signals
 from celery import Task
-from celery.apps.worker import Worker
 from celery.signals import celeryd_init
 from celery.signals import worker_init
 from celery.signals import worker_process_init
@@ -50,21 +50,22 @@ def on_task_postrun(
 
 
 @celeryd_init.connect
-def on_celeryd_init(sender: str, conf: Any = None, **kwargs: Any) -> None:
+def on_celeryd_init(sender: Any = None, conf: Any = None, **kwargs: Any) -> None:
     app_base.on_celeryd_init(sender, conf, **kwargs)
 
 
 @worker_init.connect
-def on_worker_init(sender: Worker, **kwargs: Any) -> None:
+def on_worker_init(sender: Any, **kwargs: Any) -> None:
     logger.info("worker_init signal received.")
+    logger.info(f"Multiprocessing start method: {multiprocessing.get_start_method()}")
 
     SqlEngine.set_app_name(POSTGRES_CELERY_WORKER_INDEXING_APP_NAME)
 
-    # rkuo: Transient errors keep happening in the indexing watchdog threads.
-    # "SSL connection has been closed unexpectedly"
-    # actually setting the spawn method in the cloud fixes 95% of these.
-    # setting pre ping might help even more, but not worrying about that yet
-    SqlEngine.init_engine(pool_size=sender.concurrency, max_overflow=8)  # type: ignore
+    # rkuo: been seeing transient connection exceptions here, so upping the connection count
+    # from just concurrency/concurrency to concurrency/concurrency*2
+    SqlEngine.init_engine(
+        pool_size=sender.concurrency, max_overflow=sender.concurrency * 2
+    )
 
     app_base.wait_for_redis(sender, **kwargs)
     app_base.wait_for_db(sender, **kwargs)

backend/onyx/background/celery/apps/light.py

@@ -1,9 +1,9 @@
+import multiprocessing
 from typing import Any
 
 from celery import Celery
 from celery import signals
 from celery import Task
-from celery.apps.worker import Worker
 from celery.signals import celeryd_init
 from celery.signals import worker_init
 from celery.signals import worker_ready
@@ -15,6 +15,7 @@ from onyx.db.engine import SqlEngine
 from onyx.utils.logger import setup_logger
 from shared_configs.configs import MULTI_TENANT
 
+
 logger = setup_logger()
 
 celery_app = Celery(__name__)
@@ -48,18 +49,17 @@ def on_task_postrun(
 
 
 @celeryd_init.connect
-def on_celeryd_init(sender: str, conf: Any = None, **kwargs: Any) -> None:
+def on_celeryd_init(sender: Any = None, conf: Any = None, **kwargs: Any) -> None:
     app_base.on_celeryd_init(sender, conf, **kwargs)
 
 
 @worker_init.connect
-def on_worker_init(sender: Worker, **kwargs: Any) -> None:
+def on_worker_init(sender: Any, **kwargs: Any) -> None:
     logger.info("worker_init signal received.")
-
-    logger.info(f"Concurrency: {sender.concurrency}")  # type: ignore
+    logger.info(f"Multiprocessing start method: {multiprocessing.get_start_method()}")
 
     SqlEngine.set_app_name(POSTGRES_CELERY_WORKER_LIGHT_APP_NAME)
-    SqlEngine.init_engine(pool_size=sender.concurrency, max_overflow=8)  # type: ignore
+    SqlEngine.init_engine(pool_size=sender.concurrency, max_overflow=8)
 
     app_base.wait_for_redis(sender, **kwargs)
     app_base.wait_for_db(sender, **kwargs)

backend/onyx/background/celery/apps/monitoring.py

@@ -1,95 +0,0 @@
-import multiprocessing
-from typing import Any
-
-from celery import Celery
-from celery import signals
-from celery import Task
-from celery.signals import celeryd_init
-from celery.signals import worker_init
-from celery.signals import worker_ready
-from celery.signals import worker_shutdown
-
-import onyx.background.celery.apps.app_base as app_base
-from onyx.configs.constants import POSTGRES_CELERY_WORKER_MONITORING_APP_NAME
-from onyx.db.engine import SqlEngine
-from onyx.utils.logger import setup_logger
-from shared_configs.configs import MULTI_TENANT
-
-
-logger = setup_logger()
-
-celery_app = Celery(__name__)
-celery_app.config_from_object("onyx.background.celery.configs.monitoring")
-
-
-@signals.task_prerun.connect
-def on_task_prerun(
-    sender: Any | None = None,
-    task_id: str | None = None,
-    task: Task | None = None,
-    args: tuple | None = None,
-    kwargs: dict | None = None,
-    **kwds: Any,
-) -> None:
-    app_base.on_task_prerun(sender, task_id, task, args, kwargs, **kwds)
-
-
-@signals.task_postrun.connect
-def on_task_postrun(
-    sender: Any | None = None,
-    task_id: str | None = None,
-    task: Task | None = None,
-    args: tuple | None = None,
-    kwargs: dict | None = None,
-    retval: Any | None = None,
-    state: str | None = None,
-    **kwds: Any,
-) -> None:
-    app_base.on_task_postrun(sender, task_id, task, args, kwargs, retval, state, **kwds)
-
-
-@celeryd_init.connect
-def on_celeryd_init(sender: Any = None, conf: Any = None, **kwargs: Any) -> None:
-    app_base.on_celeryd_init(sender, conf, **kwargs)
-
-
-@worker_init.connect
-def on_worker_init(sender: Any, **kwargs: Any) -> None:
-    logger.info("worker_init signal received.")
-    logger.info(f"Multiprocessing start method: {multiprocessing.get_start_method()}")
-
-    SqlEngine.set_app_name(POSTGRES_CELERY_WORKER_MONITORING_APP_NAME)
-    SqlEngine.init_engine(pool_size=sender.concurrency, max_overflow=3)
-
-    app_base.wait_for_redis(sender, **kwargs)
-    app_base.wait_for_db(sender, **kwargs)
-
-    # Less startup checks in multi-tenant case
-    if MULTI_TENANT:
-        return
-
-    app_base.on_secondary_worker_init(sender, **kwargs)
-
-
-@worker_ready.connect
-def on_worker_ready(sender: Any, **kwargs: Any) -> None:
-    app_base.on_worker_ready(sender, **kwargs)
-
-
-@worker_shutdown.connect
-def on_worker_shutdown(sender: Any, **kwargs: Any) -> None:
-    app_base.on_worker_shutdown(sender, **kwargs)
-
-
-@signals.setup_logging.connect
-def on_setup_logging(
-    loglevel: Any, logfile: Any, format: Any, colorize: Any, **kwargs: Any
-) -> None:
-    app_base.on_setup_logging(loglevel, logfile, format, colorize, **kwargs)
-
-
-celery_app.autodiscover_tasks(
-    [
-        "onyx.background.celery.tasks.monitoring",
-    ]
-)

backend/onyx/background/celery/apps/primary.py

@@ -1,4 +1,5 @@
 import logging
+import multiprocessing
 from typing import Any
 from typing import cast
 
@@ -6,7 +7,6 @@ from celery import bootsteps  # type: ignore
 from celery import Celery
 from celery import signals
 from celery import Task
-from celery.apps.worker import Worker
 from celery.exceptions import WorkerShutdown
 from celery.signals import celeryd_init
 from celery.signals import worker_init
@@ -17,7 +17,7 @@ from redis.lock import Lock as RedisLock
 import onyx.background.celery.apps.app_base as app_base
 from onyx.background.celery.apps.app_base import task_logger
 from onyx.background.celery.celery_utils import celery_is_worker_primary
-from onyx.background.celery.tasks.indexing.utils import (
+from onyx.background.celery.tasks.indexing.tasks import (
     get_unfenced_index_attempt_ids,
 )
 from onyx.configs.constants import CELERY_PRIMARY_WORKER_LOCK_TIMEOUT
@@ -73,13 +73,14 @@ def on_task_postrun(
 
 
 @celeryd_init.connect
-def on_celeryd_init(sender: str, conf: Any = None, **kwargs: Any) -> None:
+def on_celeryd_init(sender: Any = None, conf: Any = None, **kwargs: Any) -> None:
     app_base.on_celeryd_init(sender, conf, **kwargs)
 
 
 @worker_init.connect
-def on_worker_init(sender: Worker, **kwargs: Any) -> None:
+def on_worker_init(sender: Any, **kwargs: Any) -> None:
     logger.info("worker_init signal received.")
+    logger.info(f"Multiprocessing start method: {multiprocessing.get_start_method()}")
 
     SqlEngine.set_app_name(POSTGRES_CELERY_WORKER_PRIMARY_APP_NAME)
     SqlEngine.init_engine(pool_size=8, max_overflow=0)
@@ -134,7 +135,7 @@ def on_worker_init(sender: Worker, **kwargs: Any) -> None:
         raise WorkerShutdown("Primary worker lock could not be acquired!")
 
     # tacking on our own user data to the sender
-    sender.primary_worker_lock = lock  # type: ignore
+    sender.primary_worker_lock = lock
 
     # As currently designed, when this worker starts as "primary", we reinitialize redis
     # to a clean state (for our purposes, anyway)

backend/onyx/background/celery/celery_utils.py

@@ -14,7 +14,6 @@ from onyx.connectors.interfaces import PollConnector
 from onyx.connectors.interfaces import SlimConnector
 from onyx.connectors.models import Document
 from onyx.db.connector_credential_pair import get_connector_credential_pair
-from onyx.db.enums import ConnectorCredentialPairStatus
 from onyx.db.enums import TaskStatus
 from onyx.db.models import TaskQueueState
 from onyx.indexing.indexing_heartbeat import IndexingHeartbeatInterface
@@ -42,21 +41,14 @@ def _get_deletion_status(
         return None
 
     redis_connector = RedisConnector(tenant_id, cc_pair.id)
-    if redis_connector.delete.fenced:
-        return TaskQueueState(
-            task_id="",
-            task_name=redis_connector.delete.fence_key,
-            status=TaskStatus.STARTED,
-        )
+    if not redis_connector.delete.fenced:
+        return None
 
-    if cc_pair.status == ConnectorCredentialPairStatus.DELETING:
-        return TaskQueueState(
-            task_id="",
-            task_name=redis_connector.delete.fence_key,
-            status=TaskStatus.PENDING,
-        )
-
-    return None
+    return TaskQueueState(
+        task_id="",
+        task_name=redis_connector.delete.fence_key,
+        status=TaskStatus.STARTED,
+    )
 
 
 def get_deletion_attempt_snapshot(

backend/onyx/background/celery/configs/monitoring.py

@@ -1,21 +0,0 @@
-import onyx.background.celery.configs.base as shared_config
-
-broker_url = shared_config.broker_url
-broker_connection_retry_on_startup = shared_config.broker_connection_retry_on_startup
-broker_pool_limit = shared_config.broker_pool_limit
-broker_transport_options = shared_config.broker_transport_options
-
-redis_socket_keepalive = shared_config.redis_socket_keepalive
-redis_retry_on_timeout = shared_config.redis_retry_on_timeout
-redis_backend_health_check_interval = shared_config.redis_backend_health_check_interval
-
-result_backend = shared_config.result_backend
-result_expires = shared_config.result_expires  # 86400 seconds is the default
-
-task_default_priority = shared_config.task_default_priority
-task_acks_late = shared_config.task_acks_late
-
-# Monitoring worker specific settings
-worker_concurrency = 1  # Single worker is sufficient for monitoring
-worker_pool = "threads"
-worker_prefetch_multiplier = 1

backend/onyx/background/celery/tasks/beat_schedule.py

@@ -2,43 +2,24 @@ from datetime import timedelta
 from typing import Any
 
 from onyx.configs.app_configs import LLM_MODEL_UPDATE_API_URL
-from onyx.configs.constants import ONYX_CLOUD_CELERY_TASK_PREFIX
 from onyx.configs.constants import OnyxCeleryPriority
-from onyx.configs.constants import OnyxCeleryQueues
 from onyx.configs.constants import OnyxCeleryTask
-from shared_configs.configs import MULTI_TENANT
 
 # choosing 15 minutes because it roughly gives us enough time to process many tasks
 # we might be able to reduce this greatly if we can run a unified
 # loop across all tenants rather than tasks per tenant
 
-# we set expires because it isn't necessary to queue up these tasks
-# it's only important that they run relatively regularly
 BEAT_EXPIRES_DEFAULT = 15 * 60  # 15 minutes (in seconds)
 
-# tasks that only run in the cloud
-# the name attribute must start with ONYX_CELERY_CLOUD_PREFIX = "cloud" to be filtered
-# by the DynamicTenantScheduler
-cloud_tasks_to_schedule = [
-    {
-        "name": f"{ONYX_CLOUD_CELERY_TASK_PREFIX}_check-for-indexing",
-        "task": OnyxCeleryTask.CLOUD_CHECK_FOR_INDEXING,
-        "schedule": timedelta(seconds=15),
-        "options": {
-            "priority": OnyxCeleryPriority.HIGHEST,
-            "expires": BEAT_EXPIRES_DEFAULT,
-        },
-    },
-]
-
-# tasks that run in either self-hosted on cloud
+# we set expires because it isn't necessary to queue up these tasks
+# it's only important that they run relatively regularly
 tasks_to_schedule = [
     {
         "name": "check-for-vespa-sync",
         "task": OnyxCeleryTask.CHECK_FOR_VESPA_SYNC_TASK,
         "schedule": timedelta(seconds=20),
         "options": {
-            "priority": OnyxCeleryPriority.MEDIUM,
+            "priority": OnyxCeleryPriority.HIGH,
             "expires": BEAT_EXPIRES_DEFAULT,
         },
     },
@@ -47,7 +28,16 @@ tasks_to_schedule = [
         "task": OnyxCeleryTask.CHECK_FOR_CONNECTOR_DELETION,
         "schedule": timedelta(seconds=20),
         "options": {
-            "priority": OnyxCeleryPriority.MEDIUM,
+            "priority": OnyxCeleryPriority.HIGH,
+            "expires": BEAT_EXPIRES_DEFAULT,
+        },
+    },
+    {
+        "name": "check-for-indexing",
+        "task": OnyxCeleryTask.CHECK_FOR_INDEXING,
+        "schedule": timedelta(seconds=15),
+        "options": {
+            "priority": OnyxCeleryPriority.HIGH,
             "expires": BEAT_EXPIRES_DEFAULT,
         },
     },
@@ -56,7 +46,7 @@ tasks_to_schedule = [
         "task": OnyxCeleryTask.CHECK_FOR_PRUNING,
         "schedule": timedelta(seconds=15),
         "options": {
-            "priority": OnyxCeleryPriority.MEDIUM,
+            "priority": OnyxCeleryPriority.HIGH,
             "expires": BEAT_EXPIRES_DEFAULT,
         },
     },
@@ -74,26 +64,16 @@ tasks_to_schedule = [
         "task": OnyxCeleryTask.MONITOR_VESPA_SYNC,
         "schedule": timedelta(seconds=5),
         "options": {
-            "priority": OnyxCeleryPriority.MEDIUM,
+            "priority": OnyxCeleryPriority.HIGH,
             "expires": BEAT_EXPIRES_DEFAULT,
         },
     },
-    {
-        "name": "monitor-background-processes",
-        "task": OnyxCeleryTask.MONITOR_BACKGROUND_PROCESSES,
-        "schedule": timedelta(minutes=5),
-        "options": {
-            "priority": OnyxCeleryPriority.LOW,
-            "expires": BEAT_EXPIRES_DEFAULT,
-            "queue": OnyxCeleryQueues.MONITORING,
-        },
-    },
     {
         "name": "check-for-doc-permissions-sync",
         "task": OnyxCeleryTask.CHECK_FOR_DOC_PERMISSIONS_SYNC,
         "schedule": timedelta(seconds=30),
         "options": {
-            "priority": OnyxCeleryPriority.MEDIUM,
+            "priority": OnyxCeleryPriority.HIGH,
             "expires": BEAT_EXPIRES_DEFAULT,
         },
     },
@@ -102,25 +82,12 @@ tasks_to_schedule = [
         "task": OnyxCeleryTask.CHECK_FOR_EXTERNAL_GROUP_SYNC,
         "schedule": timedelta(seconds=20),
         "options": {
-            "priority": OnyxCeleryPriority.MEDIUM,
+            "priority": OnyxCeleryPriority.HIGH,
             "expires": BEAT_EXPIRES_DEFAULT,
         },
     },
 ]
 
-if not MULTI_TENANT:
-    tasks_to_schedule.append(
-        {
-            "name": "check-for-indexing",
-            "task": OnyxCeleryTask.CHECK_FOR_INDEXING,
-            "schedule": timedelta(seconds=15),
-            "options": {
-                "priority": OnyxCeleryPriority.MEDIUM,
-                "expires": BEAT_EXPIRES_DEFAULT,
-            },
-        }
-    )
-
 # Only add the LLM model update task if the API URL is configured
 if LLM_MODEL_UPDATE_API_URL:
     tasks_to_schedule.append(
@@ -136,9 +103,5 @@ if LLM_MODEL_UPDATE_API_URL:
     )
 
 
-def get_cloud_tasks_to_schedule() -> list[dict[str, Any]]:
-    return cloud_tasks_to_schedule
-
-
 def get_tasks_to_schedule() -> list[dict[str, Any]]:
     return tasks_to_schedule

backend/onyx/background/celery/tasks/connector_deletion/tasks.py

@@ -10,17 +10,14 @@ from sqlalchemy.orm import Session
 
 from onyx.background.celery.apps.app_base import task_logger
 from onyx.configs.app_configs import JOB_TIMEOUT
-from onyx.configs.constants import CELERY_GENERIC_BEAT_LOCK_TIMEOUT
+from onyx.configs.constants import CELERY_VESPA_SYNC_BEAT_LOCK_TIMEOUT
 from onyx.configs.constants import OnyxCeleryTask
 from onyx.configs.constants import OnyxRedisLocks
 from onyx.db.connector_credential_pair import get_connector_credential_pair_from_id
 from onyx.db.connector_credential_pair import get_connector_credential_pairs
 from onyx.db.engine import get_session_with_tenant
 from onyx.db.enums import ConnectorCredentialPairStatus
-from onyx.db.enums import SyncType
 from onyx.db.search_settings import get_all_search_settings
-from onyx.db.sync_record import cleanup_sync_records
-from onyx.db.sync_record import insert_sync_record
 from onyx.redis.redis_connector import RedisConnector
 from onyx.redis.redis_connector_delete import RedisConnectorDeletePayload
 from onyx.redis.redis_pool import get_redis_client
@@ -44,14 +41,14 @@ def check_for_connector_deletion_task(
 
     lock_beat: RedisLock = r.lock(
         OnyxRedisLocks.CHECK_CONNECTOR_DELETION_BEAT_LOCK,
-        timeout=CELERY_GENERIC_BEAT_LOCK_TIMEOUT,
+        timeout=CELERY_VESPA_SYNC_BEAT_LOCK_TIMEOUT,
     )
 
-    # these tasks should never overlap
-    if not lock_beat.acquire(blocking=False):
-        return None
-
     try:
+        # these tasks should never overlap
+        if not lock_beat.acquire(blocking=False):
+            return None
+
         # collect cc_pair_ids
         cc_pair_ids: list[int] = []
         with get_session_with_tenant(tenant_id) as db_session:
@@ -116,21 +113,11 @@ def try_generate_document_cc_pair_cleanup_tasks(
     # we need to load the state of the object inside the fence
     # to avoid a race condition with db.commit/fence deletion
     # at the end of this taskset
-    cc_pair = get_connector_credential_pair_from_id(
-        db_session=db_session,
-        cc_pair_id=cc_pair_id,
-    )
+    cc_pair = get_connector_credential_pair_from_id(cc_pair_id, db_session)
     if not cc_pair:
         return None
 
     if cc_pair.status != ConnectorCredentialPairStatus.DELETING:
-        # there should be no in-progress sync records if this is up to date
-        # clean it up just in case things got into a bad state
-        cleanup_sync_records(
-            db_session=db_session,
-            entity_id=cc_pair_id,
-            sync_type=SyncType.CONNECTOR_DELETION,
-        )
         return None
 
     # set a basic fence to start
@@ -139,13 +126,6 @@ def try_generate_document_cc_pair_cleanup_tasks(
         submitted=datetime.now(timezone.utc),
     )
 
-    # create before setting fence to avoid race condition where the monitoring
-    # task updates the sync record before it is created
-    insert_sync_record(
-        db_session=db_session,
-        entity_id=cc_pair_id,
-        sync_type=SyncType.CONNECTOR_DELETION,
-    )
     redis_connector.delete.set_fence(fence_payload)
 
     try:

backend/onyx/background/celery/tasks/doc_permission_syncing/tasks.py

@@ -16,15 +16,12 @@ from ee.onyx.db.connector_credential_pair import get_all_auto_sync_cc_pairs
 from ee.onyx.db.document import upsert_document_external_perms
 from ee.onyx.external_permissions.sync_params import DOC_PERMISSION_SYNC_PERIODS
 from ee.onyx.external_permissions.sync_params import DOC_PERMISSIONS_FUNC_MAP
-from ee.onyx.external_permissions.sync_params import (
-    DOC_SOURCE_TO_CHUNK_CENSORING_FUNCTION,
-)
 from onyx.access.models import DocExternalAccess
 from onyx.background.celery.apps.app_base import task_logger
 from onyx.configs.app_configs import JOB_TIMEOUT
-from onyx.configs.constants import CELERY_GENERIC_BEAT_LOCK_TIMEOUT
 from onyx.configs.constants import CELERY_PERMISSIONS_SYNC_LOCK_TIMEOUT
 from onyx.configs.constants import CELERY_TASK_WAIT_FOR_FENCE_TIMEOUT
+from onyx.configs.constants import CELERY_VESPA_SYNC_BEAT_LOCK_TIMEOUT
 from onyx.configs.constants import DANSWER_REDIS_FUNCTION_LOCK_PREFIX
 from onyx.configs.constants import DocumentSource
 from onyx.configs.constants import OnyxCeleryPriority
@@ -99,14 +96,14 @@ def check_for_doc_permissions_sync(self: Task, *, tenant_id: str | None) -> bool
 
     lock_beat: RedisLock = r.lock(
         OnyxRedisLocks.CHECK_CONNECTOR_DOC_PERMISSIONS_SYNC_BEAT_LOCK,
-        timeout=CELERY_GENERIC_BEAT_LOCK_TIMEOUT,
+        timeout=CELERY_VESPA_SYNC_BEAT_LOCK_TIMEOUT,
     )
 
-    # these tasks should never overlap
-    if not lock_beat.acquire(blocking=False):
-        return None
-
     try:
+        # these tasks should never overlap
+        if not lock_beat.acquire(blocking=False):
+            return None
+
         # get all cc pairs that need to be synced
         cc_pair_ids_to_sync: list[int] = []
         with get_session_with_tenant(tenant_id) as db_session:
@@ -279,10 +276,7 @@ def connector_permission_sync_generator_task(
 
     try:
         with get_session_with_tenant(tenant_id) as db_session:
-            cc_pair = get_connector_credential_pair_from_id(
-                db_session=db_session,
-                cc_pair_id=cc_pair_id,
-            )
+            cc_pair = get_connector_credential_pair_from_id(cc_pair_id, db_session)
             if cc_pair is None:
                 raise ValueError(
                     f"No connector credential pair found for id: {cc_pair_id}"
@@ -292,8 +286,6 @@ def connector_permission_sync_generator_task(
 
             doc_sync_func = DOC_PERMISSIONS_FUNC_MAP.get(source_type)
             if doc_sync_func is None:
-                if source_type in DOC_SOURCE_TO_CHUNK_CENSORING_FUNCTION:
-                    return None
                 raise ValueError(
                     f"No doc sync func found for {source_type} with cc_pair={cc_pair_id}"
                 )
@@ -394,7 +386,5 @@ def update_external_document_permissions_task(
             )
         return True
     except Exception:
-        logger.exception(
-            f"Error Syncing Document Permissions: connector_id={connector_id} doc_id={doc_id}"
-        )
+        logger.exception("Error Syncing Document Permissions")
         return False

backend/onyx/background/celery/tasks/external_group_syncing/tasks.py

@@ -22,7 +22,7 @@ from ee.onyx.external_permissions.sync_params import (
 from onyx.background.celery.apps.app_base import task_logger
 from onyx.configs.app_configs import JOB_TIMEOUT
 from onyx.configs.constants import CELERY_EXTERNAL_GROUP_SYNC_LOCK_TIMEOUT
-from onyx.configs.constants import CELERY_GENERIC_BEAT_LOCK_TIMEOUT
+from onyx.configs.constants import CELERY_VESPA_SYNC_BEAT_LOCK_TIMEOUT
 from onyx.configs.constants import DANSWER_REDIS_FUNCTION_LOCK_PREFIX
 from onyx.configs.constants import OnyxCeleryPriority
 from onyx.configs.constants import OnyxCeleryQueues
@@ -99,14 +99,14 @@ def check_for_external_group_sync(self: Task, *, tenant_id: str | None) -> bool
 
     lock_beat: RedisLock = r.lock(
         OnyxRedisLocks.CHECK_CONNECTOR_EXTERNAL_GROUP_SYNC_BEAT_LOCK,
-        timeout=CELERY_GENERIC_BEAT_LOCK_TIMEOUT,
+        timeout=CELERY_VESPA_SYNC_BEAT_LOCK_TIMEOUT,
     )
 
-    # these tasks should never overlap
-    if not lock_beat.acquire(blocking=False):
-        return None
-
     try:
+        # these tasks should never overlap
+        if not lock_beat.acquire(blocking=False):
+            return None
+
         cc_pair_ids_to_sync: list[int] = []
         with get_session_with_tenant(tenant_id) as db_session:
             cc_pairs = get_all_auto_sync_cc_pairs(db_session)
@@ -250,10 +250,7 @@ def connector_external_group_sync_generator_task(
             return None
 
         with get_session_with_tenant(tenant_id) as db_session:
-            cc_pair = get_connector_credential_pair_from_id(
-                db_session=db_session,
-                cc_pair_id=cc_pair_id,
-            )
+            cc_pair = get_connector_credential_pair_from_id(cc_pair_id, db_session)
             if cc_pair is None:
                 raise ValueError(
                     f"No connector credential pair found for id: {cc_pair_id}"

backend/onyx/background/celery/tasks/indexing/utils.py

@@ -1,519 +0,0 @@
-import time
-from datetime import datetime
-from datetime import timezone
-
-import redis
-from celery import Celery
-from redis import Redis
-from redis.exceptions import LockError
-from redis.lock import Lock as RedisLock
-from sqlalchemy.orm import Session
-
-from onyx.background.celery.apps.app_base import task_logger
-from onyx.background.celery.celery_redis import celery_find_task
-from onyx.background.celery.celery_redis import celery_get_unacked_task_ids
-from onyx.configs.app_configs import DISABLE_INDEX_UPDATE_ON_SWAP
-from onyx.configs.constants import CELERY_GENERIC_BEAT_LOCK_TIMEOUT
-from onyx.configs.constants import DANSWER_REDIS_FUNCTION_LOCK_PREFIX
-from onyx.configs.constants import DocumentSource
-from onyx.configs.constants import OnyxCeleryPriority
-from onyx.configs.constants import OnyxCeleryQueues
-from onyx.configs.constants import OnyxCeleryTask
-from onyx.db.engine import get_db_current_time
-from onyx.db.engine import get_session_with_tenant
-from onyx.db.enums import ConnectorCredentialPairStatus
-from onyx.db.enums import IndexingStatus
-from onyx.db.enums import IndexModelStatus
-from onyx.db.index_attempt import create_index_attempt
-from onyx.db.index_attempt import delete_index_attempt
-from onyx.db.index_attempt import get_all_index_attempts_by_status
-from onyx.db.index_attempt import get_index_attempt
-from onyx.db.index_attempt import mark_attempt_failed
-from onyx.db.models import ConnectorCredentialPair
-from onyx.db.models import IndexAttempt
-from onyx.db.models import SearchSettings
-from onyx.indexing.indexing_heartbeat import IndexingHeartbeatInterface
-from onyx.redis.redis_connector import RedisConnector
-from onyx.redis.redis_connector_index import RedisConnectorIndex
-from onyx.redis.redis_connector_index import RedisConnectorIndexPayload
-from onyx.redis.redis_pool import redis_lock_dump
-from onyx.redis.redis_pool import SCAN_ITER_COUNT_DEFAULT
-from onyx.utils.logger import setup_logger
-
-logger = setup_logger()
-
-
-def get_unfenced_index_attempt_ids(db_session: Session, r: redis.Redis) -> list[int]:
-    """Gets a list of unfenced index attempts. Should not be possible, so we'd typically
-    want to clean them up.
-
-    Unfenced = attempt not in terminal state and fence does not exist.
-    """
-    unfenced_attempts: list[int] = []
-
-    # inner/outer/inner double check pattern to avoid race conditions when checking for
-    # bad state
-    # inner = index_attempt in non terminal state
-    # outer = r.fence_key down
-
-    # check the db for index attempts in a non terminal state
-    attempts: list[IndexAttempt] = []
-    attempts.extend(
-        get_all_index_attempts_by_status(IndexingStatus.NOT_STARTED, db_session)
-    )
-    attempts.extend(
-        get_all_index_attempts_by_status(IndexingStatus.IN_PROGRESS, db_session)
-    )
-
-    for attempt in attempts:
-        fence_key = RedisConnectorIndex.fence_key_with_ids(
-            attempt.connector_credential_pair_id, attempt.search_settings_id
-        )
-
-        # if the fence is down / doesn't exist, possible error but not confirmed
-        if r.exists(fence_key):
-            continue
-
-        # Between the time the attempts are first looked up and the time we see the fence down,
-        # the attempt may have completed and taken down the fence normally.
-
-        # We need to double check that the index attempt is still in a non terminal state
-        # and matches the original state, which confirms we are really in a bad state.
-        attempt_2 = get_index_attempt(db_session, attempt.id)
-        if not attempt_2:
-            continue
-
-        if attempt.status != attempt_2.status:
-            continue
-
-        unfenced_attempts.append(attempt.id)
-
-    return unfenced_attempts
-
-
-class IndexingCallback(IndexingHeartbeatInterface):
-    PARENT_CHECK_INTERVAL = 60
-
-    def __init__(
-        self,
-        parent_pid: int,
-        stop_key: str,
-        generator_progress_key: str,
-        redis_lock: RedisLock,
-        redis_client: Redis,
-    ):
-        super().__init__()
-        self.parent_pid = parent_pid
-        self.redis_lock: RedisLock = redis_lock
-        self.stop_key: str = stop_key
-        self.generator_progress_key: str = generator_progress_key
-        self.redis_client = redis_client
-        self.started: datetime = datetime.now(timezone.utc)
-        self.redis_lock.reacquire()
-
-        self.last_tag: str = "IndexingCallback.__init__"
-        self.last_lock_reacquire: datetime = datetime.now(timezone.utc)
-        self.last_lock_monotonic = time.monotonic()
-
-        self.last_parent_check = time.monotonic()
-
-    def should_stop(self) -> bool:
-        if self.redis_client.exists(self.stop_key):
-            return True
-
-        return False
-
-    def progress(self, tag: str, amount: int) -> None:
-        # rkuo: this shouldn't be necessary yet because we spawn the process this runs inside
-        # with daemon = True. It seems likely some indexing tasks will need to spawn other processes eventually
-        # so leave this code in until we're ready to test it.
-
-        # if self.parent_pid:
-        #     # check if the parent pid is alive so we aren't running as a zombie
-        #     now = time.monotonic()
-        #     if now - self.last_parent_check > IndexingCallback.PARENT_CHECK_INTERVAL:
-        #         try:
-        #             # this is unintuitive, but it checks if the parent pid is still running
-        #             os.kill(self.parent_pid, 0)
-        #         except Exception:
-        #             logger.exception("IndexingCallback - parent pid check exceptioned")
-        #             raise
-        #         self.last_parent_check = now
-
-        try:
-            current_time = time.monotonic()
-            if current_time - self.last_lock_monotonic >= (
-                CELERY_GENERIC_BEAT_LOCK_TIMEOUT / 4
-            ):
-                self.redis_lock.reacquire()
-                self.last_lock_reacquire = datetime.now(timezone.utc)
-                self.last_lock_monotonic = time.monotonic()
-
-            self.last_tag = tag
-        except LockError:
-            logger.exception(
-                f"IndexingCallback - lock.reacquire exceptioned: "
-                f"lock_timeout={self.redis_lock.timeout} "
-                f"start={self.started} "
-                f"last_tag={self.last_tag} "
-                f"last_reacquired={self.last_lock_reacquire} "
-                f"now={datetime.now(timezone.utc)}"
-            )
-
-            redis_lock_dump(self.redis_lock, self.redis_client)
-            raise
-
-        self.redis_client.incrby(self.generator_progress_key, amount)
-
-
-def validate_indexing_fence(
-    tenant_id: str | None,
-    key_bytes: bytes,
-    reserved_tasks: set[str],
-    r_celery: Redis,
-    db_session: Session,
-) -> None:
-    """Checks for the error condition where an indexing fence is set but the associated celery tasks don't exist.
-    This can happen if the indexing worker hard crashes or is terminated.
-    Being in this bad state means the fence will never clear without help, so this function
-    gives the help.
-
-    How this works:
-    1. This function renews the active signal with a 5 minute TTL under the following conditions
-    1.2. When the task is seen in the redis queue
-    1.3. When the task is seen in the reserved / prefetched list
-
-    2. Externally, the active signal is renewed when:
-    2.1. The fence is created
-    2.2. The indexing watchdog checks the spawned task.
-
-    3. The TTL allows us to get through the transitions on fence startup
-    and when the task starts executing.
-
-    More TTL clarification: it is seemingly impossible to exactly query Celery for
-    whether a task is in the queue or currently executing.
-    1. An unknown task id is always returned as state PENDING.
-    2. Redis can be inspected for the task id, but the task id is gone between the time a worker receives the task
-    and the time it actually starts on the worker.
-    """
-    # if the fence doesn't exist, there's nothing to do
-    fence_key = key_bytes.decode("utf-8")
-    composite_id = RedisConnector.get_id_from_fence_key(fence_key)
-    if composite_id is None:
-        task_logger.warning(
-            f"validate_indexing_fence - could not parse composite_id from {fence_key}"
-        )
-        return
-
-    # parse out metadata and initialize the helper class with it
-    parts = composite_id.split("/")
-    if len(parts) != 2:
-        return
-
-    cc_pair_id = int(parts[0])
-    search_settings_id = int(parts[1])
-
-    redis_connector = RedisConnector(tenant_id, cc_pair_id)
-    redis_connector_index = redis_connector.new_index(search_settings_id)
-
-    # check to see if the fence/payload exists
-    if not redis_connector_index.fenced:
-        return
-
-    payload = redis_connector_index.payload
-    if not payload:
-        return
-
-    # OK, there's actually something for us to validate
-
-    if payload.celery_task_id is None:
-        # the fence is just barely set up.
-        if redis_connector_index.active():
-            return
-
-        # it would be odd to get here as there isn't that much that can go wrong during
-        # initial fence setup, but it's still worth making sure we can recover
-        logger.info(
-            f"validate_indexing_fence - Resetting fence in basic state without any activity: fence={fence_key}"
-        )
-        redis_connector_index.reset()
-        return
-
-    found = celery_find_task(
-        payload.celery_task_id, OnyxCeleryQueues.CONNECTOR_INDEXING, r_celery
-    )
-    if found:
-        # the celery task exists in the redis queue
-        redis_connector_index.set_active()
-        return
-
-    if payload.celery_task_id in reserved_tasks:
-        # the celery task was prefetched and is reserved within the indexing worker
-        redis_connector_index.set_active()
-        return
-
-    # we may want to enable this check if using the active task list somehow isn't good enough
-    # if redis_connector_index.generator_locked():
-    #     logger.info(f"{payload.celery_task_id} is currently executing.")
-
-    # if we get here, we didn't find any direct indication that the associated celery tasks exist,
-    # but they still might be there due to gaps in our ability to check states during transitions
-    # Checking the active signal safeguards us against these transition periods
-    # (which has a duration that allows us to bridge those gaps)
-    if redis_connector_index.active():
-        return
-
-    # celery tasks don't exist and the active signal has expired, possibly due to a crash. Clean it up.
-    logger.warning(
-        f"validate_indexing_fence - Resetting fence because no associated celery tasks were found: "
-        f"index_attempt={payload.index_attempt_id} "
-        f"cc_pair={cc_pair_id} "
-        f"search_settings={search_settings_id} "
-        f"fence={fence_key}"
-    )
-    if payload.index_attempt_id:
-        try:
-            mark_attempt_failed(
-                payload.index_attempt_id,
-                db_session,
-                "validate_indexing_fence - Canceling index attempt due to missing celery tasks: "
-                f"index_attempt={payload.index_attempt_id}",
-            )
-        except Exception:
-            logger.exception(
-                "validate_indexing_fence - Exception while marking index attempt as failed: "
-                f"index_attempt={payload.index_attempt_id}",
-            )
-
-    redis_connector_index.reset()
-    return
-
-
-def validate_indexing_fences(
-    tenant_id: str | None,
-    celery_app: Celery,
-    r: Redis,
-    r_celery: Redis,
-    lock_beat: RedisLock,
-) -> None:
-    reserved_indexing_tasks = celery_get_unacked_task_ids(
-        OnyxCeleryQueues.CONNECTOR_INDEXING, r_celery
-    )
-
-    # validate all existing indexing jobs
-    for key_bytes in r.scan_iter(
-        RedisConnectorIndex.FENCE_PREFIX + "*", count=SCAN_ITER_COUNT_DEFAULT
-    ):
-        lock_beat.reacquire()
-        with get_session_with_tenant(tenant_id) as db_session:
-            validate_indexing_fence(
-                tenant_id,
-                key_bytes,
-                reserved_indexing_tasks,
-                r_celery,
-                db_session,
-            )
-    return
-
-
-def _should_index(
-    cc_pair: ConnectorCredentialPair,
-    last_index: IndexAttempt | None,
-    search_settings_instance: SearchSettings,
-    search_settings_primary: bool,
-    secondary_index_building: bool,
-    db_session: Session,
-) -> bool:
-    """Checks various global settings and past indexing attempts to determine if
-    we should try to start indexing the cc pair / search setting combination.
-
-    Note that tactical checks such as preventing overlap with a currently running task
-    are not handled here.
-
-    Return True if we should try to index, False if not.
-    """
-    connector = cc_pair.connector
-
-    # uncomment for debugging
-    # task_logger.info(f"_should_index: "
-    #                  f"cc_pair={cc_pair.id} "
-    #                  f"connector={cc_pair.connector_id} "
-    #                  f"refresh_freq={connector.refresh_freq}")
-
-    # don't kick off indexing for `NOT_APPLICABLE` sources
-    if connector.source == DocumentSource.NOT_APPLICABLE:
-        return False
-
-    # User can still manually create single indexing attempts via the UI for the
-    # currently in use index
-    if DISABLE_INDEX_UPDATE_ON_SWAP:
-        if (
-            search_settings_instance.status == IndexModelStatus.PRESENT
-            and secondary_index_building
-        ):
-            return False
-
-    # When switching over models, always index at least once
-    if search_settings_instance.status == IndexModelStatus.FUTURE:
-        if last_index:
-            # No new index if the last index attempt succeeded
-            # Once is enough. The model will never be able to swap otherwise.
-            if last_index.status == IndexingStatus.SUCCESS:
-                return False
-
-            # No new index if the last index attempt is waiting to start
-            if last_index.status == IndexingStatus.NOT_STARTED:
-                return False
-
-            # No new index if the last index attempt is running
-            if last_index.status == IndexingStatus.IN_PROGRESS:
-                return False
-        else:
-            if (
-                connector.id == 0 or connector.source == DocumentSource.INGESTION_API
-            ):  # Ingestion API
-                return False
-        return True
-
-    # If the connector is paused or is the ingestion API, don't index
-    # NOTE: during an embedding model switch over, the following logic
-    # is bypassed by the above check for a future model
-    if (
-        not cc_pair.status.is_active()
-        or connector.id == 0
-        or connector.source == DocumentSource.INGESTION_API
-    ):
-        return False
-
-    if search_settings_primary:
-        if cc_pair.indexing_trigger is not None:
-            # if a manual indexing trigger is on the cc pair, honor it for primary search settings
-            return True
-
-    # if no attempt has ever occurred, we should index regardless of refresh_freq
-    if not last_index:
-        return True
-
-    if connector.refresh_freq is None:
-        return False
-
-    current_db_time = get_db_current_time(db_session)
-    time_since_index = current_db_time - last_index.time_updated
-    if time_since_index.total_seconds() < connector.refresh_freq:
-        return False
-
-    return True
-
-
-def try_creating_indexing_task(
-    celery_app: Celery,
-    cc_pair: ConnectorCredentialPair,
-    search_settings: SearchSettings,
-    reindex: bool,
-    db_session: Session,
-    r: Redis,
-    tenant_id: str | None,
-) -> int | None:
-    """Checks for any conditions that should block the indexing task from being
-    created, then creates the task.
-
-    Does not check for scheduling related conditions as this function
-    is used to trigger indexing immediately.
-    """
-
-    LOCK_TIMEOUT = 30
-    index_attempt_id: int | None = None
-
-    # we need to serialize any attempt to trigger indexing since it can be triggered
-    # either via celery beat or manually (API call)
-    lock: RedisLock = r.lock(
-        DANSWER_REDIS_FUNCTION_LOCK_PREFIX + "try_creating_indexing_task",
-        timeout=LOCK_TIMEOUT,
-    )
-
-    acquired = lock.acquire(blocking_timeout=LOCK_TIMEOUT / 2)
-    if not acquired:
-        return None
-
-    try:
-        redis_connector = RedisConnector(tenant_id, cc_pair.id)
-        redis_connector_index = redis_connector.new_index(search_settings.id)
-
-        # skip if already indexing
-        if redis_connector_index.fenced:
-            return None
-
-        # skip indexing if the cc_pair is deleting
-        if redis_connector.delete.fenced:
-            return None
-
-        db_session.refresh(cc_pair)
-        if cc_pair.status == ConnectorCredentialPairStatus.DELETING:
-            return None
-
-        # add a long running generator task to the queue
-        redis_connector_index.generator_clear()
-
-        # set a basic fence to start
-        payload = RedisConnectorIndexPayload(
-            index_attempt_id=None,
-            started=None,
-            submitted=datetime.now(timezone.utc),
-            celery_task_id=None,
-        )
-
-        redis_connector_index.set_active()
-        redis_connector_index.set_fence(payload)
-
-        # create the index attempt for tracking purposes
-        # code elsewhere checks for index attempts without an associated redis key
-        # and cleans them up
-        # therefore we must create the attempt and the task after the fence goes up
-        index_attempt_id = create_index_attempt(
-            cc_pair.id,
-            search_settings.id,
-            from_beginning=reindex,
-            db_session=db_session,
-        )
-
-        custom_task_id = redis_connector_index.generate_generator_task_id()
-
-        # when the task is sent, we have yet to finish setting up the fence
-        # therefore, the task must contain code that blocks until the fence is ready
-        result = celery_app.send_task(
-            OnyxCeleryTask.CONNECTOR_INDEXING_PROXY_TASK,
-            kwargs=dict(
-                index_attempt_id=index_attempt_id,
-                cc_pair_id=cc_pair.id,
-                search_settings_id=search_settings.id,
-                tenant_id=tenant_id,
-            ),
-            queue=OnyxCeleryQueues.CONNECTOR_INDEXING,
-            task_id=custom_task_id,
-            priority=OnyxCeleryPriority.MEDIUM,
-        )
-        if not result:
-            raise RuntimeError("send_task for connector_indexing_proxy_task failed.")
-
-        # now fill out the fence with the rest of the data
-        redis_connector_index.set_active()
-
-        payload.index_attempt_id = index_attempt_id
-        payload.celery_task_id = result.id
-        redis_connector_index.set_fence(payload)
-    except Exception:
-        task_logger.exception(
-            f"try_creating_indexing_task - Unexpected exception: "
-            f"cc_pair={cc_pair.id} "
-            f"search_settings={search_settings.id}"
-        )
-
-        if index_attempt_id is not None:
-            delete_index_attempt(db_session, index_attempt_id)
-        redis_connector_index.set_fence(None)
-        return None
-    finally:
-        if lock.owned():
-            lock.release()
-
-    return index_attempt_id

backend/onyx/background/celery/tasks/monitoring/tasks.py

@@ -1,458 +0,0 @@
-import json
-from collections.abc import Callable
-from datetime import timedelta
-from typing import Any
-
-from celery import shared_task
-from celery import Task
-from celery.exceptions import SoftTimeLimitExceeded
-from pydantic import BaseModel
-from redis import Redis
-from redis.lock import Lock as RedisLock
-from sqlalchemy import select
-from sqlalchemy.orm import Session
-
-from onyx.background.celery.apps.app_base import task_logger
-from onyx.background.celery.tasks.vespa.tasks import celery_get_queue_length
-from onyx.configs.constants import OnyxCeleryQueues
-from onyx.configs.constants import OnyxCeleryTask
-from onyx.configs.constants import OnyxRedisLocks
-from onyx.db.engine import get_db_current_time
-from onyx.db.engine import get_session_with_tenant
-from onyx.db.enums import IndexingStatus
-from onyx.db.enums import SyncType
-from onyx.db.models import ConnectorCredentialPair
-from onyx.db.models import DocumentSet
-from onyx.db.models import IndexAttempt
-from onyx.db.models import SyncRecord
-from onyx.db.models import UserGroup
-from onyx.redis.redis_pool import get_redis_client
-from onyx.utils.telemetry import optional_telemetry
-from onyx.utils.telemetry import RecordType
-
-_MONITORING_SOFT_TIME_LIMIT = 60 * 5  # 5 minutes
-_MONITORING_TIME_LIMIT = _MONITORING_SOFT_TIME_LIMIT + 60  # 6 minutes
-
-_CONNECTOR_INDEX_ATTEMPT_START_LATENCY_KEY_FMT = (
-    "monitoring_connector_index_attempt_start_latency:{cc_pair_id}:{index_attempt_id}"
-)
-
-_CONNECTOR_INDEX_ATTEMPT_RUN_SUCCESS_KEY_FMT = (
-    "monitoring_connector_index_attempt_run_success:{cc_pair_id}:{index_attempt_id}"
-)
-
-
-def _mark_metric_as_emitted(redis_std: Redis, key: str) -> None:
-    """Mark a metric as having been emitted by setting a Redis key with expiration"""
-    redis_std.set(key, "1", ex=24 * 60 * 60)  # Expire after 1 day
-
-
-def _has_metric_been_emitted(redis_std: Redis, key: str) -> bool:
-    """Check if a metric has been emitted by checking for existence of Redis key"""
-    return bool(redis_std.exists(key))
-
-
-class Metric(BaseModel):
-    key: str | None  # only required if we need to store that we have emitted this metric
-    name: str
-    value: Any
-    tags: dict[str, str]
-
-    def log(self) -> None:
-        """Log the metric in a standardized format"""
-        data = {
-            "metric": self.name,
-            "value": self.value,
-            "tags": self.tags,
-        }
-        task_logger.info(json.dumps(data))
-
-    def emit(self, tenant_id: str | None) -> None:
-        # Convert value to appropriate type based on the input value
-        bool_value = None
-        float_value = None
-        int_value = None
-        string_value = None
-        # NOTE: have to do bool first, since `isinstance(True, int)` is true
-        # e.g. bool is a subclass of int
-        if isinstance(self.value, bool):
-            bool_value = self.value
-        elif isinstance(self.value, int):
-            int_value = self.value
-        elif isinstance(self.value, float):
-            float_value = self.value
-        elif isinstance(self.value, str):
-            string_value = self.value
-        else:
-            task_logger.error(
-                f"Invalid metric value type: {type(self.value)} "
-                f"({self.value}) for metric {self.name}."
-            )
-            return
-
-        # don't send None values over the wire
-        data = {
-            k: v
-            for k, v in {
-                "metric_name": self.name,
-                "float_value": float_value,
-                "int_value": int_value,
-                "string_value": string_value,
-                "bool_value": bool_value,
-                "tags": self.tags,
-            }.items()
-            if v is not None
-        }
-        optional_telemetry(
-            record_type=RecordType.METRIC,
-            data=data,
-            tenant_id=tenant_id,
-        )
-
-
-def _collect_queue_metrics(redis_celery: Redis) -> list[Metric]:
-    """Collect metrics about queue lengths for different Celery queues"""
-    metrics = []
-    queue_mappings = {
-        "celery_queue_length": "celery",
-        "indexing_queue_length": "indexing",
-        "sync_queue_length": "sync",
-        "deletion_queue_length": "deletion",
-        "pruning_queue_length": "pruning",
-        "permissions_sync_queue_length": OnyxCeleryQueues.CONNECTOR_DOC_PERMISSIONS_SYNC,
-        "external_group_sync_queue_length": OnyxCeleryQueues.CONNECTOR_EXTERNAL_GROUP_SYNC,
-        "permissions_upsert_queue_length": OnyxCeleryQueues.DOC_PERMISSIONS_UPSERT,
-    }
-
-    for name, queue in queue_mappings.items():
-        metrics.append(
-            Metric(
-                key=None,
-                name=name,
-                value=celery_get_queue_length(queue, redis_celery),
-                tags={"queue": name},
-            )
-        )
-
-    return metrics
-
-
-def _build_connector_start_latency_metric(
-    cc_pair: ConnectorCredentialPair,
-    recent_attempt: IndexAttempt,
-    second_most_recent_attempt: IndexAttempt | None,
-    redis_std: Redis,
-) -> Metric | None:
-    if not recent_attempt.time_started:
-        return None
-
-    # check if we already emitted a metric for this index attempt
-    metric_key = _CONNECTOR_INDEX_ATTEMPT_START_LATENCY_KEY_FMT.format(
-        cc_pair_id=cc_pair.id,
-        index_attempt_id=recent_attempt.id,
-    )
-    if _has_metric_been_emitted(redis_std, metric_key):
-        task_logger.info(
-            f"Skipping metric for connector {cc_pair.connector.id} "
-            f"index attempt {recent_attempt.id} because it has already been "
-            "emitted"
-        )
-        return None
-
-    # Connector start latency
-    # first run case - we should start as soon as it's created
-    if not second_most_recent_attempt:
-        desired_start_time = cc_pair.connector.time_created
-    else:
-        if not cc_pair.connector.refresh_freq:
-            task_logger.error(
-                "Found non-initial index attempt for connector "
-                "without refresh_freq. This should never happen."
-            )
-            return None
-
-        desired_start_time = second_most_recent_attempt.time_updated + timedelta(
-            seconds=cc_pair.connector.refresh_freq
-        )
-
-    start_latency = (recent_attempt.time_started - desired_start_time).total_seconds()
-
-    return Metric(
-        key=metric_key,
-        name="connector_start_latency",
-        value=start_latency,
-        tags={},
-    )
-
-
-def _build_run_success_metrics(
-    cc_pair: ConnectorCredentialPair,
-    recent_attempts: list[IndexAttempt],
-    redis_std: Redis,
-) -> list[Metric]:
-    metrics = []
-    for attempt in recent_attempts:
-        metric_key = _CONNECTOR_INDEX_ATTEMPT_RUN_SUCCESS_KEY_FMT.format(
-            cc_pair_id=cc_pair.id,
-            index_attempt_id=attempt.id,
-        )
-
-        if _has_metric_been_emitted(redis_std, metric_key):
-            task_logger.info(
-                f"Skipping metric for connector {cc_pair.connector.id} "
-                f"index attempt {attempt.id} because it has already been "
-                "emitted"
-            )
-            continue
-
-        if attempt.status in [
-            IndexingStatus.SUCCESS,
-            IndexingStatus.FAILED,
-            IndexingStatus.CANCELED,
-        ]:
-            metrics.append(
-                Metric(
-                    key=metric_key,
-                    name="connector_run_succeeded",
-                    value=attempt.status == IndexingStatus.SUCCESS,
-                    tags={"source": str(cc_pair.connector.source)},
-                )
-            )
-
-    return metrics
-
-
-def _collect_connector_metrics(db_session: Session, redis_std: Redis) -> list[Metric]:
-    """Collect metrics about connector runs from the past hour"""
-    # NOTE: use get_db_current_time since the IndexAttempt times are set based on DB time
-    one_hour_ago = get_db_current_time(db_session) - timedelta(hours=1)
-
-    # Get all connector credential pairs
-    cc_pairs = db_session.scalars(select(ConnectorCredentialPair)).all()
-
-    metrics = []
-    for cc_pair in cc_pairs:
-        # Get all attempts in the last hour
-        recent_attempts = (
-            db_session.query(IndexAttempt)
-            .filter(
-                IndexAttempt.connector_credential_pair_id == cc_pair.id,
-                IndexAttempt.time_created >= one_hour_ago,
-            )
-            .order_by(IndexAttempt.time_created.desc())
-            .all()
-        )
-        most_recent_attempt = recent_attempts[0] if recent_attempts else None
-        second_most_recent_attempt = (
-            recent_attempts[1] if len(recent_attempts) > 1 else None
-        )
-
-        # if no metric to emit, skip
-        if most_recent_attempt is None:
-            continue
-
-        # Connector start latency
-        start_latency_metric = _build_connector_start_latency_metric(
-            cc_pair, most_recent_attempt, second_most_recent_attempt, redis_std
-        )
-        if start_latency_metric:
-            metrics.append(start_latency_metric)
-
-        # Connector run success/failure
-        run_success_metrics = _build_run_success_metrics(
-            cc_pair, recent_attempts, redis_std
-        )
-        metrics.extend(run_success_metrics)
-
-    return metrics
-
-
-def _collect_sync_metrics(db_session: Session, redis_std: Redis) -> list[Metric]:
-    """Collect metrics about document set and group syncing speed"""
-    # NOTE: use get_db_current_time since the SyncRecord times are set based on DB time
-    one_hour_ago = get_db_current_time(db_session) - timedelta(hours=1)
-
-    # Get all sync records from the last hour
-    recent_sync_records = db_session.scalars(
-        select(SyncRecord)
-        .where(SyncRecord.sync_start_time >= one_hour_ago)
-        .order_by(SyncRecord.sync_start_time.desc())
-    ).all()
-
-    metrics = []
-    for sync_record in recent_sync_records:
-        # Skip if no end time (sync still in progress)
-        if not sync_record.sync_end_time:
-            continue
-
-        # Check if we already emitted a metric for this sync record
-        metric_key = (
-            f"sync_speed:{sync_record.sync_type}:"
-            f"{sync_record.entity_id}:{sync_record.id}"
-        )
-        if _has_metric_been_emitted(redis_std, metric_key):
-            task_logger.debug(
-                f"Skipping metric for sync record {sync_record.id} "
-                "because it has already been emitted"
-            )
-            continue
-
-        # Calculate sync duration in minutes
-        sync_duration_mins = (
-            sync_record.sync_end_time - sync_record.sync_start_time
-        ).total_seconds() / 60.0
-
-        # Calculate sync speed (docs/min) - avoid division by zero
-        sync_speed = (
-            sync_record.num_docs_synced / sync_duration_mins
-            if sync_duration_mins > 0
-            else None
-        )
-
-        if sync_speed is None:
-            task_logger.error(
-                "Something went wrong with sync speed calculation. "
-                f"Sync record: {sync_record.id}"
-            )
-            continue
-
-        metrics.append(
-            Metric(
-                key=metric_key,
-                name="sync_speed_docs_per_min",
-                value=sync_speed,
-                tags={
-                    "sync_type": str(sync_record.sync_type),
-                    "status": str(sync_record.sync_status),
-                },
-            )
-        )
-
-        # Add sync start latency metric
-        start_latency_key = (
-            f"sync_start_latency:{sync_record.sync_type}"
-            f":{sync_record.entity_id}:{sync_record.id}"
-        )
-        if _has_metric_been_emitted(redis_std, start_latency_key):
-            task_logger.debug(
-                f"Skipping start latency metric for sync record {sync_record.id} "
-                "because it has already been emitted"
-            )
-            continue
-
-        # Get the entity's last update time based on sync type
-        entity: DocumentSet | UserGroup | None = None
-        if sync_record.sync_type == SyncType.DOCUMENT_SET:
-            entity = db_session.scalar(
-                select(DocumentSet).where(DocumentSet.id == sync_record.entity_id)
-            )
-        elif sync_record.sync_type == SyncType.USER_GROUP:
-            entity = db_session.scalar(
-                select(UserGroup).where(UserGroup.id == sync_record.entity_id)
-            )
-        else:
-            # Skip other sync types
-            task_logger.debug(
-                f"Skipping sync record {sync_record.id} "
-                f"with type {sync_record.sync_type} "
-                f"and id {sync_record.entity_id} "
-                "because it is not a document set or user group"
-            )
-            continue
-
-        if entity is None:
-            task_logger.error(
-                f"Could not find entity for sync record {sync_record.id} "
-                f"with type {sync_record.sync_type} and id {sync_record.entity_id}"
-            )
-            continue
-
-        # Calculate start latency in seconds
-        start_latency = (
-            sync_record.sync_start_time - entity.time_last_modified_by_user
-        ).total_seconds()
-        if start_latency < 0:
-            task_logger.error(
-                f"Start latency is negative for sync record {sync_record.id} "
-                f"with type {sync_record.sync_type} and id {sync_record.entity_id}."
-                "This is likely because the entity was updated between the time the "
-                "time the sync finished and this job ran. Skipping."
-            )
-            continue
-
-        metrics.append(
-            Metric(
-                key=start_latency_key,
-                name="sync_start_latency_seconds",
-                value=start_latency,
-                tags={
-                    "sync_type": str(sync_record.sync_type),
-                },
-            )
-        )
-
-    return metrics
-
-
-@shared_task(
-    name=OnyxCeleryTask.MONITOR_BACKGROUND_PROCESSES,
-    soft_time_limit=_MONITORING_SOFT_TIME_LIMIT,
-    time_limit=_MONITORING_TIME_LIMIT,
-    queue=OnyxCeleryQueues.MONITORING,
-    bind=True,
-)
-def monitor_background_processes(self: Task, *, tenant_id: str | None) -> None:
-    """Collect and emit metrics about background processes.
-    This task runs periodically to gather metrics about:
-    - Queue lengths for different Celery queues
-    - Connector run metrics (start latency, success rate)
-    - Syncing speed metrics
-    - Worker status and task counts
-    """
-    task_logger.info("Starting background monitoring")
-    r = get_redis_client(tenant_id=tenant_id)
-
-    lock_monitoring: RedisLock = r.lock(
-        OnyxRedisLocks.MONITOR_BACKGROUND_PROCESSES_LOCK,
-        timeout=_MONITORING_SOFT_TIME_LIMIT,
-    )
-
-    # these tasks should never overlap
-    if not lock_monitoring.acquire(blocking=False):
-        task_logger.info("Skipping monitoring task because it is already running")
-        return None
-
-    try:
-        # Get Redis client for Celery broker
-        redis_celery = self.app.broker_connection().channel().client  # type: ignore
-        redis_std = get_redis_client(tenant_id=tenant_id)
-
-        # Define metric collection functions and their dependencies
-        metric_functions: list[Callable[[], list[Metric]]] = [
-            lambda: _collect_queue_metrics(redis_celery),
-            lambda: _collect_connector_metrics(db_session, redis_std),
-            lambda: _collect_sync_metrics(db_session, redis_std),
-        ]
-        # Collect and log each metric
-        with get_session_with_tenant(tenant_id) as db_session:
-            for metric_fn in metric_functions:
-                metrics = metric_fn()
-                for metric in metrics:
-                    metric.log()
-                    metric.emit(tenant_id)
-                    if metric.key:
-                        _mark_metric_as_emitted(redis_std, metric.key)
-
-        task_logger.info("Successfully collected background metrics")
-    except SoftTimeLimitExceeded:
-        task_logger.info(
-            "Soft time limit exceeded, task is being terminated gracefully."
-        )
-    except Exception as e:
-        task_logger.exception("Error collecting background process metrics")
-        raise e
-    finally:
-        if lock_monitoring.owned():
-            lock_monitoring.release()
-
-        task_logger.info("Background monitoring task finished")

backend/onyx/background/celery/tasks/pruning/tasks.py

@@ -13,11 +13,11 @@ from sqlalchemy.orm import Session
 
 from onyx.background.celery.apps.app_base import task_logger
 from onyx.background.celery.celery_utils import extract_ids_from_runnable_connector
-from onyx.background.celery.tasks.indexing.utils import IndexingCallback
+from onyx.background.celery.tasks.indexing.tasks import IndexingCallback
 from onyx.configs.app_configs import ALLOW_SIMULTANEOUS_PRUNING
 from onyx.configs.app_configs import JOB_TIMEOUT
-from onyx.configs.constants import CELERY_GENERIC_BEAT_LOCK_TIMEOUT
 from onyx.configs.constants import CELERY_PRUNING_LOCK_TIMEOUT
+from onyx.configs.constants import CELERY_VESPA_SYNC_BEAT_LOCK_TIMEOUT
 from onyx.configs.constants import DANSWER_REDIS_FUNCTION_LOCK_PREFIX
 from onyx.configs.constants import OnyxCeleryPriority
 from onyx.configs.constants import OnyxCeleryQueues
@@ -86,14 +86,14 @@ def check_for_pruning(self: Task, *, tenant_id: str | None) -> bool | None:
 
     lock_beat: RedisLock = r.lock(
         OnyxRedisLocks.CHECK_PRUNE_BEAT_LOCK,
-        timeout=CELERY_GENERIC_BEAT_LOCK_TIMEOUT,
+        timeout=CELERY_VESPA_SYNC_BEAT_LOCK_TIMEOUT,
     )
 
-    # these tasks should never overlap
-    if not lock_beat.acquire(blocking=False):
-        return None
-
     try:
+        # these tasks should never overlap
+        if not lock_beat.acquire(blocking=False):
+            return None
+
         cc_pair_ids: list[int] = []
         with get_session_with_tenant(tenant_id) as db_session:
             cc_pairs = get_connector_credential_pairs(db_session)
@@ -103,10 +103,7 @@ def check_for_pruning(self: Task, *, tenant_id: str | None) -> bool | None:
         for cc_pair_id in cc_pair_ids:
             lock_beat.reacquire()
             with get_session_with_tenant(tenant_id) as db_session:
-                cc_pair = get_connector_credential_pair_from_id(
-                    db_session=db_session,
-                    cc_pair_id=cc_pair_id,
-                )
+                cc_pair = get_connector_credential_pair_from_id(cc_pair_id, db_session)
                 if not cc_pair:
                     continue
 

backend/onyx/background/celery/tasks/shared/RetryDocumentIndex.py

@@ -28,35 +28,13 @@ class RetryDocumentIndex:
         wait=wait_random_exponential(multiplier=1, max=MAX_WAIT),
         stop=stop_after_delay(STOP_AFTER),
     )
-    def delete_single(
-        self,
-        doc_id: str,
-        *,
-        tenant_id: str | None,
-        chunk_count: int | None,
-    ) -> int:
-        return self.index.delete_single(
-            doc_id,
-            tenant_id=tenant_id,
-            chunk_count=chunk_count,
-        )
+    def delete_single(self, doc_id: str) -> int:
+        return self.index.delete_single(doc_id)
 
     @retry(
         retry=retry_if_exception_type(httpx.ReadTimeout),
         wait=wait_random_exponential(multiplier=1, max=MAX_WAIT),
         stop=stop_after_delay(STOP_AFTER),
     )
-    def update_single(
-        self,
-        doc_id: str,
-        *,
-        tenant_id: str | None,
-        chunk_count: int | None,
-        fields: VespaDocumentFields,
-    ) -> int:
-        return self.index.update_single(
-            doc_id,
-            tenant_id=tenant_id,
-            chunk_count=chunk_count,
-            fields=fields,
-        )
+    def update_single(self, doc_id: str, fields: VespaDocumentFields) -> int:
+        return self.index.update_single(doc_id, fields)

backend/onyx/background/celery/tasks/shared/tasks.py

@@ -12,7 +12,6 @@ from onyx.background.celery.tasks.shared.RetryDocumentIndex import RetryDocument
 from onyx.configs.constants import OnyxCeleryTask
 from onyx.db.document import delete_document_by_connector_credential_pair__no_commit
 from onyx.db.document import delete_documents_complete__no_commit
-from onyx.db.document import fetch_chunk_count_for_document
 from onyx.db.document import get_document
 from onyx.db.document import get_document_connector_count
 from onyx.db.document import mark_document_as_modified
@@ -81,13 +80,7 @@ def document_by_cc_pair_cleanup_task(
                 # delete it from vespa and the db
                 action = "delete"
 
-                chunk_count = fetch_chunk_count_for_document(document_id, db_session)
-
-                chunks_affected = retry_index.delete_single(
-                    document_id,
-                    tenant_id=tenant_id,
-                    chunk_count=chunk_count,
-                )
+                chunks_affected = retry_index.delete_single(document_id)
                 delete_documents_complete__no_commit(
                     db_session=db_session,
                     document_ids=[document_id],
@@ -117,12 +110,7 @@ def document_by_cc_pair_cleanup_task(
                 )
 
                 # update Vespa. OK if doc doesn't exist. Raises exception otherwise.
-                chunks_affected = retry_index.update_single(
-                    document_id,
-                    tenant_id=tenant_id,
-                    chunk_count=doc.chunk_count,
-                    fields=fields,
-                )
+                chunks_affected = retry_index.update_single(document_id, fields=fields)
 
                 # there are still other cc_pair references to the doc, so just resync to Vespa
                 delete_document_by_connector_credential_pair__no_commit(

backend/onyx/background/celery/tasks/vespa/tasks.py

@@ -1,11 +1,9 @@
 import random
 import time
 import traceback
-from collections.abc import Callable
 from datetime import datetime
 from datetime import timezone
 from http import HTTPStatus
-from typing import Any
 from typing import cast
 
 import httpx
@@ -28,7 +26,6 @@ from onyx.background.celery.tasks.shared.RetryDocumentIndex import RetryDocument
 from onyx.background.celery.tasks.shared.tasks import LIGHT_SOFT_TIME_LIMIT
 from onyx.background.celery.tasks.shared.tasks import LIGHT_TIME_LIMIT
 from onyx.configs.app_configs import JOB_TIMEOUT
-from onyx.configs.app_configs import VESPA_SYNC_MAX_TASKS
 from onyx.configs.constants import CELERY_VESPA_SYNC_BEAT_LOCK_TIMEOUT
 from onyx.configs.constants import OnyxCeleryQueues
 from onyx.configs.constants import OnyxCeleryTask
@@ -54,16 +51,10 @@ from onyx.db.document_set import get_document_set_by_id
 from onyx.db.document_set import mark_document_set_as_synced
 from onyx.db.engine import get_session_with_tenant
 from onyx.db.enums import IndexingStatus
-from onyx.db.enums import SyncStatus
-from onyx.db.enums import SyncType
 from onyx.db.index_attempt import delete_index_attempts
 from onyx.db.index_attempt import get_index_attempt
 from onyx.db.index_attempt import mark_attempt_failed
 from onyx.db.models import DocumentSet
-from onyx.db.models import UserGroup
-from onyx.db.sync_record import cleanup_sync_records
-from onyx.db.sync_record import insert_sync_record
-from onyx.db.sync_record import update_sync_record_status
 from onyx.document_index.document_index_utils import get_both_index_names
 from onyx.document_index.factory import get_default_document_index
 from onyx.document_index.interfaces import VespaDocumentFields
@@ -79,7 +70,6 @@ from onyx.redis.redis_connector_prune import RedisConnectorPrune
 from onyx.redis.redis_document_set import RedisDocumentSet
 from onyx.redis.redis_pool import get_redis_client
 from onyx.redis.redis_pool import redis_lock_dump
-from onyx.redis.redis_pool import SCAN_ITER_COUNT_DEFAULT
 from onyx.redis.redis_usergroup import RedisUserGroup
 from onyx.utils.logger import setup_logger
 from onyx.utils.variable_functionality import fetch_versioned_implementation
@@ -113,14 +103,14 @@ def check_for_vespa_sync_task(self: Task, *, tenant_id: str | None) -> bool | No
         timeout=CELERY_VESPA_SYNC_BEAT_LOCK_TIMEOUT,
     )
 
-    # these tasks should never overlap
-    if not lock_beat.acquire(blocking=False):
-        return None
-
     try:
+        # these tasks should never overlap
+        if not lock_beat.acquire(blocking=False):
+            return None
+
         with get_session_with_tenant(tenant_id) as db_session:
             try_generate_stale_document_sync_tasks(
-                self.app, VESPA_SYNC_MAX_TASKS, db_session, r, lock_beat, tenant_id
+                self.app, db_session, r, lock_beat, tenant_id
             )
 
         # region document set scan
@@ -195,7 +185,6 @@ def check_for_vespa_sync_task(self: Task, *, tenant_id: str | None) -> bool | No
 
 def try_generate_stale_document_sync_tasks(
     celery_app: Celery,
-    max_tasks: int,
     db_session: Session,
     r: Redis,
     lock_beat: RedisLock,
@@ -226,16 +215,11 @@ def try_generate_stale_document_sync_tasks(
     # rkuo: we could technically sync all stale docs in one big pass.
     # but I feel it's more understandable to group the docs by cc_pair
     total_tasks_generated = 0
-    tasks_remaining = max_tasks
     cc_pairs = get_connector_credential_pairs(db_session)
     for cc_pair in cc_pairs:
-        lock_beat.reacquire()
-
         rc = RedisConnectorCredentialPair(tenant_id, cc_pair.id)
         rc.set_skip_docs(docs_to_skip)
-        result = rc.generate_tasks(
-            tasks_remaining, celery_app, db_session, r, lock_beat, tenant_id
-        )
+        result = rc.generate_tasks(celery_app, db_session, r, lock_beat, tenant_id)
 
         if result is None:
             continue
@@ -249,19 +233,10 @@ def try_generate_stale_document_sync_tasks(
         )
 
         total_tasks_generated += result[0]
-        tasks_remaining -= result[0]
-        if tasks_remaining <= 0:
-            break
 
-    if tasks_remaining <= 0:
-        task_logger.info(
-            f"RedisConnector.generate_tasks reached the task generation limit: "
-            f"total_tasks_generated={total_tasks_generated} max_tasks={max_tasks}"
-        )
-    else:
-        task_logger.info(
-            f"RedisConnector.generate_tasks finished for all cc_pairs. total_tasks_generated={total_tasks_generated}"
-        )
+    task_logger.info(
+        f"RedisConnector.generate_tasks finished for all cc_pairs. total_tasks_generated={total_tasks_generated}"
+    )
 
     r.set(RedisConnectorCredentialPair.get_fence_key(), total_tasks_generated)
     return total_tasks_generated
@@ -285,21 +260,11 @@ def try_generate_document_set_sync_tasks(
 
     # don't generate sync tasks if we're up to date
     # race condition with the monitor/cleanup function if we use a cached result!
-    document_set = get_document_set_by_id(
-        db_session=db_session,
-        document_set_id=document_set_id,
-    )
+    document_set = get_document_set_by_id(db_session, document_set_id)
     if not document_set:
         return None
 
     if document_set.is_up_to_date:
-        # there should be no in-progress sync records if this is up to date
-        # clean it up just in case things got into a bad state
-        cleanup_sync_records(
-            db_session=db_session,
-            entity_id=document_set_id,
-            sync_type=SyncType.DOCUMENT_SET,
-        )
         return None
 
     # add tasks to celery and build up the task set to monitor in redis
@@ -310,9 +275,7 @@ def try_generate_document_set_sync_tasks(
     )
 
     # Add all documents that need to be updated into the queue
-    result = rds.generate_tasks(
-        VESPA_SYNC_MAX_TASKS, celery_app, db_session, r, lock_beat, tenant_id
-    )
+    result = rds.generate_tasks(celery_app, db_session, r, lock_beat, tenant_id)
     if result is None:
         return None
 
@@ -328,13 +291,6 @@ def try_generate_document_set_sync_tasks(
         f"document_set={document_set.id} tasks_generated={tasks_generated}"
     )
 
-    # create before setting fence to avoid race condition where the monitoring
-    # task updates the sync record before it is created
-    insert_sync_record(
-        db_session=db_session,
-        entity_id=document_set_id,
-        sync_type=SyncType.DOCUMENT_SET,
-    )
     # set this only after all tasks have been added
     rds.set_fence(tasks_generated)
     return tasks_generated
@@ -356,9 +312,8 @@ def try_generate_user_group_sync_tasks(
         return None
 
     # race condition with the monitor/cleanup function if we use a cached result!
-    fetch_user_group = cast(
-        Callable[[Session, int], UserGroup | None],
-        fetch_versioned_implementation("onyx.db.user_group", "fetch_user_group"),
+    fetch_user_group = fetch_versioned_implementation(
+        "onyx.db.user_group", "fetch_user_group"
     )
 
     usergroup = fetch_user_group(db_session, usergroup_id)
@@ -366,13 +321,6 @@ def try_generate_user_group_sync_tasks(
         return None
 
     if usergroup.is_up_to_date:
-        # there should be no in-progress sync records if this is up to date
-        # clean it up just in case things got into a bad state
-        cleanup_sync_records(
-            db_session=db_session,
-            entity_id=usergroup_id,
-            sync_type=SyncType.USER_GROUP,
-        )
         return None
 
     # add tasks to celery and build up the task set to monitor in redis
@@ -382,9 +330,7 @@ def try_generate_user_group_sync_tasks(
     task_logger.info(
         f"RedisUserGroup.generate_tasks starting. usergroup_id={usergroup.id}"
     )
-    result = rug.generate_tasks(
-        VESPA_SYNC_MAX_TASKS, celery_app, db_session, r, lock_beat, tenant_id
-    )
+    result = rug.generate_tasks(celery_app, db_session, r, lock_beat, tenant_id)
     if result is None:
         return None
 
@@ -400,16 +346,8 @@ def try_generate_user_group_sync_tasks(
         f"usergroup={usergroup.id} tasks_generated={tasks_generated}"
     )
 
-    # create before setting fence to avoid race condition where the monitoring
-    # task updates the sync record before it is created
-    insert_sync_record(
-        db_session=db_session,
-        entity_id=usergroup_id,
-        sync_type=SyncType.USER_GROUP,
-    )
     # set this only after all tasks have been added
     rug.set_fence(tasks_generated)
-
     return tasks_generated
 
 
@@ -459,13 +397,6 @@ def monitor_document_set_taskset(
         f"remaining={count} initial={initial_count}"
     )
     if count > 0:
-        update_sync_record_status(
-            db_session=db_session,
-            entity_id=document_set_id,
-            sync_type=SyncType.DOCUMENT_SET,
-            sync_status=SyncStatus.IN_PROGRESS,
-            num_docs_synced=count,
-        )
         return
 
     document_set = cast(
@@ -484,13 +415,6 @@ def monitor_document_set_taskset(
             task_logger.info(
                 f"Successfully synced document set: document_set={document_set_id}"
             )
-        update_sync_record_status(
-            db_session=db_session,
-            entity_id=document_set_id,
-            sync_type=SyncType.DOCUMENT_SET,
-            sync_status=SyncStatus.SUCCESS,
-            num_docs_synced=initial_count,
-        )
 
     rds.reset()
 
@@ -524,21 +448,10 @@ def monitor_connector_deletion_taskset(
         f"Connector deletion progress: cc_pair={cc_pair_id} remaining={remaining} initial={fence_data.num_tasks}"
     )
     if remaining > 0:
-        with get_session_with_tenant(tenant_id) as db_session:
-            update_sync_record_status(
-                db_session=db_session,
-                entity_id=cc_pair_id,
-                sync_type=SyncType.CONNECTOR_DELETION,
-                sync_status=SyncStatus.IN_PROGRESS,
-                num_docs_synced=remaining,
-            )
         return
 
     with get_session_with_tenant(tenant_id) as db_session:
-        cc_pair = get_connector_credential_pair_from_id(
-            db_session=db_session,
-            cc_pair_id=cc_pair_id,
-        )
+        cc_pair = get_connector_credential_pair_from_id(cc_pair_id, db_session)
         if not cc_pair:
             task_logger.warning(
                 f"Connector deletion - cc_pair not found: cc_pair={cc_pair_id}"
@@ -610,29 +523,11 @@ def monitor_connector_deletion_taskset(
                 )
                 db_session.delete(connector)
             db_session.commit()
-
-            update_sync_record_status(
-                db_session=db_session,
-                entity_id=cc_pair_id,
-                sync_type=SyncType.CONNECTOR_DELETION,
-                sync_status=SyncStatus.SUCCESS,
-                num_docs_synced=fence_data.num_tasks,
-            )
-
         except Exception as e:
             db_session.rollback()
             stack_trace = traceback.format_exc()
             error_message = f"Error: {str(e)}\n\nStack Trace:\n{stack_trace}"
             add_deletion_failure_message(db_session, cc_pair_id, error_message)
-
-            update_sync_record_status(
-                db_session=db_session,
-                entity_id=cc_pair_id,
-                sync_type=SyncType.CONNECTOR_DELETION,
-                sync_status=SyncStatus.FAILED,
-                num_docs_synced=fence_data.num_tasks,
-            )
-
             task_logger.exception(
                 f"Connector deletion exceptioned: "
                 f"cc_pair={cc_pair_id} connector={cc_pair.connector_id} credential={cc_pair.credential_id}"
@@ -857,7 +752,7 @@ def monitor_ccpair_indexing_taskset(
 
 
 @shared_task(name=OnyxCeleryTask.MONITOR_VESPA_SYNC, soft_time_limit=300, bind=True)
-def monitor_vespa_sync(self: Task, tenant_id: str | None) -> bool | None:
+def monitor_vespa_sync(self: Task, tenant_id: str | None) -> bool:
     """This is a celery beat task that monitors and finalizes metadata sync tasksets.
     It scans for fence values and then gets the counts of any associated tasksets.
     If the count is 0, that means all tasks finished and we should clean up.
@@ -871,7 +766,7 @@ def monitor_vespa_sync(self: Task, tenant_id: str | None) -> bool | None:
 
     time_start = time.monotonic()
 
-    timings: dict[str, Any] = {}
+    timings: dict[str, float] = {}
     timings["start"] = time_start
 
     r = get_redis_client(tenant_id=tenant_id)
@@ -881,15 +776,16 @@ def monitor_vespa_sync(self: Task, tenant_id: str | None) -> bool | None:
         timeout=CELERY_VESPA_SYNC_BEAT_LOCK_TIMEOUT,
     )
 
-    # prevent overlapping tasks
-    if not lock_beat.acquire(blocking=False):
-        return None
-
     try:
+        # prevent overlapping tasks
+        if not lock_beat.acquire(blocking=False):
+            task_logger.info("monitor_vespa_sync exiting due to overlap")
+            return False
+
         # print current queue lengths
         phase_start = time.monotonic()
         # we don't need every tenant polling redis for this info.
-        if not MULTI_TENANT or random.randint(1, 10) == 10:
+        if not MULTI_TENANT or random.randint(1, 100) == 100:
             r_celery = self.app.broker_connection().channel().client  # type: ignore
             n_celery = celery_get_queue_length("celery", r_celery)
             n_indexing = celery_get_queue_length(
@@ -930,7 +826,6 @@ def monitor_vespa_sync(self: Task, tenant_id: str | None) -> bool | None:
                 f"permissions_upsert={n_permissions_upsert} "
             )
         timings["queues"] = time.monotonic() - phase_start
-        timings["queues_ttl"] = r.ttl(OnyxRedisLocks.MONITOR_VESPA_SYNC_BEAT_LOCK)
 
         # scan and monitor activity to completion
         phase_start = time.monotonic()
@@ -938,37 +833,24 @@ def monitor_vespa_sync(self: Task, tenant_id: str | None) -> bool | None:
         if r.exists(RedisConnectorCredentialPair.get_fence_key()):
             monitor_connector_taskset(r)
         timings["connector"] = time.monotonic() - phase_start
-        timings["connector_ttl"] = r.ttl(OnyxRedisLocks.MONITOR_VESPA_SYNC_BEAT_LOCK)
 
         phase_start = time.monotonic()
-        lock_beat.reacquire()
-        for key_bytes in r.scan_iter(
-            RedisConnectorDelete.FENCE_PREFIX + "*", count=SCAN_ITER_COUNT_DEFAULT
-        ):
-            monitor_connector_deletion_taskset(tenant_id, key_bytes, r)
+        for key_bytes in r.scan_iter(RedisConnectorDelete.FENCE_PREFIX + "*"):
             lock_beat.reacquire()
+            monitor_connector_deletion_taskset(tenant_id, key_bytes, r)
 
         timings["connector_deletion"] = time.monotonic() - phase_start
-        timings["connector_deletion_ttl"] = r.ttl(
-            OnyxRedisLocks.MONITOR_VESPA_SYNC_BEAT_LOCK
-        )
 
         phase_start = time.monotonic()
-        lock_beat.reacquire()
-        for key_bytes in r.scan_iter(
-            RedisDocumentSet.FENCE_PREFIX + "*", count=SCAN_ITER_COUNT_DEFAULT
-        ):
+        for key_bytes in r.scan_iter(RedisDocumentSet.FENCE_PREFIX + "*"):
+            lock_beat.reacquire()
             with get_session_with_tenant(tenant_id) as db_session:
                 monitor_document_set_taskset(tenant_id, key_bytes, r, db_session)
-            lock_beat.reacquire()
-        timings["documentset"] = time.monotonic() - phase_start
-        timings["documentset_ttl"] = r.ttl(OnyxRedisLocks.MONITOR_VESPA_SYNC_BEAT_LOCK)
+        timings["document_set"] = time.monotonic() - phase_start
 
         phase_start = time.monotonic()
-        lock_beat.reacquire()
-        for key_bytes in r.scan_iter(
-            RedisUserGroup.FENCE_PREFIX + "*", count=SCAN_ITER_COUNT_DEFAULT
-        ):
+        for key_bytes in r.scan_iter(RedisUserGroup.FENCE_PREFIX + "*"):
+            lock_beat.reacquire()
             monitor_usergroup_taskset = fetch_versioned_implementation_with_fallback(
                 "onyx.background.celery.tasks.vespa.tasks",
                 "monitor_usergroup_taskset",
@@ -976,45 +858,29 @@ def monitor_vespa_sync(self: Task, tenant_id: str | None) -> bool | None:
             )
             with get_session_with_tenant(tenant_id) as db_session:
                 monitor_usergroup_taskset(tenant_id, key_bytes, r, db_session)
-            lock_beat.reacquire()
         timings["usergroup"] = time.monotonic() - phase_start
-        timings["usergroup_ttl"] = r.ttl(OnyxRedisLocks.MONITOR_VESPA_SYNC_BEAT_LOCK)
 
         phase_start = time.monotonic()
-        lock_beat.reacquire()
-        for key_bytes in r.scan_iter(
-            RedisConnectorPrune.FENCE_PREFIX + "*", count=SCAN_ITER_COUNT_DEFAULT
-        ):
+        for key_bytes in r.scan_iter(RedisConnectorPrune.FENCE_PREFIX + "*"):
+            lock_beat.reacquire()
             with get_session_with_tenant(tenant_id) as db_session:
                 monitor_ccpair_pruning_taskset(tenant_id, key_bytes, r, db_session)
-            lock_beat.reacquire()
         timings["pruning"] = time.monotonic() - phase_start
-        timings["pruning_ttl"] = r.ttl(OnyxRedisLocks.MONITOR_VESPA_SYNC_BEAT_LOCK)
 
         phase_start = time.monotonic()
-        lock_beat.reacquire()
-        for key_bytes in r.scan_iter(
-            RedisConnectorIndex.FENCE_PREFIX + "*", count=SCAN_ITER_COUNT_DEFAULT
-        ):
+        for key_bytes in r.scan_iter(RedisConnectorIndex.FENCE_PREFIX + "*"):
+            lock_beat.reacquire()
             with get_session_with_tenant(tenant_id) as db_session:
                 monitor_ccpair_indexing_taskset(tenant_id, key_bytes, r, db_session)
-            lock_beat.reacquire()
         timings["indexing"] = time.monotonic() - phase_start
-        timings["indexing_ttl"] = r.ttl(OnyxRedisLocks.MONITOR_VESPA_SYNC_BEAT_LOCK)
 
         phase_start = time.monotonic()
-        lock_beat.reacquire()
-        for key_bytes in r.scan_iter(
-            RedisConnectorPermissionSync.FENCE_PREFIX + "*",
-            count=SCAN_ITER_COUNT_DEFAULT,
-        ):
+        for key_bytes in r.scan_iter(RedisConnectorPermissionSync.FENCE_PREFIX + "*"):
+            lock_beat.reacquire()
             with get_session_with_tenant(tenant_id) as db_session:
                 monitor_ccpair_permissions_taskset(tenant_id, key_bytes, r, db_session)
-            lock_beat.reacquire()
 
         timings["permissions"] = time.monotonic() - phase_start
-        timings["permissions_ttl"] = r.ttl(OnyxRedisLocks.MONITOR_VESPA_SYNC_BEAT_LOCK)
-
     except SoftTimeLimitExceeded:
         task_logger.info(
             "Soft time limit exceeded, task is being terminated gracefully."
@@ -1023,10 +889,18 @@ def monitor_vespa_sync(self: Task, tenant_id: str | None) -> bool | None:
         if lock_beat.owned():
             lock_beat.release()
         else:
+            t = timings
             task_logger.error(
                 "monitor_vespa_sync - Lock not owned on completion: "
                 f"tenant={tenant_id} "
-                f"timings={timings}"
+                f"queues={t.get('queues')} "
+                f"connector={t.get('connector')} "
+                f"connector_deletion={t.get('connector_deletion')} "
+                f"document_set={t.get('document_set')} "
+                f"usergroup={t.get('usergroup')} "
+                f"pruning={t.get('pruning')} "
+                f"indexing={t.get('indexing')} "
+                f"permissions={t.get('permissions')}"
             )
             redis_lock_dump(lock_beat, r)
 
@@ -1075,25 +949,18 @@ def vespa_metadata_sync_task(
             )
 
             # update Vespa. OK if doc doesn't exist. Raises exception otherwise.
-            chunks_affected = retry_index.update_single(
-                document_id,
-                tenant_id=tenant_id,
-                chunk_count=doc.chunk_count,
-                fields=fields,
-            )
+            chunks_affected = retry_index.update_single(document_id, fields)
 
             # update db last. Worst case = we crash right before this and
             # the sync might repeat again later
             mark_document_as_synced(document_id, db_session)
 
-            # this code checks for and removes a per document sync key that is
-            # used to block out the same doc from continualy resyncing
-            # a quick hack that is only needed for production issues
-            # redis_syncing_key = RedisConnectorCredentialPair.make_redis_syncing_key(
-            #     document_id
-            # )
-            # r = get_redis_client(tenant_id=tenant_id)
-            # r.delete(redis_syncing_key)
+            redis_syncing_key = RedisConnectorCredentialPair.make_redis_syncing_key(
+                document_id
+            )
+            r = get_redis_client(tenant_id=tenant_id)
+            r.delete(redis_syncing_key)
+            # r.hdel(RedisConnectorCredentialPair.SYNCING_HASH, document_id)
 
             task_logger.info(f"doc={document_id} action=sync chunks={chunks_affected}")
     except SoftTimeLimitExceeded:

backend/onyx/background/celery/versioned_apps/monitoring.py

@@ -1,15 +0,0 @@
-"""Factory stub for running celery worker / celery beat."""
-from celery import Celery
-
-from onyx.utils.variable_functionality import set_is_ee_based_on_env_variable
-
-set_is_ee_based_on_env_variable()
-
-
-def get_app() -> Celery:
-    from onyx.background.celery.apps.monitoring import celery_app
-
-    return celery_app
-
-
-app = get_app()

backend/onyx/background/indexing/job_client.py

@@ -4,10 +4,9 @@ not follow the expected behavior, etc.
 
 NOTE: cannot use Celery directly due to
 https://github.com/celery/celery/issues/7007#issuecomment-1740139367"""
-import multiprocessing as mp
 from collections.abc import Callable
 from dataclasses import dataclass
-from multiprocessing.context import SpawnProcess
+from multiprocessing import Process
 from typing import Any
 from typing import Literal
 from typing import Optional
@@ -47,9 +46,7 @@ def _initializer(
     SqlEngine.set_app_name(POSTGRES_CELERY_WORKER_INDEXING_CHILD_APP_NAME)
 
     # Initialize a new engine with desired parameters
-    SqlEngine.init_engine(
-        pool_size=4, max_overflow=12, pool_recycle=60, pool_pre_ping=True
-    )
+    SqlEngine.init_engine(pool_size=4, max_overflow=12, pool_recycle=60)
 
     # Proceed with executing the target function
     return func(*args, **kwargs)
@@ -66,7 +63,7 @@ class SimpleJob:
     """Drop in replacement for `dask.distributed.Future`"""
 
     id: int
-    process: Optional["SpawnProcess"] = None
+    process: Optional["Process"] = None
 
     def cancel(self) -> bool:
         return self.release()
@@ -134,10 +131,7 @@ class SimpleJobClient:
         job_id = self.job_id_counter
         self.job_id_counter += 1
 
-        # this approach allows us to always "spawn" a new process regardless of
-        # get_start_method's current setting
-        ctx = mp.get_context("spawn")
-        process = ctx.Process(target=_run_in_process, args=(func, args), daemon=True)
+        process = Process(target=_run_in_process, args=(func, args), daemon=True)
         job = SimpleJob(id=job_id, process=process)
         process.start()
 

backend/onyx/background/indexing/run_indexing.py

@@ -4,7 +4,6 @@ from datetime import datetime
 from datetime import timedelta
 from datetime import timezone
 
-from pydantic import BaseModel
 from sqlalchemy.orm import Session
 
 from onyx.background.indexing.checkpointing import get_time_windows_for_index_attempt
@@ -12,7 +11,6 @@ from onyx.background.indexing.tracer import OnyxTracer
 from onyx.configs.app_configs import INDEXING_SIZE_WARNING_THRESHOLD
 from onyx.configs.app_configs import INDEXING_TRACER_INTERVAL
 from onyx.configs.app_configs import POLL_CONNECTOR_OFFSET
-from onyx.configs.constants import DocumentSource
 from onyx.configs.constants import MilestoneRecordType
 from onyx.connectors.connector_runner import ConnectorRunner
 from onyx.connectors.factory import instantiate_connector
@@ -23,14 +21,12 @@ from onyx.db.connector_credential_pair import get_last_successful_attempt_time
 from onyx.db.connector_credential_pair import update_connector_credential_pair
 from onyx.db.engine import get_session_with_tenant
 from onyx.db.enums import ConnectorCredentialPairStatus
-from onyx.db.index_attempt import get_index_attempt
 from onyx.db.index_attempt import mark_attempt_canceled
 from onyx.db.index_attempt import mark_attempt_failed
 from onyx.db.index_attempt import mark_attempt_partially_succeeded
 from onyx.db.index_attempt import mark_attempt_succeeded
 from onyx.db.index_attempt import transition_attempt_to_in_progress
 from onyx.db.index_attempt import update_docs_indexed
-from onyx.db.models import ConnectorCredentialPair
 from onyx.db.models import IndexAttempt
 from onyx.db.models import IndexingStatus
 from onyx.db.models import IndexModelStatus
@@ -79,8 +75,7 @@ def _get_connector_runner(
         # it will never succeed
 
         cc_pair = get_connector_credential_pair_from_id(
-            db_session=db_session,
-            cc_pair_id=attempt.connector_credential_pair.id,
+            attempt.connector_credential_pair.id, db_session
         )
         if cc_pair and cc_pair.status == ConnectorCredentialPairStatus.ACTIVE:
             update_connector_credential_pair(
@@ -101,17 +96,10 @@ def strip_null_characters(doc_batch: list[Document]) -> list[Document]:
     for doc in doc_batch:
         cleaned_doc = doc.model_copy()
 
-        # Postgres cannot handle NUL characters in text fields
         if "\x00" in cleaned_doc.id:
             logger.warning(f"NUL characters found in document ID: {cleaned_doc.id}")
             cleaned_doc.id = cleaned_doc.id.replace("\x00", "")
 
-        if cleaned_doc.title and "\x00" in cleaned_doc.title:
-            logger.warning(
-                f"NUL characters found in document title: {cleaned_doc.title}"
-            )
-            cleaned_doc.title = cleaned_doc.title.replace("\x00", "")
-
         if "\x00" in cleaned_doc.semantic_identifier:
             logger.warning(
                 f"NUL characters found in document semantic identifier: {cleaned_doc.semantic_identifier}"
@@ -127,9 +115,6 @@ def strip_null_characters(doc_batch: list[Document]) -> list[Document]:
                 )
                 section.link = section.link.replace("\x00", "")
 
-            # since text can be longer, just replace to avoid double scan
-            section.text = section.text.replace("\x00", "")
-
         cleaned_batch.append(cleaned_doc)
 
     return cleaned_batch
@@ -139,21 +124,9 @@ class ConnectorStopSignal(Exception):
     """A custom exception used to signal a stop in processing."""
 
 
-class RunIndexingContext(BaseModel):
-    index_name: str
-    cc_pair_id: int
-    connector_id: int
-    credential_id: int
-    source: DocumentSource
-    earliest_index_time: float
-    from_beginning: bool
-    is_primary: bool
-    search_settings_status: IndexModelStatus
-
-
 def _run_indexing(
     db_session: Session,
-    index_attempt_id: int,
+    index_attempt: IndexAttempt,
     tenant_id: str | None,
     callback: IndexingHeartbeatInterface | None = None,
 ) -> None:
@@ -167,76 +140,61 @@ def _run_indexing(
     """
     start_time = time.time()
 
-    with get_session_with_tenant(tenant_id) as db_session_temp:
-        index_attempt_start = get_index_attempt(db_session_temp, index_attempt_id)
-        if not index_attempt_start:
-            raise ValueError(
-                f"Index attempt {index_attempt_id} does not exist in DB. This should not be possible."
-            )
-
-        if index_attempt_start.search_settings is None:
-            raise ValueError(
-                "Search settings must be set for indexing. This should not be possible."
-            )
-
-        # search_settings = index_attempt_start.search_settings
-        db_connector = index_attempt_start.connector_credential_pair.connector
-        db_credential = index_attempt_start.connector_credential_pair.credential
-        ctx = RunIndexingContext(
-            index_name=index_attempt_start.search_settings.index_name,
-            cc_pair_id=index_attempt_start.connector_credential_pair.id,
-            connector_id=db_connector.id,
-            credential_id=db_credential.id,
-            source=db_connector.source,
-            earliest_index_time=(
-                db_connector.indexing_start.timestamp()
-                if db_connector.indexing_start
-                else 0
-            ),
-            from_beginning=index_attempt_start.from_beginning,
-            # Only update cc-pair status for primary index jobs
-            # Secondary index syncs at the end when swapping
-            is_primary=(
-                index_attempt_start.search_settings.status == IndexModelStatus.PRESENT
-            ),
-            search_settings_status=index_attempt_start.search_settings.status,
+    if index_attempt.search_settings is None:
+        raise ValueError(
+            "Search settings must be set for indexing. This should not be possible."
         )
 
-        last_successful_index_time = (
-            ctx.earliest_index_time
-            if ctx.from_beginning
-            else get_last_successful_attempt_time(
-                connector_id=ctx.connector_id,
-                credential_id=ctx.credential_id,
-                earliest_index=ctx.earliest_index_time,
-                search_settings=index_attempt_start.search_settings,
-                db_session=db_session_temp,
-            )
-        )
+    search_settings = index_attempt.search_settings
 
-        embedding_model = DefaultIndexingEmbedder.from_db_search_settings(
-            search_settings=index_attempt_start.search_settings,
-            callback=callback,
-        )
+    index_name = search_settings.index_name
+
+    # Only update cc-pair status for primary index jobs
+    # Secondary index syncs at the end when swapping
+    is_primary = search_settings.status == IndexModelStatus.PRESENT
 
     # Indexing is only done into one index at a time
     document_index = get_default_document_index(
-        primary_index_name=ctx.index_name, secondary_index_name=None
+        primary_index_name=index_name, secondary_index_name=None
+    )
+
+    embedding_model = DefaultIndexingEmbedder.from_db_search_settings(
+        search_settings=search_settings,
+        callback=callback,
     )
 
     indexing_pipeline = build_indexing_pipeline(
-        attempt_id=index_attempt_id,
+        attempt_id=index_attempt.id,
         embedder=embedding_model,
         document_index=document_index,
         ignore_time_skip=(
-            ctx.from_beginning
-            or (ctx.search_settings_status == IndexModelStatus.FUTURE)
+            index_attempt.from_beginning
+            or (search_settings.status == IndexModelStatus.FUTURE)
         ),
         db_session=db_session,
         tenant_id=tenant_id,
         callback=callback,
     )
 
+    db_cc_pair = index_attempt.connector_credential_pair
+    db_connector = index_attempt.connector_credential_pair.connector
+    db_credential = index_attempt.connector_credential_pair.credential
+    earliest_index_time = (
+        db_connector.indexing_start.timestamp() if db_connector.indexing_start else 0
+    )
+
+    last_successful_index_time = (
+        earliest_index_time
+        if index_attempt.from_beginning
+        else get_last_successful_attempt_time(
+            connector_id=db_connector.id,
+            credential_id=db_credential.id,
+            earliest_index=earliest_index_time,
+            search_settings=index_attempt.search_settings,
+            db_session=db_session,
+        )
+    )
+
     if INDEXING_TRACER_INTERVAL > 0:
         logger.debug(f"Memory tracer starting: interval={INDEXING_TRACER_INTERVAL}")
         tracer = OnyxTracer()
@@ -244,8 +202,8 @@ def _run_indexing(
         tracer.snap()
 
     index_attempt_md = IndexAttemptMetadata(
-        connector_id=ctx.connector_id,
-        credential_id=ctx.credential_id,
+        connector_id=db_connector.id,
+        credential_id=db_credential.id,
     )
 
     batch_num = 0
@@ -261,31 +219,21 @@ def _run_indexing(
             source_type=db_connector.source,
         )
     ):
-        cc_pair_loop: ConnectorCredentialPair | None = None
-        index_attempt_loop: IndexAttempt | None = None
-
         try:
             window_start = max(
                 window_start - timedelta(minutes=POLL_CONNECTOR_OFFSET),
                 datetime(1970, 1, 1, tzinfo=timezone.utc),
             )
 
-            with get_session_with_tenant(tenant_id) as db_session_temp:
-                index_attempt_loop_start = get_index_attempt(
-                    db_session_temp, index_attempt_id
-                )
-                if not index_attempt_loop_start:
-                    raise RuntimeError(
-                        f"Index attempt {index_attempt_id} not found in DB."
-                    )
+            connector_runner = _get_connector_runner(
+                db_session=db_session,
+                attempt=index_attempt,
+                start_time=window_start,
+                end_time=window_end,
+                tenant_id=tenant_id,
+            )
 
-                connector_runner = _get_connector_runner(
-                    db_session=db_session_temp,
-                    attempt=index_attempt_loop_start,
-                    start_time=window_start,
-                    end_time=window_end,
-                    tenant_id=tenant_id,
-                )
+            all_connector_doc_ids: set[str] = set()
 
             tracer_counter = 0
             if INDEXING_TRACER_INTERVAL > 0:
@@ -300,38 +248,24 @@ def _run_indexing(
                         raise ConnectorStopSignal("Connector stop signal detected")
 
                 # TODO: should we move this into the above callback instead?
-                with get_session_with_tenant(tenant_id) as db_session_temp:
-                    cc_pair_loop = get_connector_credential_pair_from_id(
-                        db_session_temp,
-                        ctx.cc_pair_id,
+                db_session.refresh(db_cc_pair)
+                if (
+                    (
+                        db_cc_pair.status == ConnectorCredentialPairStatus.PAUSED
+                        and search_settings.status != IndexModelStatus.FUTURE
                     )
-                    if not cc_pair_loop:
-                        raise RuntimeError(f"CC pair {ctx.cc_pair_id} not found in DB.")
+                    # if it's deleting, we don't care if this is a secondary index
+                    or db_cc_pair.status == ConnectorCredentialPairStatus.DELETING
+                ):
+                    # let the `except` block handle this
+                    raise RuntimeError("Connector was disabled mid run")
 
-                    if (
-                        (
-                            cc_pair_loop.status == ConnectorCredentialPairStatus.PAUSED
-                            and ctx.search_settings_status != IndexModelStatus.FUTURE
-                        )
-                        # if it's deleting, we don't care if this is a secondary index
-                        or cc_pair_loop.status == ConnectorCredentialPairStatus.DELETING
-                    ):
-                        # let the `except` block handle this
-                        raise RuntimeError("Connector was disabled mid run")
-
-                    index_attempt_loop = get_index_attempt(
-                        db_session_temp, index_attempt_id
+                db_session.refresh(index_attempt)
+                if index_attempt.status != IndexingStatus.IN_PROGRESS:
+                    # Likely due to user manually disabling it or model swap
+                    raise RuntimeError(
+                        f"Index Attempt was canceled, status is {index_attempt.status}"
                     )
-                    if not index_attempt_loop:
-                        raise RuntimeError(
-                            f"Index attempt {index_attempt_id} not found in DB."
-                        )
-
-                    if index_attempt_loop.status != IndexingStatus.IN_PROGRESS:
-                        # Likely due to user manually disabling it or model swap
-                        raise RuntimeError(
-                            f"Index Attempt was canceled, status is {index_attempt_loop.status}"
-                        )
 
                 batch_description = []
 
@@ -355,15 +289,16 @@ def _run_indexing(
                 index_attempt_md.batch_num = batch_num + 1  # use 1-index for this
 
                 # real work happens here!
-                index_pipeline_result = indexing_pipeline(
+                new_docs, total_batch_chunks = indexing_pipeline(
                     document_batch=doc_batch_cleaned,
                     index_attempt_metadata=index_attempt_md,
                 )
 
                 batch_num += 1
-                net_doc_change += index_pipeline_result.new_docs
-                chunk_count += index_pipeline_result.total_chunks
-                document_count += index_pipeline_result.total_docs
+                net_doc_change += new_docs
+                chunk_count += total_batch_chunks
+                document_count += len(doc_batch_cleaned)
+                all_connector_doc_ids.update(doc.id for doc in doc_batch_cleaned)
 
                 # commit transaction so that the `update` below begins
                 # with a brand new transaction. Postgres uses the start
@@ -372,19 +307,18 @@ def _run_indexing(
                 # be inaccurate
                 db_session.commit()
 
-                # This new value is updated every batch, so UI can refresh per batch update
-                with get_session_with_tenant(tenant_id) as db_session_temp:
-                    update_docs_indexed(
-                        db_session=db_session_temp,
-                        index_attempt_id=index_attempt_id,
-                        total_docs_indexed=document_count,
-                        new_docs_indexed=net_doc_change,
-                        docs_removed_from_index=0,
-                    )
-
                 if callback:
                     callback.progress("_run_indexing", len(doc_batch_cleaned))
 
+                # This new value is updated every batch, so UI can refresh per batch update
+                update_docs_indexed(
+                    db_session=db_session,
+                    index_attempt=index_attempt,
+                    total_docs_indexed=document_count,
+                    new_docs_indexed=net_doc_change,
+                    docs_removed_from_index=0,
+                )
+
                 tracer_counter += 1
                 if (
                     INDEXING_TRACER_INTERVAL > 0
@@ -397,35 +331,33 @@ def _run_indexing(
                     tracer.log_previous_diff(INDEXING_TRACER_NUM_PRINT_ENTRIES)
 
             run_end_dt = window_end
-            if ctx.is_primary:
-                with get_session_with_tenant(tenant_id) as db_session_temp:
-                    update_connector_credential_pair(
-                        db_session=db_session_temp,
-                        connector_id=ctx.connector_id,
-                        credential_id=ctx.credential_id,
-                        net_docs=net_doc_change,
-                        run_dt=run_end_dt,
-                    )
+            if is_primary:
+                update_connector_credential_pair(
+                    db_session=db_session,
+                    connector_id=db_connector.id,
+                    credential_id=db_credential.id,
+                    net_docs=net_doc_change,
+                    run_dt=run_end_dt,
+                )
         except Exception as e:
             logger.exception(
                 f"Connector run exceptioned after elapsed time: {time.time() - start_time} seconds"
             )
 
             if isinstance(e, ConnectorStopSignal):
-                with get_session_with_tenant(tenant_id) as db_session_temp:
-                    mark_attempt_canceled(
-                        index_attempt_id,
-                        db_session_temp,
-                        reason=str(e),
-                    )
+                mark_attempt_canceled(
+                    index_attempt.id,
+                    db_session,
+                    reason=str(e),
+                )
 
-                    if ctx.is_primary:
-                        update_connector_credential_pair(
-                            db_session=db_session_temp,
-                            connector_id=ctx.connector_id,
-                            credential_id=ctx.credential_id,
-                            net_docs=net_doc_change,
-                        )
+                if is_primary:
+                    update_connector_credential_pair(
+                        db_session=db_session,
+                        connector_id=db_connector.id,
+                        credential_id=db_credential.id,
+                        net_docs=net_doc_change,
+                    )
 
                 if INDEXING_TRACER_INTERVAL > 0:
                     tracer.stop()
@@ -440,29 +372,23 @@ def _run_indexing(
                 # to give better clarity in the UI, as the next run will never happen.
                 if (
                     ind == 0
-                    or (
-                        cc_pair_loop is not None and not cc_pair_loop.status.is_active()
-                    )
-                    or (
-                        index_attempt_loop is not None
-                        and index_attempt_loop.status != IndexingStatus.IN_PROGRESS
-                    )
+                    or not db_cc_pair.status.is_active()
+                    or index_attempt.status != IndexingStatus.IN_PROGRESS
                 ):
-                    with get_session_with_tenant(tenant_id) as db_session_temp:
-                        mark_attempt_failed(
-                            index_attempt_id,
-                            db_session_temp,
-                            failure_reason=str(e),
-                            full_exception_trace=traceback.format_exc(),
-                        )
+                    mark_attempt_failed(
+                        index_attempt.id,
+                        db_session,
+                        failure_reason=str(e),
+                        full_exception_trace=traceback.format_exc(),
+                    )
 
-                        if ctx.is_primary:
-                            update_connector_credential_pair(
-                                db_session=db_session_temp,
-                                connector_id=ctx.connector_id,
-                                credential_id=ctx.credential_id,
-                                net_docs=net_doc_change,
-                            )
+                    if is_primary:
+                        update_connector_credential_pair(
+                            db_session=db_session,
+                            connector_id=db_connector.id,
+                            credential_id=db_credential.id,
+                            net_docs=net_doc_change,
+                        )
 
                     if INDEXING_TRACER_INTERVAL > 0:
                         tracer.stop()
@@ -485,58 +411,56 @@ def _run_indexing(
         index_attempt_md.num_exceptions > 0
         and index_attempt_md.num_exceptions >= batch_num
     ):
-        with get_session_with_tenant(tenant_id) as db_session_temp:
-            mark_attempt_failed(
-                index_attempt_id,
-                db_session_temp,
-                failure_reason="All batches exceptioned.",
-            )
-            if ctx.is_primary:
-                update_connector_credential_pair(
-                    db_session=db_session_temp,
-                    connector_id=ctx.connector_id,
-                    credential_id=ctx.credential_id,
-                )
-            raise Exception(
-                f"Connector failed - All batches exceptioned: batches={batch_num}"
+        mark_attempt_failed(
+            index_attempt.id,
+            db_session,
+            failure_reason="All batches exceptioned.",
+        )
+        if is_primary:
+            update_connector_credential_pair(
+                db_session=db_session,
+                connector_id=index_attempt.connector_credential_pair.connector.id,
+                credential_id=index_attempt.connector_credential_pair.credential.id,
             )
+        raise Exception(
+            f"Connector failed - All batches exceptioned: batches={batch_num}"
+        )
 
     elapsed_time = time.time() - start_time
 
-    with get_session_with_tenant(tenant_id) as db_session_temp:
-        if index_attempt_md.num_exceptions == 0:
-            mark_attempt_succeeded(index_attempt_id, db_session_temp)
+    if index_attempt_md.num_exceptions == 0:
+        mark_attempt_succeeded(index_attempt, db_session)
 
-            create_milestone_and_report(
-                user=None,
-                distinct_id=tenant_id or "N/A",
-                event_type=MilestoneRecordType.CONNECTOR_SUCCEEDED,
-                properties=None,
-                db_session=db_session_temp,
-            )
+        create_milestone_and_report(
+            user=None,
+            distinct_id=tenant_id or "N/A",
+            event_type=MilestoneRecordType.CONNECTOR_SUCCEEDED,
+            properties=None,
+            db_session=db_session,
+        )
 
-            logger.info(
-                f"Connector succeeded: "
-                f"docs={document_count} chunks={chunk_count} elapsed={elapsed_time:.2f}s"
-            )
-        else:
-            mark_attempt_partially_succeeded(index_attempt_id, db_session_temp)
-            logger.info(
-                f"Connector completed with some errors: "
-                f"exceptions={index_attempt_md.num_exceptions} "
-                f"batches={batch_num} "
-                f"docs={document_count} "
-                f"chunks={chunk_count} "
-                f"elapsed={elapsed_time:.2f}s"
-            )
+        logger.info(
+            f"Connector succeeded: "
+            f"docs={document_count} chunks={chunk_count} elapsed={elapsed_time:.2f}s"
+        )
+    else:
+        mark_attempt_partially_succeeded(index_attempt, db_session)
+        logger.info(
+            f"Connector completed with some errors: "
+            f"exceptions={index_attempt_md.num_exceptions} "
+            f"batches={batch_num} "
+            f"docs={document_count} "
+            f"chunks={chunk_count} "
+            f"elapsed={elapsed_time:.2f}s"
+        )
 
-        if ctx.is_primary:
-            update_connector_credential_pair(
-                db_session=db_session_temp,
-                connector_id=ctx.connector_id,
-                credential_id=ctx.credential_id,
-                run_dt=run_end_dt,
-            )
+    if is_primary:
+        update_connector_credential_pair(
+            db_session=db_session,
+            connector_id=db_connector.id,
+            credential_id=db_credential.id,
+            run_dt=run_end_dt,
+        )
 
 
 def run_indexing_entrypoint(
@@ -556,35 +480,27 @@ def run_indexing_entrypoint(
             index_attempt_id, connector_credential_pair_id
         )
         with get_session_with_tenant(tenant_id) as db_session:
-            # TODO: remove long running session entirely
             attempt = transition_attempt_to_in_progress(index_attempt_id, db_session)
 
             tenant_str = ""
             if tenant_id is not None:
                 tenant_str = f" for tenant {tenant_id}"
 
-            connector_name = attempt.connector_credential_pair.connector.name
-            connector_config = (
-                attempt.connector_credential_pair.connector.connector_specific_config
+            logger.info(
+                f"Indexing starting{tenant_str}: "
+                f"connector='{attempt.connector_credential_pair.connector.name}' "
+                f"config='{attempt.connector_credential_pair.connector.connector_specific_config}' "
+                f"credentials='{attempt.connector_credential_pair.connector_id}'"
             )
-            credential_id = attempt.connector_credential_pair.credential_id
 
-        logger.info(
-            f"Indexing starting{tenant_str}: "
-            f"connector='{connector_name}' "
-            f"config='{connector_config}' "
-            f"credentials='{credential_id}'"
-        )
+            _run_indexing(db_session, attempt, tenant_id, callback)
 
-        with get_session_with_tenant(tenant_id) as db_session:
-            _run_indexing(db_session, index_attempt_id, tenant_id, callback)
-
-        logger.info(
-            f"Indexing finished{tenant_str}: "
-            f"connector='{connector_name}' "
-            f"config='{connector_config}' "
-            f"credentials='{credential_id}'"
-        )
+            logger.info(
+                f"Indexing finished{tenant_str}: "
+                f"connector='{attempt.connector_credential_pair.connector.name}' "
+                f"config='{attempt.connector_credential_pair.connector.connector_specific_config}' "
+                f"credentials='{attempt.connector_credential_pair.connector_id}'"
+            )
     except Exception as e:
         logger.exception(
             f"Indexing job with ID '{index_attempt_id}' for tenant {tenant_id} failed due to {e}"

backend/onyx/chat/answer.py

@@ -12,10 +12,10 @@ from onyx.chat.models import AnswerStyleConfig
 from onyx.chat.models import CitationInfo
 from onyx.chat.models import OnyxAnswerPiece
 from onyx.chat.models import PromptConfig
-from onyx.chat.prompt_builder.answer_prompt_builder import AnswerPromptBuilder
-from onyx.chat.prompt_builder.answer_prompt_builder import default_build_system_message
-from onyx.chat.prompt_builder.answer_prompt_builder import default_build_user_message
-from onyx.chat.prompt_builder.answer_prompt_builder import LLMCall
+from onyx.chat.prompt_builder.build import AnswerPromptBuilder
+from onyx.chat.prompt_builder.build import default_build_system_message
+from onyx.chat.prompt_builder.build import default_build_user_message
+from onyx.chat.prompt_builder.build import LLMCall
 from onyx.chat.stream_processing.answer_response_handler import (
     CitationResponseHandler,
 )
@@ -212,6 +212,19 @@ class Answer:
             current_llm_call
         ) or ([], [])
 
+        # Quotes are no longer supported
+        # answer_handler: AnswerResponseHandler
+        # if self.answer_style_config.citation_config:
+        #     answer_handler = CitationResponseHandler(
+        #         context_docs=search_result,
+        #         doc_id_to_rank_map=map_document_id_order(search_result),
+        #     )
+        # elif self.answer_style_config.quotes_config:
+        #     answer_handler = QuotesResponseHandler(
+        #         context_docs=search_result,
+        #     )
+        # else:
+        #     raise ValueError("No answer style config provided")
         answer_handler = CitationResponseHandler(
             context_docs=final_search_results,
             final_doc_id_to_rank_map=map_document_id_order(final_search_results),
@@ -252,13 +265,11 @@ class Answer:
                 user_query=self.question,
                 prompt_config=self.prompt_config,
                 files=self.latest_query_files,
-                single_message_history=self.single_message_history,
             ),
             message_history=self.message_history,
             llm_config=self.llm.config,
-            raw_user_query=self.question,
-            raw_user_uploaded_files=self.latest_query_files or [],
             single_message_history=self.single_message_history,
+            raw_user_text=self.question,
         )
         prompt_builder.update_system_prompt(
             default_build_system_message(self.prompt_config)

backend/onyx/chat/chat_utils.py

@@ -25,7 +25,7 @@ from onyx.db.models import Persona
 from onyx.db.models import Prompt
 from onyx.db.models import Tool
 from onyx.db.models import User
-from onyx.db.prompts import get_prompts_by_ids
+from onyx.db.persona import get_prompts_by_ids
 from onyx.llm.models import PreviousMessage
 from onyx.natural_language_processing.utils import BaseTokenizer
 from onyx.server.query_and_chat.models import CreateChatMessageRequest

backend/onyx/chat/llm_response_handler.py

@@ -7,7 +7,7 @@ from langchain_core.messages import BaseMessage
 from onyx.chat.models import ResponsePart
 from onyx.chat.models import StreamStopInfo
 from onyx.chat.models import StreamStopReason
-from onyx.chat.prompt_builder.answer_prompt_builder import LLMCall
+from onyx.chat.prompt_builder.build import LLMCall
 from onyx.chat.stream_processing.answer_response_handler import AnswerResponseHandler
 from onyx.chat.tool_handling.tool_response_handler import ToolResponseHandler
 

backend/onyx/chat/models.py

@@ -8,6 +8,7 @@ from typing import TYPE_CHECKING
 from pydantic import BaseModel
 from pydantic import ConfigDict
 from pydantic import Field
+from pydantic import model_validator
 
 from onyx.configs.constants import DocumentSource
 from onyx.configs.constants import MessageType
@@ -260,8 +261,13 @@ class CitationConfig(BaseModel):
     all_docs_useful: bool = False
 
 
+class QuotesConfig(BaseModel):
+    pass
+
+
 class AnswerStyleConfig(BaseModel):
-    citation_config: CitationConfig
+    citation_config: CitationConfig | None = None
+    quotes_config: QuotesConfig | None = None
     document_pruning_config: DocumentPruningConfig = Field(
         default_factory=DocumentPruningConfig
     )
@@ -270,6 +276,20 @@ class AnswerStyleConfig(BaseModel):
     # right now, only used by the simple chat API
     structured_response_format: dict | None = None
 
+    @model_validator(mode="after")
+    def check_quotes_and_citation(self) -> "AnswerStyleConfig":
+        if self.citation_config is None and self.quotes_config is None:
+            raise ValueError(
+                "One of `citation_config` or `quotes_config` must be provided"
+            )
+
+        if self.citation_config is not None and self.quotes_config is not None:
+            raise ValueError(
+                "Only one of `citation_config` or `quotes_config` must be provided"
+            )
+
+        return self
+
 
 class PromptConfig(BaseModel):
     """Final representation of the Prompt configuration passed

backend/onyx/chat/process_message.py

@@ -302,11 +302,6 @@ def stream_chat_message_objects(
     enforce_chat_session_id_for_search_docs: bool = True,
     bypass_acl: bool = False,
     include_contexts: bool = False,
-    # a string which represents the history of a conversation. Used in cases like
-    # Slack threads where the conversation cannot be represented by a chain of User/Assistant
-    # messages.
-    # NOTE: is not stored in the database at all.
-    single_message_history: str | None = None,
 ) -> ChatPacketStream:
     """Streams in order:
     1. [conditional] Retrieved documents if a search needs to be run
@@ -712,7 +707,6 @@ def stream_chat_message_objects(
             ],
             tools=tools,
             force_use_tool=_get_force_search_settings(new_msg_req, tools),
-            single_message_history=single_message_history,
         )
 
         reference_db_search_docs = None

backend/onyx/chat/prompt_builder/build.py

@@ -17,7 +17,6 @@ from onyx.llm.utils import check_message_tokens
 from onyx.llm.utils import message_to_prompt_and_imgs
 from onyx.natural_language_processing.utils import get_tokenizer
 from onyx.prompts.chat_prompts import CHAT_USER_CONTEXT_FREE_PROMPT
-from onyx.prompts.direct_qa_prompts import HISTORY_BLOCK
 from onyx.prompts.prompt_utils import add_date_time_to_prompt
 from onyx.prompts.prompt_utils import drop_messages_history_overflow
 from onyx.tools.force import ForceUseTool
@@ -43,22 +42,11 @@ def default_build_system_message(
 
 
 def default_build_user_message(
-    user_query: str,
-    prompt_config: PromptConfig,
-    files: list[InMemoryChatFile] = [],
-    single_message_history: str | None = None,
+    user_query: str, prompt_config: PromptConfig, files: list[InMemoryChatFile] = []
 ) -> HumanMessage:
-    history_block = (
-        HISTORY_BLOCK.format(history_str=single_message_history)
-        if single_message_history
-        else ""
-    )
-
     user_prompt = (
         CHAT_USER_CONTEXT_FREE_PROMPT.format(
-            history_block=history_block,
-            task_prompt=prompt_config.task_prompt,
-            user_query=user_query,
+            task_prompt=prompt_config.task_prompt, user_query=user_query
         )
         if prompt_config.task_prompt
         else user_query
@@ -76,8 +64,7 @@ class AnswerPromptBuilder:
         user_message: HumanMessage,
         message_history: list[PreviousMessage],
         llm_config: LLMConfig,
-        raw_user_query: str,
-        raw_user_uploaded_files: list[InMemoryChatFile],
+        raw_user_text: str,
         single_message_history: str | None = None,
     ) -> None:
         self.max_tokens = compute_max_llm_input_tokens(llm_config)
@@ -96,6 +83,10 @@ class AnswerPromptBuilder:
             self.history_token_cnts,
         ) = translate_history_to_basemessages(message_history)
 
+        # for cases where like the QA flow where we want to condense the chat history
+        # into a single message rather than a sequence of User / Assistant messages
+        self.single_message_history = single_message_history
+
         self.system_message_and_token_cnt: tuple[SystemMessage, int] | None = None
         self.user_message_and_token_cnt = (
             user_message,
@@ -104,10 +95,7 @@ class AnswerPromptBuilder:
 
         self.new_messages_and_token_cnts: list[tuple[BaseMessage, int]] = []
 
-        # used for building a new prompt after a tool-call
-        self.raw_user_query = raw_user_query
-        self.raw_user_uploaded_files = raw_user_uploaded_files
-        self.single_message_history = single_message_history
+        self.raw_user_message = raw_user_text
 
     def update_system_prompt(self, system_message: SystemMessage | None) -> None:
         if not system_message:

backend/onyx/chat/prompt_builder/citations_prompt.py

@@ -1,13 +1,12 @@
 from langchain.schema.messages import HumanMessage
 from langchain.schema.messages import SystemMessage
-from sqlalchemy.orm import Session
 
 from onyx.chat.models import LlmDoc
 from onyx.chat.models import PromptConfig
 from onyx.configs.model_configs import GEN_AI_SINGLE_USER_MESSAGE_EXPECTED_MAX_TOKENS
 from onyx.context.search.models import InferenceChunk
 from onyx.db.models import Persona
-from onyx.db.prompts import get_default_prompt
+from onyx.db.persona import get_default_prompt__read_only
 from onyx.db.search_settings import get_multilingual_expansion
 from onyx.llm.factory import get_llms_for_persona
 from onyx.llm.factory import get_main_llm_from_tuple
@@ -98,12 +97,11 @@ def compute_max_document_tokens(
 
 
 def compute_max_document_tokens_for_persona(
-    db_session: Session,
     persona: Persona,
     actual_user_input: str | None = None,
     max_llm_token_override: int | None = None,
 ) -> int:
-    prompt = persona.prompts[0] if persona.prompts else get_default_prompt(db_session)
+    prompt = persona.prompts[0] if persona.prompts else get_default_prompt__read_only()
     return compute_max_document_tokens(
         prompt_config=PromptConfig.from_model(prompt),
         llm_config=get_main_llm_from_tuple(get_llms_for_persona(persona)).config,
@@ -146,7 +144,9 @@ def build_citations_user_message(
     )
 
     history_block = (
-        HISTORY_BLOCK.format(history_str=history_message) if history_message else ""
+        HISTORY_BLOCK.format(history_str=history_message) + "\n"
+        if history_message
+        else ""
     )
     query, img_urls = message_to_prompt_and_imgs(message)
 

backend/onyx/chat/prompt_builder/utils.py

@@ -7,6 +7,26 @@ from onyx.db.models import ChatMessage
 from onyx.file_store.models import InMemoryChatFile
 from onyx.llm.models import PreviousMessage
 from onyx.llm.utils import build_content_with_imgs
+from onyx.prompts.direct_qa_prompts import PARAMATERIZED_PROMPT
+from onyx.prompts.direct_qa_prompts import PARAMATERIZED_PROMPT_WITHOUT_CONTEXT
+
+
+def build_dummy_prompt(
+    system_prompt: str, task_prompt: str, retrieval_disabled: bool
+) -> str:
+    if retrieval_disabled:
+        return PARAMATERIZED_PROMPT_WITHOUT_CONTEXT.format(
+            user_query="<USER_QUERY>",
+            system_prompt=system_prompt,
+            task_prompt=task_prompt,
+        ).strip()
+
+    return PARAMATERIZED_PROMPT.format(
+        context_docs_str="<CONTEXT_DOCS>",
+        user_query="<USER_QUERY>",
+        system_prompt=system_prompt,
+        task_prompt=task_prompt,
+    ).strip()
 
 
 def translate_onyx_msg_to_langchain(

backend/onyx/chat/tool_handling/tool_response_handler.py

@@ -5,7 +5,7 @@ from langchain_core.messages import BaseMessage
 from langchain_core.messages import ToolCall
 
 from onyx.chat.models import ResponsePart
-from onyx.chat.prompt_builder.answer_prompt_builder import LLMCall
+from onyx.chat.prompt_builder.build import LLMCall
 from onyx.llm.interfaces import LLM
 from onyx.tools.force import ForceUseTool
 from onyx.tools.message import build_tool_message
@@ -62,7 +62,7 @@ class ToolResponseHandler:
                 llm_call.force_use_tool.args
                 if llm_call.force_use_tool.args is not None
                 else tool.get_args_for_non_tool_calling_llm(
-                    query=llm_call.prompt_builder.raw_user_query,
+                    query=llm_call.prompt_builder.raw_user_message,
                     history=llm_call.prompt_builder.raw_message_history,
                     llm=llm,
                     force_run=True,
@@ -76,7 +76,7 @@ class ToolResponseHandler:
         else:
             tool_options = check_which_tools_should_run_for_non_tool_calling_llm(
                 tools=llm_call.tools,
-                query=llm_call.prompt_builder.raw_user_query,
+                query=llm_call.prompt_builder.raw_user_message,
                 history=llm_call.prompt_builder.raw_message_history,
                 llm=llm,
             )
@@ -95,7 +95,7 @@ class ToolResponseHandler:
                 select_single_tool_for_non_tool_calling_llm(
                     tools_and_args=available_tools_and_args,
                     history=llm_call.prompt_builder.raw_message_history,
-                    query=llm_call.prompt_builder.raw_user_query,
+                    query=llm_call.prompt_builder.raw_user_message,
                     llm=llm,
                 )
                 if available_tools_and_args

backend/onyx/configs/app_configs.py

@@ -17,7 +17,6 @@ APP_PORT = 8080
 # prefix from requests directed towards the API server. In these cases, set this to `/api`
 APP_API_PREFIX = os.environ.get("API_PREFIX", "")
 
-SKIP_WARM_UP = os.environ.get("SKIP_WARM_UP", "").lower() == "true"
 
 #####
 # User Facing Features Configs
@@ -196,6 +195,7 @@ REDIS_PASSWORD = os.environ.get("REDIS_PASSWORD") or ""
 
 REDIS_AUTH_KEY_PREFIX = "fastapi_users_token:"
 
+
 # Rate limiting for auth endpoints
 RATE_LIMIT_WINDOW_SECONDS: int | None = None
 _rate_limit_window_seconds_str = os.environ.get("RATE_LIMIT_WINDOW_SECONDS")
@@ -213,7 +213,6 @@ if _rate_limit_max_requests_str is not None:
     except ValueError:
         pass
 
-AUTH_RATE_LIMITING_ENABLED = RATE_LIMIT_MAX_REQUESTS and RATE_LIMIT_WINDOW_SECONDS
 # Used for general redis things
 REDIS_DB_NUMBER = int(os.environ.get("REDIS_DB_NUMBER", 0))
 
@@ -281,11 +280,6 @@ try:
 except ValueError:
     CELERY_WORKER_INDEXING_CONCURRENCY = CELERY_WORKER_INDEXING_CONCURRENCY_DEFAULT
 
-# The maximum number of tasks that can be queued up to sync to Vespa in a single pass
-VESPA_SYNC_MAX_TASKS = 1024
-
-DB_YIELD_PER_DEFAULT = 64
-
 #####
 # Connector Configs
 #####

backend/onyx/configs/chat_configs.py

@@ -1,6 +1,6 @@
 import os
 
-INPUT_PROMPT_YAML = "./onyx/seeding/input_prompts.yaml"
+
 PROMPTS_YAML = "./onyx/seeding/prompts.yaml"
 PERSONAS_YAML = "./onyx/seeding/personas.yaml"
 

backend/onyx/configs/constants.py

@@ -47,7 +47,6 @@ POSTGRES_CELERY_WORKER_PRIMARY_APP_NAME = "celery_worker_primary"
 POSTGRES_CELERY_WORKER_LIGHT_APP_NAME = "celery_worker_light"
 POSTGRES_CELERY_WORKER_HEAVY_APP_NAME = "celery_worker_heavy"
 POSTGRES_CELERY_WORKER_INDEXING_APP_NAME = "celery_worker_indexing"
-POSTGRES_CELERY_WORKER_MONITORING_APP_NAME = "celery_worker_monitoring"
 POSTGRES_CELERY_WORKER_INDEXING_CHILD_APP_NAME = "celery_worker_indexing_child"
 POSTGRES_PERMISSIONS_APP_NAME = "permissions"
 POSTGRES_UNKNOWN_APP_NAME = "unknown"
@@ -79,8 +78,6 @@ KV_DOCUMENTS_SEEDED_KEY = "documents_seeded"
 
 # NOTE: we use this timeout / 4 in various places to refresh a lock
 # might be worth separating this timeout into separate timeouts for each situation
-CELERY_GENERIC_BEAT_LOCK_TIMEOUT = 120
-
 CELERY_VESPA_SYNC_BEAT_LOCK_TIMEOUT = 120
 
 CELERY_PRIMARY_WORKER_LOCK_TIMEOUT = 120
@@ -200,7 +197,6 @@ class SessionType(str, Enum):
 class QAFeedbackType(str, Enum):
     LIKE = "like"  # User likes the answer, used for metrics
     DISLIKE = "dislike"  # User dislikes the answer, used for metrics
-    MIXED = "mixed"  # User likes some answers and dislikes other, used for chat session metrics
 
 
 class SearchFeedbackType(str, Enum):
@@ -264,9 +260,6 @@ class OnyxCeleryQueues:
     # Indexing queue
     CONNECTOR_INDEXING = "connector_indexing"
 
-    # Monitoring queue
-    MONITORING = "monitoring"
-
 
 class OnyxRedisLocks:
     PRIMARY_WORKER = "da_lock:primary_worker"
@@ -281,7 +274,6 @@ class OnyxRedisLocks:
         "da_lock:check_connector_external_group_sync_beat"
     )
     MONITOR_VESPA_SYNC_BEAT_LOCK = "da_lock:monitor_vespa_sync_beat"
-    MONITOR_BACKGROUND_PROCESSES_LOCK = "da_lock:monitor_background_processes"
 
     CONNECTOR_DOC_PERMISSIONS_SYNC_LOCK_PREFIX = (
         "da_lock:connector_doc_permissions_sync"
@@ -294,8 +286,6 @@ class OnyxRedisLocks:
     SLACK_BOT_HEARTBEAT_PREFIX = "da_heartbeat:slack_bot"
     ANONYMOUS_USER_ENABLED = "anonymous_user_enabled"
 
-    CLOUD_CHECK_INDEXING_BEAT_LOCK = "da_lock:cloud_check_indexing_beat"
-
 
 class OnyxRedisSignals:
     VALIDATE_INDEXING_FENCES = "signal:validate_indexing_fences"
@@ -309,13 +299,6 @@ class OnyxCeleryPriority(int, Enum):
     LOWEST = auto()
 
 
-# a prefix used to distinguish system wide tasks in the cloud
-ONYX_CLOUD_CELERY_TASK_PREFIX = "cloud"
-
-# the tenant id we use for system level redis operations
-ONYX_CLOUD_TENANT_ID = "cloud"
-
-
 class OnyxCeleryTask:
     CHECK_FOR_CONNECTOR_DELETION = "check_for_connector_deletion_task"
     CHECK_FOR_VESPA_SYNC_TASK = "check_for_vespa_sync_task"
@@ -325,7 +308,6 @@ class OnyxCeleryTask:
     CHECK_FOR_EXTERNAL_GROUP_SYNC = "check_for_external_group_sync"
     CHECK_FOR_LLM_MODEL_UPDATE = "check_for_llm_model_update"
     MONITOR_VESPA_SYNC = "monitor_vespa_sync"
-    MONITOR_BACKGROUND_PROCESSES = "monitor_background_processes"
     KOMBU_MESSAGE_CLEANUP_TASK = "kombu_message_cleanup_task"
     CONNECTOR_PERMISSION_SYNC_GENERATOR_TASK = (
         "connector_permission_sync_generator_task"
@@ -343,8 +325,6 @@ class OnyxCeleryTask:
     CHECK_TTL_MANAGEMENT_TASK = "check_ttl_management_task"
     AUTOGENERATE_USAGE_REPORT_TASK = "autogenerate_usage_report_task"
 
-    CLOUD_CHECK_FOR_INDEXING = f"{ONYX_CLOUD_CELERY_TASK_PREFIX}_check_for_indexing"
-
 
 REDIS_SOCKET_KEEPALIVE_OPTIONS = {}
 REDIS_SOCKET_KEEPALIVE_OPTIONS[socket.TCP_KEEPINTVL] = 15

backend/onyx/connectors/confluence/onyx_confluence.py

@@ -121,7 +121,6 @@ def handle_confluence_rate_limit(confluence_call: F) -> F:
 
 
 _DEFAULT_PAGINATION_LIMIT = 1000
-_MINIMUM_PAGINATION_LIMIT = 50
 
 
 class OnyxConfluence(Confluence):
@@ -135,6 +134,32 @@ class OnyxConfluence(Confluence):
         super(OnyxConfluence, self).__init__(url, *args, **kwargs)
         self._wrap_methods()
 
+    def get_current_user(self, expand: str | None = None) -> Any:
+        """
+        Implements a method that isn't in the third party client.
+
+        Get information about the current user
+        :param expand: OPTIONAL expand for get status of user.
+                Possible param is "status". Results are "Active, Deactivated"
+        :return: Returns the user details
+        """
+
+        from atlassian.errors import ApiPermissionError  # type:ignore
+
+        url = "rest/api/user/current"
+        params = {}
+        if expand:
+            params["expand"] = expand
+        try:
+            response = self.get(url, params=params)
+        except HTTPError as e:
+            if e.response.status_code == 403:
+                raise ApiPermissionError(
+                    "The calling user does not have permission", reason=e
+                )
+            raise
+        return response
+
     def _wrap_methods(self) -> None:
         """
         For each attribute that is callable (i.e., a method) and doesn't start with an underscore,
@@ -161,67 +186,27 @@ class OnyxConfluence(Confluence):
         url_suffix += f"{connection_char}limit={limit}"
 
         while url_suffix:
-            logger.debug(f"Making confluence call to {url_suffix}")
             try:
-                raw_response = self.get(
-                    path=url_suffix,
-                    advanced_mode=True,
-                )
-            except Exception as e:
-                logger.exception(f"Error in confluence call to {url_suffix}")
-                raise e
-
-            try:
-                raw_response.raise_for_status()
+                logger.debug(f"Making confluence call to {url_suffix}")
+                next_response = self.get(url_suffix)
             except Exception as e:
                 logger.warning(f"Error in confluence call to {url_suffix}")
 
                 # If the problematic expansion is in the url, replace it
                 # with the replacement expansion and try again
                 # If that fails, raise the error
-                if _PROBLEMATIC_EXPANSIONS in url_suffix:
-                    logger.warning(
-                        f"Replacing {_PROBLEMATIC_EXPANSIONS} with {_REPLACEMENT_EXPANSIONS}"
-                        " and trying again."
-                    )
-                    url_suffix = url_suffix.replace(
-                        _PROBLEMATIC_EXPANSIONS,
-                        _REPLACEMENT_EXPANSIONS,
-                    )
-                    continue
-                if (
-                    raw_response.status_code == 500
-                    and limit > _MINIMUM_PAGINATION_LIMIT
-                ):
-                    new_limit = limit // 2
-                    logger.warning(
-                        f"Error in confluence call to {url_suffix} \n"
-                        f"Raw Response Text: {raw_response.text} \n"
-                        f"Full Response: {raw_response.__dict__} \n"
-                        f"Error: {e} \n"
-                        f"Reducing limit from {limit} to {new_limit} and trying again."
-                    )
-                    url_suffix = url_suffix.replace(
-                        f"limit={limit}", f"limit={new_limit}"
-                    )
-                    limit = new_limit
-                    continue
-
-                logger.exception(
-                    f"Error in confluence call to {url_suffix} \n"
-                    f"Raw Response Text: {raw_response.text} \n"
-                    f"Full Response: {raw_response.__dict__} \n"
-                    f"Error: {e} \n"
+                if _PROBLEMATIC_EXPANSIONS not in url_suffix:
+                    logger.exception(f"Error in confluence call to {url_suffix}")
+                    raise e
+                logger.warning(
+                    f"Replacing {_PROBLEMATIC_EXPANSIONS} with {_REPLACEMENT_EXPANSIONS}"
+                    " and trying again."
                 )
-                raise e
-
-            try:
-                next_response = raw_response.json()
-            except Exception as e:
-                logger.exception(
-                    f"Failed to parse response as JSON. Response: {raw_response.__dict__}"
+                url_suffix = url_suffix.replace(
+                    _PROBLEMATIC_EXPANSIONS,
+                    _REPLACEMENT_EXPANSIONS,
                 )
-                raise e
+                continue
 
             # yield the results individually
             yield from next_response.get("results", [])
@@ -328,62 +313,6 @@ class OnyxConfluence(Confluence):
         group_name = quote(group_name)
         yield from self._paginate_url(f"rest/api/group/{group_name}/member", limit)
 
-    def get_all_space_permissions_server(
-        self,
-        space_key: str,
-    ) -> list[dict[str, Any]]:
-        """
-        This is a confluence server specific method that can be used to
-        fetch the permissions of a space.
-        This is better logging than calling the get_space_permissions method
-        because it returns a jsonrpc response.
-        TODO: Make this call these endpoints for newer confluence versions:
-        - /rest/api/space/{spaceKey}/permissions
-        - /rest/api/space/{spaceKey}/permissions/anonymous
-        """
-        url = "rpc/json-rpc/confluenceservice-v2"
-        data = {
-            "jsonrpc": "2.0",
-            "method": "getSpacePermissionSets",
-            "id": 7,
-            "params": [space_key],
-        }
-        response = self.post(url, data=data)
-        logger.debug(f"jsonrpc response: {response}")
-        if not response.get("result"):
-            logger.warning(
-                f"No jsonrpc response for space permissions for space {space_key}"
-                f"\nResponse: {response}"
-            )
-
-        return response.get("result", [])
-
-    def get_current_user(self, expand: str | None = None) -> Any:
-        """
-        Implements a method that isn't in the third party client.
-
-        Get information about the current user
-        :param expand: OPTIONAL expand for get status of user.
-                Possible param is "status". Results are "Active, Deactivated"
-        :return: Returns the user details
-        """
-
-        from atlassian.errors import ApiPermissionError  # type:ignore
-
-        url = "rest/api/user/current"
-        params = {}
-        if expand:
-            params["expand"] = expand
-        try:
-            response = self.get(url, params=params)
-        except HTTPError as e:
-            if e.response.status_code == 403:
-                raise ApiPermissionError(
-                    "The calling user does not have permission", reason=e
-                )
-            raise
-        return response
-
 
 def _validate_connector_configuration(
     credentials: dict[str, Any],

backend/onyx/connectors/confluence/utils.py

@@ -32,13 +32,11 @@ def get_user_email_from_username__server(
             response = confluence_client.get_mobile_parameters(user_name)
             email = response.get("email")
         except Exception:
+            # For now, we'll just return a string that indicates failure
+            # We may want to revert to returning None in the future
+            # email = None
+            email = f"FAILED TO GET CONFLUENCE EMAIL FOR {user_name}"
             logger.warning(f"failed to get confluence email for {user_name}")
-            # For now, we'll just return None and log a warning. This means
-            # we will keep retrying to get the email every group sync.
-            email = None
-            # We may want to just return a string that indicates failure so we dont
-            # keep retrying
-            # email = f"FAILED TO GET CONFLUENCE EMAIL FOR {user_name}"
         _USER_EMAIL_CACHE[user_name] = email
     return _USER_EMAIL_CACHE[user_name]
 

backend/onyx/connectors/egnyte/connector.py

@@ -224,7 +224,7 @@ class EgnyteConnector(LoadConnector, PollConnector, OAuthConnector):
     def _get_files_list(
         self,
         path: str,
-    ) -> Generator[dict[str, Any], None, None]:
+    ) -> list[dict[str, Any]]:
         if not self.access_token or not self.domain:
             raise ConnectorMissingCredentialError("Egnyte")
 
@@ -245,46 +245,48 @@ class EgnyteConnector(LoadConnector, PollConnector, OAuthConnector):
             raise RuntimeError(f"Failed to fetch files from Egnyte: {response.text}")
 
         data = response.json()
+        all_files: list[dict[str, Any]] = []
 
-        # Yield files from current directory
-        for file in data.get("files", []):
-            yield file
+        # Add files from current directory
+        all_files.extend(data.get("files", []))
 
         # Recursively traverse folders
-        for folder in data.get("folders", []):
-            yield from self._get_files_list(folder["path"])
+        for item in data.get("folders", []):
+            all_files.extend(self._get_files_list(item["path"]))
 
-    def _should_index_file(
+        return all_files
+
+    def _filter_files(
         self,
-        file: dict[str, Any],
+        files: list[dict[str, Any]],
         start_time: datetime | None = None,
         end_time: datetime | None = None,
-    ) -> bool:
-        """Return True if file should be included based on filters."""
-        if file["is_folder"]:
-            return False
+    ) -> list[dict[str, Any]]:
+        filtered_files = []
+        for file in files:
+            if file["is_folder"]:
+                continue
 
-        file_modified = _parse_last_modified(file["last_modified"])
-        if start_time and file_modified < start_time:
-            return False
-        if end_time and file_modified > end_time:
-            return False
+            file_modified = _parse_last_modified(file["last_modified"])
+            if start_time and file_modified < start_time:
+                continue
+            if end_time and file_modified > end_time:
+                continue
 
-        return True
+            filtered_files.append(file)
+
+        return filtered_files
 
     def _process_files(
         self,
         start_time: datetime | None = None,
         end_time: datetime | None = None,
     ) -> Generator[list[Document], None, None]:
+        files = self._get_files_list(self.folder_path)
+        files = self._filter_files(files, start_time, end_time)
+
         current_batch: list[Document] = []
-
-        # Iterate through yielded files and filter them
-        for file in self._get_files_list(self.folder_path):
-            if not self._should_index_file(file, start_time, end_time):
-                logger.debug(f"Skipping file '{file['path']}'.")
-                continue
-
+        for file in files:
             try:
                 # Set up request with streaming enabled
                 headers = {

backend/onyx/connectors/fireflies/connector.py

@@ -30,14 +30,13 @@ _FIREFLIES_API_QUERY = """
         transcripts(fromDate: $fromDate, toDate: $toDate, limit: $limit, skip: $skip) {
             id
             title
-            organizer_email
+            host_email
             participants
             date
             transcript_url
             sentences {
                 text
                 speaker_name
-                start_time
             }
         }
     }
@@ -45,34 +44,16 @@ _FIREFLIES_API_QUERY = """
 
 
 def _create_doc_from_transcript(transcript: dict) -> Document | None:
-    sections: List[Section] = []
-    current_speaker_name = None
-    current_link = ""
-    current_text = ""
+    meeting_text = ""
+    sentences = transcript.get("sentences", [])
+    if sentences:
+        for sentence in sentences:
+            meeting_text += sentence.get("speaker_name") or "Unknown Speaker"
+            meeting_text += ": " + sentence.get("text", "") + "\n\n"
+    else:
+        return None
 
-    for sentence in transcript["sentences"]:
-        if sentence["speaker_name"] != current_speaker_name:
-            if current_speaker_name is not None:
-                sections.append(
-                    Section(
-                        link=current_link,
-                        text=current_text.strip(),
-                    )
-                )
-            current_speaker_name = sentence.get("speaker_name") or "Unknown Speaker"
-            current_link = f"{transcript['transcript_url']}?t={sentence['start_time']}"
-            current_text = f"{current_speaker_name}: "
-
-        cleaned_text = sentence["text"].replace("\xa0", " ")
-        current_text += f"{cleaned_text} "
-
-    # Sometimes these links (links with a timestamp) do not work, it is a bug with Fireflies.
-    sections.append(
-        Section(
-            link=current_link,
-            text=current_text.strip(),
-        )
-    )
+    meeting_link = transcript["transcript_url"]
 
     fireflies_id = _FIREFLIES_ID_PREFIX + transcript["id"]
 
@@ -81,22 +62,27 @@ def _create_doc_from_transcript(transcript: dict) -> Document | None:
     meeting_date_unix = transcript["date"]
     meeting_date = datetime.fromtimestamp(meeting_date_unix / 1000, tz=timezone.utc)
 
-    meeting_organizer_email = transcript["organizer_email"]
-    organizer_email_user_info = [BasicExpertInfo(email=meeting_organizer_email)]
+    meeting_host_email = transcript["host_email"]
+    host_email_user_info = [BasicExpertInfo(email=meeting_host_email)]
 
     meeting_participants_email_list = []
     for participant in transcript.get("participants", []):
-        if participant != meeting_organizer_email and participant:
+        if participant != meeting_host_email and participant:
             meeting_participants_email_list.append(BasicExpertInfo(email=participant))
 
     return Document(
         id=fireflies_id,
-        sections=sections,
+        sections=[
+            Section(
+                link=meeting_link,
+                text=meeting_text,
+            )
+        ],
         source=DocumentSource.FIREFLIES,
         semantic_identifier=meeting_title,
         metadata={},
         doc_updated_at=meeting_date,
-        primary_owners=organizer_email_user_info,
+        primary_owners=host_email_user_info,
         secondary_owners=meeting_participants_email_list,
     )
 

backend/onyx/connectors/google_drive/connector.py

@@ -258,7 +258,7 @@ class GoogleDriveConnector(LoadConnector, PollConnector, SlimConnector):
                         user_emails.append(email)
         return user_emails
 
-    def get_all_drive_ids(self) -> set[str]:
+    def _get_all_drive_ids(self) -> set[str]:
         primary_drive_service = get_drive_service(
             creds=self.creds,
             user_email=self.primary_admin_email,
@@ -353,7 +353,7 @@ class GoogleDriveConnector(LoadConnector, PollConnector, SlimConnector):
     ) -> Iterator[GoogleDriveFileType]:
         all_org_emails: list[str] = self._get_all_user_emails()
 
-        all_drive_ids: set[str] = self.get_all_drive_ids()
+        all_drive_ids: set[str] = self._get_all_drive_ids()
 
         drive_ids_to_retrieve: set[str] = set()
         folder_ids_to_retrieve: set[str] = set()
@@ -437,7 +437,7 @@ class GoogleDriveConnector(LoadConnector, PollConnector, SlimConnector):
             # If all 3 are true, we already yielded from get_all_files_for_oauth
             return
 
-        all_drive_ids = self.get_all_drive_ids()
+        all_drive_ids = self._get_all_drive_ids()
         drive_ids_to_retrieve: set[str] = set()
         folder_ids_to_retrieve: set[str] = set()
         if self._requested_shared_drive_ids or self._requested_folder_ids:

backend/onyx/connectors/google_drive/doc_conversion.py

@@ -252,7 +252,6 @@ def build_slim_document(file: GoogleDriveFileType) -> SlimDocument | None:
         id=file["webViewLink"],
         perm_sync_data={
             "doc_id": file.get("id"),
-            "drive_id": file.get("driveId"),
             "permissions": file.get("permissions", []),
             "permission_ids": file.get("permissionIds", []),
             "name": file.get("name"),

backend/onyx/connectors/google_drive/file_retrieval.py

@@ -19,7 +19,7 @@ FILE_FIELDS = (
     "shortcutDetails, owners(emailAddress), size)"
 )
 SLIM_FILE_FIELDS = (
-    "nextPageToken, files(mimeType, driveId, id, name, permissions(emailAddress, type), "
+    "nextPageToken, files(mimeType, id, name, permissions(emailAddress, type), "
     "permissionIds, webViewLink, owners(emailAddress))"
 )
 FOLDER_FIELDS = "nextPageToken, files(id, name, permissions, modifiedTime, webViewLink, shortcutDetails)"

backend/onyx/connectors/google_utils/google_kv.py

@@ -17,9 +17,6 @@ from onyx.configs.constants import KV_GOOGLE_DRIVE_CRED_KEY
 from onyx.configs.constants import KV_GOOGLE_DRIVE_SERVICE_ACCOUNT_KEY
 from onyx.connectors.google_utils.resources import get_drive_service
 from onyx.connectors.google_utils.resources import get_gmail_service
-from onyx.connectors.google_utils.shared_constants import (
-    DB_CREDENTIALS_AUTHENTICATION_METHOD,
-)
 from onyx.connectors.google_utils.shared_constants import (
     DB_CREDENTIALS_DICT_SERVICE_ACCOUNT_KEY,
 )
@@ -32,9 +29,6 @@ from onyx.connectors.google_utils.shared_constants import (
 from onyx.connectors.google_utils.shared_constants import (
     GOOGLE_SCOPES,
 )
-from onyx.connectors.google_utils.shared_constants import (
-    GoogleOAuthAuthenticationMethod,
-)
 from onyx.connectors.google_utils.shared_constants import (
     MISSING_SCOPES_ERROR_STR,
 )
@@ -102,7 +96,6 @@ def update_credential_access_tokens(
     user: User,
     db_session: Session,
     source: DocumentSource,
-    auth_method: GoogleOAuthAuthenticationMethod,
 ) -> OAuthCredentials | None:
     app_credentials = get_google_app_cred(source)
     flow = InstalledAppFlow.from_client_config(
@@ -126,7 +119,6 @@ def update_credential_access_tokens(
     new_creds_dict = {
         DB_CREDENTIALS_DICT_TOKEN_KEY: token_json_str,
         DB_CREDENTIALS_PRIMARY_ADMIN_KEY: email,
-        DB_CREDENTIALS_AUTHENTICATION_METHOD: auth_method.value,
     }
 
     if not update_credential_json(credential_id, new_creds_dict, user, db_session):
@@ -137,7 +129,6 @@ def update_credential_access_tokens(
 def build_service_account_creds(
     source: DocumentSource,
     primary_admin_email: str | None = None,
-    name: str | None = None,
 ) -> CredentialBase:
     service_account_key = get_service_account_key(source=source)
 
@@ -147,15 +138,10 @@ def build_service_account_creds(
     if primary_admin_email:
         credential_dict[DB_CREDENTIALS_PRIMARY_ADMIN_KEY] = primary_admin_email
 
-    credential_dict[
-        DB_CREDENTIALS_AUTHENTICATION_METHOD
-    ] = GoogleOAuthAuthenticationMethod.UPLOADED.value
-
     return CredentialBase(
         credential_json=credential_dict,
         admin_public=True,
         source=source,
-        name=name,
     )
 
 

backend/onyx/connectors/salesforce/connector.py

@@ -4,29 +4,34 @@ from typing import Any
 from simple_salesforce import Salesforce
 
 from onyx.configs.app_configs import INDEX_BATCH_SIZE
+from onyx.configs.constants import DocumentSource
+from onyx.connectors.cross_connector_utils.miscellaneous_utils import time_str_to_utc
 from onyx.connectors.interfaces import GenerateDocumentsOutput
 from onyx.connectors.interfaces import GenerateSlimDocumentOutput
 from onyx.connectors.interfaces import LoadConnector
 from onyx.connectors.interfaces import PollConnector
 from onyx.connectors.interfaces import SecondsSinceUnixEpoch
 from onyx.connectors.interfaces import SlimConnector
+from onyx.connectors.models import BasicExpertInfo
 from onyx.connectors.models import ConnectorMissingCredentialError
 from onyx.connectors.models import Document
 from onyx.connectors.models import SlimDocument
-from onyx.connectors.salesforce.doc_conversion import convert_sf_object_to_doc
-from onyx.connectors.salesforce.doc_conversion import ID_PREFIX
+from onyx.connectors.salesforce.doc_conversion import extract_section
 from onyx.connectors.salesforce.salesforce_calls import fetch_all_csvs_in_parallel
 from onyx.connectors.salesforce.salesforce_calls import get_all_children_of_sf_type
 from onyx.connectors.salesforce.sqlite_functions import get_affected_parent_ids_by_type
+from onyx.connectors.salesforce.sqlite_functions import get_child_ids
 from onyx.connectors.salesforce.sqlite_functions import get_record
 from onyx.connectors.salesforce.sqlite_functions import init_db
 from onyx.connectors.salesforce.sqlite_functions import update_sf_db_with_csv
+from onyx.connectors.salesforce.utils import SalesforceObject
 from onyx.utils.logger import setup_logger
 
 logger = setup_logger()
 
 
 _DEFAULT_PARENT_OBJECT_TYPES = ["Account"]
+_ID_PREFIX = "SALESFORCE_"
 
 
 class SalesforceConnector(LoadConnector, PollConnector, SlimConnector):
@@ -60,6 +65,46 @@ class SalesforceConnector(LoadConnector, PollConnector, SlimConnector):
             raise ConnectorMissingCredentialError("Salesforce")
         return self._sf_client
 
+    def _extract_primary_owners(
+        self, sf_object: SalesforceObject
+    ) -> list[BasicExpertInfo] | None:
+        object_dict = sf_object.data
+        if not (last_modified_by_id := object_dict.get("LastModifiedById")):
+            return None
+        if not (last_modified_by := get_record(last_modified_by_id)):
+            return None
+        if not (last_modified_by_name := last_modified_by.data.get("Name")):
+            return None
+        primary_owners = [BasicExpertInfo(display_name=last_modified_by_name)]
+        return primary_owners
+
+    def _convert_object_instance_to_document(
+        self, sf_object: SalesforceObject
+    ) -> Document:
+        object_dict = sf_object.data
+        salesforce_id = object_dict["Id"]
+        onyx_salesforce_id = f"{_ID_PREFIX}{salesforce_id}"
+        base_url = f"https://{self.sf_client.sf_instance}"
+        extracted_doc_updated_at = time_str_to_utc(object_dict["LastModifiedDate"])
+        extracted_semantic_identifier = object_dict.get("Name", "Unknown Object")
+
+        sections = [extract_section(sf_object, base_url)]
+        for id in get_child_ids(sf_object.id):
+            if not (child_object := get_record(id)):
+                continue
+            sections.append(extract_section(child_object, base_url))
+
+        doc = Document(
+            id=onyx_salesforce_id,
+            sections=sections,
+            source=DocumentSource.SALESFORCE,
+            semantic_identifier=extracted_semantic_identifier,
+            doc_updated_at=extracted_doc_updated_at,
+            primary_owners=self._extract_primary_owners(sf_object),
+            metadata={},
+        )
+        return doc
+
     def _fetch_from_salesforce(
         self,
         start: SecondsSinceUnixEpoch | None = None,
@@ -81,9 +126,6 @@ class SalesforceConnector(LoadConnector, PollConnector, SlimConnector):
                 f"Found {len(child_types)} child types for {parent_object_type}"
             )
 
-        # Always want to make sure user is grabbed for permissioning purposes
-        all_object_types.add("User")
-
         logger.info(f"Found total of {len(all_object_types)} object types to fetch")
         logger.debug(f"All object types: {all_object_types}")
 
@@ -127,6 +169,9 @@ class SalesforceConnector(LoadConnector, PollConnector, SlimConnector):
                 logger.debug(
                     f"Added {len(new_ids)} new/updated records for {object_type}"
                 )
+                # Remove the csv file after it has been used
+                # to successfully update the db
+                os.remove(csv_path)
 
         logger.info(f"Found {len(updated_ids)} total updated records")
         logger.info(
@@ -151,10 +196,7 @@ class SalesforceConnector(LoadConnector, PollConnector, SlimConnector):
                     continue
 
                 docs_to_yield.append(
-                    convert_sf_object_to_doc(
-                        sf_object=parent_object,
-                        sf_instance=self.sf_client.sf_instance,
-                    )
+                    self._convert_object_instance_to_document(parent_object)
                 )
                 docs_processed += 1
 
@@ -183,7 +225,7 @@ class SalesforceConnector(LoadConnector, PollConnector, SlimConnector):
             query_result = self.sf_client.query_all(query)
             doc_metadata_list.extend(
                 SlimDocument(
-                    id=f"{ID_PREFIX}{instance_dict.get('Id', '')}",
+                    id=f"{_ID_PREFIX}{instance_dict.get('Id', '')}",
                     perm_sync_data={},
                 )
                 for instance_dict in query_result["records"]

backend/onyx/connectors/salesforce/doc_conversion.py

@@ -1,18 +1,8 @@
 import re
+from collections import OrderedDict
 
-from onyx.configs.constants import DocumentSource
-from onyx.connectors.cross_connector_utils.miscellaneous_utils import time_str_to_utc
-from onyx.connectors.models import BasicExpertInfo
-from onyx.connectors.models import Document
 from onyx.connectors.models import Section
-from onyx.connectors.salesforce.sqlite_functions import get_child_ids
-from onyx.connectors.salesforce.sqlite_functions import get_record
 from onyx.connectors.salesforce.utils import SalesforceObject
-from onyx.utils.logger import setup_logger
-
-logger = setup_logger()
-
-ID_PREFIX = "SALESFORCE_"
 
 # All of these types of keys are handled by specific fields in the doc
 # conversion process (E.g. URLs) or are not useful for the user (E.g. UUIDs)
@@ -113,72 +103,54 @@ def _extract_dict_text(raw_dict: dict) -> str:
     return natural_language_for_dict
 
 
-def _extract_section(salesforce_object: SalesforceObject, base_url: str) -> Section:
+def extract_section(salesforce_object: SalesforceObject, base_url: str) -> Section:
     return Section(
         text=_extract_dict_text(salesforce_object.data),
         link=f"{base_url}/{salesforce_object.id}",
     )
 
 
-def _extract_primary_owners(
-    sf_object: SalesforceObject,
-) -> list[BasicExpertInfo] | None:
-    object_dict = sf_object.data
-    if not (last_modified_by_id := object_dict.get("LastModifiedById")):
-        logger.warning(f"No LastModifiedById found for {sf_object.id}")
-        return None
-    if not (last_modified_by := get_record(last_modified_by_id)):
-        logger.warning(f"No LastModifiedBy found for {last_modified_by_id}")
-        return None
-
-    user_data = last_modified_by.data
-    expert_info = BasicExpertInfo(
-        first_name=user_data.get("FirstName"),
-        last_name=user_data.get("LastName"),
-        email=user_data.get("Email"),
-        display_name=user_data.get("Name"),
+def _field_value_is_child_object(field_value: dict) -> bool:
+    """
+    Checks if the field value is a child object.
+    """
+    return (
+        isinstance(field_value, OrderedDict)
+        and "records" in field_value.keys()
+        and isinstance(field_value["records"], list)
+        and len(field_value["records"]) > 0
+        and "Id" in field_value["records"][0].keys()
     )
 
-    # Check if all fields are None
-    if all(
-        value is None
-        for value in [
-            expert_info.first_name,
-            expert_info.last_name,
-            expert_info.email,
-            expert_info.display_name,
-        ]
-    ):
-        logger.warning(f"No identifying information found for user {user_data}")
-        return None
 
-    return [expert_info]
+def _extract_sections(salesforce_object: dict, base_url: str) -> list[Section]:
+    """
+    This goes through the salesforce_object and extracts the top level fields as a Section.
+    It also goes through the child objects and extracts them as Sections.
+    """
+    top_level_dict = {}
 
-
-def convert_sf_object_to_doc(
-    sf_object: SalesforceObject,
-    sf_instance: str,
-) -> Document:
-    object_dict = sf_object.data
-    salesforce_id = object_dict["Id"]
-    onyx_salesforce_id = f"{ID_PREFIX}{salesforce_id}"
-    base_url = f"https://{sf_instance}"
-    extracted_doc_updated_at = time_str_to_utc(object_dict["LastModifiedDate"])
-    extracted_semantic_identifier = object_dict.get("Name", "Unknown Object")
-
-    sections = [_extract_section(sf_object, base_url)]
-    for id in get_child_ids(sf_object.id):
-        if not (child_object := get_record(id)):
+    child_object_sections = []
+    for field_name, field_value in salesforce_object.items():
+        # If the field value is not a child object, add it to the top level dict
+        # to turn into text for the top level section
+        if not _field_value_is_child_object(field_value):
+            top_level_dict[field_name] = field_value
             continue
-        sections.append(_extract_section(child_object, base_url))
 
-    doc = Document(
-        id=onyx_salesforce_id,
-        sections=sections,
-        source=DocumentSource.SALESFORCE,
-        semantic_identifier=extracted_semantic_identifier,
-        doc_updated_at=extracted_doc_updated_at,
-        primary_owners=_extract_primary_owners(sf_object),
-        metadata={},
+        # If the field value is a child object, extract the child objects and add them as sections
+        for record in field_value["records"]:
+            child_object_id = record["Id"]
+            child_object_sections.append(
+                Section(
+                    text=f"Child Object(s): {field_name}\n{_extract_dict_text(record)}",
+                    link=f"{base_url}/{child_object_id}",
+                )
+            )
+
+    top_level_id = salesforce_object["Id"]
+    top_level_section = Section(
+        text=_extract_dict_text(top_level_dict),
+        link=f"{base_url}/{top_level_id}",
     )
-    return doc
+    return [top_level_section, *child_object_sections]

backend/onyx/connectors/salesforce/salesforce_calls.py

@@ -77,28 +77,25 @@ def _get_all_queryable_fields_of_sf_type(
     object_description = _get_sf_type_object_json(sf_client, sf_type)
     fields: list[dict[str, Any]] = object_description["fields"]
     valid_fields: set[str] = set()
-    field_names_to_remove: set[str] = set()
+    compound_field_names: set[str] = set()
     for field in fields:
         if compound_field_name := field.get("compoundFieldName"):
-            # We do want to get name fields even if they are compound
-            if not field.get("nameField"):
-                field_names_to_remove.add(compound_field_name)
+            compound_field_names.add(compound_field_name)
         if field.get("type", "base64") == "base64":
             continue
         if field_name := field.get("name"):
             valid_fields.add(field_name)
 
-    return list(valid_fields - field_names_to_remove)
+    return list(valid_fields - compound_field_names)
 
 
-def _check_if_object_type_is_empty(
-    sf_client: Salesforce, sf_type: str, time_filter: str
-) -> bool:
+def _check_if_object_type_is_empty(sf_client: Salesforce, sf_type: str) -> bool:
     """
-    Use the rest api to check to make sure the query will result in a non-empty response
+    Send a small query to check if the object type is empty so we don't
+    perform extra bulk queries
     """
     try:
-        query = f"SELECT Count() FROM {sf_type}{time_filter} LIMIT 1"
+        query = f"SELECT Count() FROM {sf_type} LIMIT 1"
         result = sf_client.query(query)
         if result["totalSize"] == 0:
             return False
@@ -137,7 +134,7 @@ def _bulk_retrieve_from_salesforce(
     sf_type: str,
     time_filter: str,
 ) -> tuple[str, list[str] | None]:
-    if not _check_if_object_type_is_empty(sf_client, sf_type, time_filter):
+    if not _check_if_object_type_is_empty(sf_client, sf_type):
         return sf_type, None
 
     if existing_csvs := _check_for_existing_csvs(sf_type):

backend/onyx/connectors/salesforce/shelve_stuff/old_test_salesforce_shelves.py

@@ -1,737 +0,0 @@
-import csv
-import os
-import shutil
-
-from onyx.connectors.salesforce.shelve_stuff.shelve_functions import find_ids_by_type
-from onyx.connectors.salesforce.shelve_stuff.shelve_functions import (
-    get_affected_parent_ids_by_type,
-)
-from onyx.connectors.salesforce.shelve_stuff.shelve_functions import get_child_ids
-from onyx.connectors.salesforce.shelve_stuff.shelve_functions import get_record
-from onyx.connectors.salesforce.shelve_stuff.shelve_functions import (
-    update_sf_db_with_csv,
-)
-from onyx.connectors.salesforce.utils import BASE_DATA_PATH
-from onyx.connectors.salesforce.utils import get_object_type_path
-
-_VALID_SALESFORCE_IDS = [
-    "001bm00000fd9Z3AAI",
-    "001bm00000fdYTdAAM",
-    "001bm00000fdYTeAAM",
-    "001bm00000fdYTfAAM",
-    "001bm00000fdYTgAAM",
-    "001bm00000fdYThAAM",
-    "001bm00000fdYTiAAM",
-    "001bm00000fdYTjAAM",
-    "001bm00000fdYTkAAM",
-    "001bm00000fdYTlAAM",
-    "001bm00000fdYTmAAM",
-    "001bm00000fdYTnAAM",
-    "001bm00000fdYToAAM",
-    "500bm00000XoOxtAAF",
-    "500bm00000XoOxuAAF",
-    "500bm00000XoOxvAAF",
-    "500bm00000XoOxwAAF",
-    "500bm00000XoOxxAAF",
-    "500bm00000XoOxyAAF",
-    "500bm00000XoOxzAAF",
-    "500bm00000XoOy0AAF",
-    "500bm00000XoOy1AAF",
-    "500bm00000XoOy2AAF",
-    "500bm00000XoOy3AAF",
-    "500bm00000XoOy4AAF",
-    "500bm00000XoOy5AAF",
-    "500bm00000XoOy6AAF",
-    "500bm00000XoOy7AAF",
-    "500bm00000XoOy8AAF",
-    "500bm00000XoOy9AAF",
-    "500bm00000XoOyAAAV",
-    "500bm00000XoOyBAAV",
-    "500bm00000XoOyCAAV",
-    "500bm00000XoOyDAAV",
-    "500bm00000XoOyEAAV",
-    "500bm00000XoOyFAAV",
-    "500bm00000XoOyGAAV",
-    "500bm00000XoOyHAAV",
-    "500bm00000XoOyIAAV",
-    "003bm00000EjHCjAAN",
-    "003bm00000EjHCkAAN",
-    "003bm00000EjHClAAN",
-    "003bm00000EjHCmAAN",
-    "003bm00000EjHCnAAN",
-    "003bm00000EjHCoAAN",
-    "003bm00000EjHCpAAN",
-    "003bm00000EjHCqAAN",
-    "003bm00000EjHCrAAN",
-    "003bm00000EjHCsAAN",
-    "003bm00000EjHCtAAN",
-    "003bm00000EjHCuAAN",
-    "003bm00000EjHCvAAN",
-    "003bm00000EjHCwAAN",
-    "003bm00000EjHCxAAN",
-    "003bm00000EjHCyAAN",
-    "003bm00000EjHCzAAN",
-    "003bm00000EjHD0AAN",
-    "003bm00000EjHD1AAN",
-    "003bm00000EjHD2AAN",
-    "550bm00000EXc2tAAD",
-    "006bm000006kyDpAAI",
-    "006bm000006kyDqAAI",
-    "006bm000006kyDrAAI",
-    "006bm000006kyDsAAI",
-    "006bm000006kyDtAAI",
-    "006bm000006kyDuAAI",
-    "006bm000006kyDvAAI",
-    "006bm000006kyDwAAI",
-    "006bm000006kyDxAAI",
-    "006bm000006kyDyAAI",
-    "006bm000006kyDzAAI",
-    "006bm000006kyE0AAI",
-    "006bm000006kyE1AAI",
-    "006bm000006kyE2AAI",
-    "006bm000006kyE3AAI",
-    "006bm000006kyE4AAI",
-    "006bm000006kyE5AAI",
-    "006bm000006kyE6AAI",
-    "006bm000006kyE7AAI",
-    "006bm000006kyE8AAI",
-    "006bm000006kyE9AAI",
-    "006bm000006kyEAAAY",
-    "006bm000006kyEBAAY",
-    "006bm000006kyECAAY",
-    "006bm000006kyEDAAY",
-    "006bm000006kyEEAAY",
-    "006bm000006kyEFAAY",
-    "006bm000006kyEGAAY",
-    "006bm000006kyEHAAY",
-    "006bm000006kyEIAAY",
-    "006bm000006kyEJAAY",
-    "005bm000009zy0TAAQ",
-    "005bm000009zy25AAA",
-    "005bm000009zy26AAA",
-    "005bm000009zy28AAA",
-    "005bm000009zy29AAA",
-    "005bm000009zy2AAAQ",
-    "005bm000009zy2BAAQ",
-]
-
-
-def clear_sf_db() -> None:
-    """
-    Clears the SF DB by deleting all files in the data directory.
-    """
-    shutil.rmtree(BASE_DATA_PATH)
-
-
-def create_csv_file(
-    object_type: str, records: list[dict], filename: str = "test_data.csv"
-) -> None:
-    """
-    Creates a CSV file for the given object type and records.
-
-    Args:
-        object_type: The Salesforce object type (e.g. "Account", "Contact")
-        records: List of dictionaries containing the record data
-        filename: Name of the CSV file to create (default: test_data.csv)
-    """
-    if not records:
-        return
-
-    # Get all unique fields from records
-    fields: set[str] = set()
-    for record in records:
-        fields.update(record.keys())
-    fields = set(sorted(list(fields)))  # Sort for consistent order
-
-    # Create CSV file
-    csv_path = os.path.join(get_object_type_path(object_type), filename)
-    with open(csv_path, "w", newline="", encoding="utf-8") as f:
-        writer = csv.DictWriter(f, fieldnames=fields)
-        writer.writeheader()
-        for record in records:
-            writer.writerow(record)
-
-    # Update the database with the CSV
-    update_sf_db_with_csv(object_type, csv_path)
-
-
-def create_csv_with_example_data() -> None:
-    """
-    Creates CSV files with example data, organized by object type.
-    """
-    example_data: dict[str, list[dict]] = {
-        "Account": [
-            {
-                "Id": _VALID_SALESFORCE_IDS[0],
-                "Name": "Acme Inc.",
-                "BillingCity": "New York",
-                "Industry": "Technology",
-            },
-            {
-                "Id": _VALID_SALESFORCE_IDS[1],
-                "Name": "Globex Corp",
-                "BillingCity": "Los Angeles",
-                "Industry": "Manufacturing",
-            },
-            {
-                "Id": _VALID_SALESFORCE_IDS[2],
-                "Name": "Initech",
-                "BillingCity": "Austin",
-                "Industry": "Software",
-            },
-            {
-                "Id": _VALID_SALESFORCE_IDS[3],
-                "Name": "TechCorp Solutions",
-                "BillingCity": "San Francisco",
-                "Industry": "Software",
-                "AnnualRevenue": 5000000,
-            },
-            {
-                "Id": _VALID_SALESFORCE_IDS[4],
-                "Name": "BioMed Research",
-                "BillingCity": "Boston",
-                "Industry": "Healthcare",
-                "AnnualRevenue": 12000000,
-            },
-            {
-                "Id": _VALID_SALESFORCE_IDS[5],
-                "Name": "Green Energy Co",
-                "BillingCity": "Portland",
-                "Industry": "Energy",
-                "AnnualRevenue": 8000000,
-            },
-            {
-                "Id": _VALID_SALESFORCE_IDS[6],
-                "Name": "DataFlow Analytics",
-                "BillingCity": "Seattle",
-                "Industry": "Technology",
-                "AnnualRevenue": 3000000,
-            },
-            {
-                "Id": _VALID_SALESFORCE_IDS[7],
-                "Name": "Cloud Nine Services",
-                "BillingCity": "Denver",
-                "Industry": "Cloud Computing",
-                "AnnualRevenue": 7000000,
-            },
-        ],
-        "Contact": [
-            {
-                "Id": _VALID_SALESFORCE_IDS[40],
-                "FirstName": "John",
-                "LastName": "Doe",
-                "Email": "john.doe@acme.com",
-                "Title": "CEO",
-            },
-            {
-                "Id": _VALID_SALESFORCE_IDS[41],
-                "FirstName": "Jane",
-                "LastName": "Smith",
-                "Email": "jane.smith@acme.com",
-                "Title": "CTO",
-            },
-            {
-                "Id": _VALID_SALESFORCE_IDS[42],
-                "FirstName": "Bob",
-                "LastName": "Johnson",
-                "Email": "bob.j@globex.com",
-                "Title": "Sales Director",
-            },
-            {
-                "Id": _VALID_SALESFORCE_IDS[43],
-                "FirstName": "Sarah",
-                "LastName": "Chen",
-                "Email": "sarah.chen@techcorp.com",
-                "Title": "Product Manager",
-                "Phone": "415-555-0101",
-            },
-            {
-                "Id": _VALID_SALESFORCE_IDS[44],
-                "FirstName": "Michael",
-                "LastName": "Rodriguez",
-                "Email": "m.rodriguez@biomed.com",
-                "Title": "Research Director",
-                "Phone": "617-555-0202",
-            },
-            {
-                "Id": _VALID_SALESFORCE_IDS[45],
-                "FirstName": "Emily",
-                "LastName": "Green",
-                "Email": "emily.g@greenenergy.com",
-                "Title": "Sustainability Lead",
-                "Phone": "503-555-0303",
-            },
-            {
-                "Id": _VALID_SALESFORCE_IDS[46],
-                "FirstName": "David",
-                "LastName": "Kim",
-                "Email": "david.kim@dataflow.com",
-                "Title": "Data Scientist",
-                "Phone": "206-555-0404",
-            },
-            {
-                "Id": _VALID_SALESFORCE_IDS[47],
-                "FirstName": "Rachel",
-                "LastName": "Taylor",
-                "Email": "r.taylor@cloudnine.com",
-                "Title": "Cloud Architect",
-                "Phone": "303-555-0505",
-            },
-        ],
-        "Opportunity": [
-            {
-                "Id": _VALID_SALESFORCE_IDS[62],
-                "Name": "Acme Server Upgrade",
-                "Amount": 50000,
-                "Stage": "Prospecting",
-                "CloseDate": "2024-06-30",
-            },
-            {
-                "Id": _VALID_SALESFORCE_IDS[63],
-                "Name": "Globex Manufacturing Line",
-                "Amount": 150000,
-                "Stage": "Negotiation",
-                "CloseDate": "2024-03-15",
-            },
-            {
-                "Id": _VALID_SALESFORCE_IDS[64],
-                "Name": "Initech Software License",
-                "Amount": 75000,
-                "Stage": "Closed Won",
-                "CloseDate": "2024-01-30",
-            },
-            {
-                "Id": _VALID_SALESFORCE_IDS[65],
-                "Name": "TechCorp AI Implementation",
-                "Amount": 250000,
-                "Stage": "Needs Analysis",
-                "CloseDate": "2024-08-15",
-                "Probability": 60,
-            },
-            {
-                "Id": _VALID_SALESFORCE_IDS[66],
-                "Name": "BioMed Lab Equipment",
-                "Amount": 500000,
-                "Stage": "Value Proposition",
-                "CloseDate": "2024-09-30",
-                "Probability": 75,
-            },
-            {
-                "Id": _VALID_SALESFORCE_IDS[67],
-                "Name": "Green Energy Solar Project",
-                "Amount": 750000,
-                "Stage": "Proposal",
-                "CloseDate": "2024-07-15",
-                "Probability": 80,
-            },
-            {
-                "Id": _VALID_SALESFORCE_IDS[68],
-                "Name": "DataFlow Analytics Platform",
-                "Amount": 180000,
-                "Stage": "Negotiation",
-                "CloseDate": "2024-05-30",
-                "Probability": 90,
-            },
-            {
-                "Id": _VALID_SALESFORCE_IDS[69],
-                "Name": "Cloud Nine Infrastructure",
-                "Amount": 300000,
-                "Stage": "Qualification",
-                "CloseDate": "2024-10-15",
-                "Probability": 40,
-            },
-        ],
-    }
-
-    # Create CSV files for each object type
-    for object_type, records in example_data.items():
-        create_csv_file(object_type, records)
-
-
-def test_query() -> None:
-    """
-    Tests querying functionality by verifying:
-    1. All expected Account IDs are found
-    2. Each Account's data matches what was inserted
-    """
-    # Expected test data for verification
-    expected_accounts: dict[str, dict[str, str | int]] = {
-        _VALID_SALESFORCE_IDS[0]: {
-            "Name": "Acme Inc.",
-            "BillingCity": "New York",
-            "Industry": "Technology",
-        },
-        _VALID_SALESFORCE_IDS[1]: {
-            "Name": "Globex Corp",
-            "BillingCity": "Los Angeles",
-            "Industry": "Manufacturing",
-        },
-        _VALID_SALESFORCE_IDS[2]: {
-            "Name": "Initech",
-            "BillingCity": "Austin",
-            "Industry": "Software",
-        },
-        _VALID_SALESFORCE_IDS[3]: {
-            "Name": "TechCorp Solutions",
-            "BillingCity": "San Francisco",
-            "Industry": "Software",
-            "AnnualRevenue": 5000000,
-        },
-        _VALID_SALESFORCE_IDS[4]: {
-            "Name": "BioMed Research",
-            "BillingCity": "Boston",
-            "Industry": "Healthcare",
-            "AnnualRevenue": 12000000,
-        },
-        _VALID_SALESFORCE_IDS[5]: {
-            "Name": "Green Energy Co",
-            "BillingCity": "Portland",
-            "Industry": "Energy",
-            "AnnualRevenue": 8000000,
-        },
-        _VALID_SALESFORCE_IDS[6]: {
-            "Name": "DataFlow Analytics",
-            "BillingCity": "Seattle",
-            "Industry": "Technology",
-            "AnnualRevenue": 3000000,
-        },
-        _VALID_SALESFORCE_IDS[7]: {
-            "Name": "Cloud Nine Services",
-            "BillingCity": "Denver",
-            "Industry": "Cloud Computing",
-            "AnnualRevenue": 7000000,
-        },
-    }
-
-    # Get all Account IDs
-    account_ids = find_ids_by_type("Account")
-
-    # Verify we found all expected accounts
-    assert len(account_ids) == len(
-        expected_accounts
-    ), f"Expected {len(expected_accounts)} accounts, found {len(account_ids)}"
-    assert set(account_ids) == set(
-        expected_accounts.keys()
-    ), "Found account IDs don't match expected IDs"
-
-    # Verify each account's data
-    for acc_id in account_ids:
-        combined = get_record(acc_id)
-        assert combined is not None, f"Could not find account {acc_id}"
-
-        expected = expected_accounts[acc_id]
-
-        # Verify account data matches
-        for key, value in expected.items():
-            value = str(value)
-            assert (
-                combined.data[key] == value
-            ), f"Account {acc_id} field {key} expected {value}, got {combined.data[key]}"
-
-    print("All query tests passed successfully!")
-
-
-def test_upsert() -> None:
-    """
-    Tests upsert functionality by:
-    1. Updating an existing account
-    2. Creating a new account
-    3. Verifying both operations were successful
-    """
-    # Create CSV for updating an existing account and adding a new one
-    update_data: list[dict[str, str | int]] = [
-        {
-            "Id": _VALID_SALESFORCE_IDS[0],
-            "Name": "Acme Inc. Updated",
-            "BillingCity": "New York",
-            "Industry": "Technology",
-            "Description": "Updated company info",
-        },
-        {
-            "Id": _VALID_SALESFORCE_IDS[2],
-            "Name": "New Company Inc.",
-            "BillingCity": "Miami",
-            "Industry": "Finance",
-            "AnnualRevenue": 1000000,
-        },
-    ]
-
-    create_csv_file("Account", update_data, "update_data.csv")
-
-    # Verify the update worked
-    updated_record = get_record(_VALID_SALESFORCE_IDS[0])
-    assert updated_record is not None, "Updated record not found"
-    assert updated_record.data["Name"] == "Acme Inc. Updated", "Name not updated"
-    assert (
-        updated_record.data["Description"] == "Updated company info"
-    ), "Description not added"
-
-    # Verify the new record was created
-    new_record = get_record(_VALID_SALESFORCE_IDS[2])
-    assert new_record is not None, "New record not found"
-    assert new_record.data["Name"] == "New Company Inc.", "New record name incorrect"
-    assert new_record.data["AnnualRevenue"] == "1000000", "New record revenue incorrect"
-
-    print("All upsert tests passed successfully!")
-
-
-def test_relationships() -> None:
-    """
-    Tests relationship shelf updates and queries by:
-    1. Creating test data with relationships
-    2. Verifying the relationships are correctly stored
-    3. Testing relationship queries
-    """
-    # Create test data for each object type
-    test_data: dict[str, list[dict[str, str | int]]] = {
-        "Case": [
-            {
-                "Id": _VALID_SALESFORCE_IDS[13],
-                "AccountId": _VALID_SALESFORCE_IDS[0],
-                "Subject": "Test Case 1",
-            },
-            {
-                "Id": _VALID_SALESFORCE_IDS[14],
-                "AccountId": _VALID_SALESFORCE_IDS[0],
-                "Subject": "Test Case 2",
-            },
-        ],
-        "Contact": [
-            {
-                "Id": _VALID_SALESFORCE_IDS[48],
-                "AccountId": _VALID_SALESFORCE_IDS[0],
-                "FirstName": "Test",
-                "LastName": "Contact",
-            }
-        ],
-        "Opportunity": [
-            {
-                "Id": _VALID_SALESFORCE_IDS[62],
-                "AccountId": _VALID_SALESFORCE_IDS[0],
-                "Name": "Test Opportunity",
-                "Amount": 100000,
-            }
-        ],
-    }
-
-    # Create and update CSV files for each object type
-    for object_type, records in test_data.items():
-        create_csv_file(object_type, records, "relationship_test.csv")
-
-    # Test relationship queries
-    # All these objects should be children of Acme Inc.
-    child_ids = get_child_ids(_VALID_SALESFORCE_IDS[0])
-    assert len(child_ids) == 4, f"Expected 4 child objects, found {len(child_ids)}"
-    assert _VALID_SALESFORCE_IDS[13] in child_ids, "Case 1 not found in relationship"
-    assert _VALID_SALESFORCE_IDS[14] in child_ids, "Case 2 not found in relationship"
-    assert _VALID_SALESFORCE_IDS[48] in child_ids, "Contact not found in relationship"
-    assert (
-        _VALID_SALESFORCE_IDS[62] in child_ids
-    ), "Opportunity not found in relationship"
-
-    # Test querying relationships for a different account (should be empty)
-    other_account_children = get_child_ids(_VALID_SALESFORCE_IDS[1])
-    assert (
-        len(other_account_children) == 0
-    ), "Expected no children for different account"
-
-    print("All relationship tests passed successfully!")
-
-
-def test_account_with_children() -> None:
-    """
-    Tests querying all accounts and retrieving their child objects.
-    This test verifies that:
-    1. All accounts can be retrieved
-    2. Child objects are correctly linked
-    3. Child object data is complete and accurate
-    """
-    # First get all account IDs
-    account_ids = find_ids_by_type("Account")
-    assert len(account_ids) > 0, "No accounts found"
-
-    # For each account, get its children and verify the data
-    for account_id in account_ids:
-        account = get_record(account_id)
-        assert account is not None, f"Could not find account {account_id}"
-
-        # Get all child objects
-        child_ids = get_child_ids(account_id)
-
-        # For Acme Inc., verify specific relationships
-        if account_id == _VALID_SALESFORCE_IDS[0]:  # Acme Inc.
-            assert (
-                len(child_ids) == 4
-            ), f"Expected 4 children for Acme Inc., found {len(child_ids)}"
-
-            # Get all child records
-            child_records = []
-            for child_id in child_ids:
-                child_record = get_record(child_id)
-                if child_record is not None:
-                    child_records.append(child_record)
-            # Verify Cases
-            cases = [r for r in child_records if r.type == "Case"]
-            assert (
-                len(cases) == 2
-            ), f"Expected 2 cases for Acme Inc., found {len(cases)}"
-            case_subjects = {case.data["Subject"] for case in cases}
-            assert "Test Case 1" in case_subjects, "Test Case 1 not found"
-            assert "Test Case 2" in case_subjects, "Test Case 2 not found"
-
-            # Verify Contacts
-            contacts = [r for r in child_records if r.type == "Contact"]
-            assert (
-                len(contacts) == 1
-            ), f"Expected 1 contact for Acme Inc., found {len(contacts)}"
-            contact = contacts[0]
-            assert contact.data["FirstName"] == "Test", "Contact first name mismatch"
-            assert contact.data["LastName"] == "Contact", "Contact last name mismatch"
-
-            # Verify Opportunities
-            opportunities = [r for r in child_records if r.type == "Opportunity"]
-            assert (
-                len(opportunities) == 1
-            ), f"Expected 1 opportunity for Acme Inc., found {len(opportunities)}"
-            opportunity = opportunities[0]
-            assert (
-                opportunity.data["Name"] == "Test Opportunity"
-            ), "Opportunity name mismatch"
-            assert opportunity.data["Amount"] == "100000", "Opportunity amount mismatch"
-
-    print("All account with children tests passed successfully!")
-
-
-def test_relationship_updates() -> None:
-    """
-    Tests that relationships are properly updated when a child object's parent reference changes.
-    This test verifies:
-    1. Initial relationship is created correctly
-    2. When parent reference is updated, old relationship is removed
-    3. New relationship is created correctly
-    """
-    # Create initial test data - Contact linked to Acme Inc.
-    initial_contact = [
-        {
-            "Id": _VALID_SALESFORCE_IDS[40],
-            "AccountId": _VALID_SALESFORCE_IDS[0],
-            "FirstName": "Test",
-            "LastName": "Contact",
-        }
-    ]
-    create_csv_file("Contact", initial_contact, "initial_contact.csv")
-
-    # Verify initial relationship
-    acme_children = get_child_ids(_VALID_SALESFORCE_IDS[0])
-    assert (
-        _VALID_SALESFORCE_IDS[40] in acme_children
-    ), "Initial relationship not created"
-
-    # Update contact to be linked to Globex Corp instead
-    updated_contact = [
-        {
-            "Id": _VALID_SALESFORCE_IDS[40],
-            "AccountId": _VALID_SALESFORCE_IDS[1],
-            "FirstName": "Test",
-            "LastName": "Contact",
-        }
-    ]
-    create_csv_file("Contact", updated_contact, "updated_contact.csv")
-
-    # Verify old relationship is removed
-    acme_children = get_child_ids(_VALID_SALESFORCE_IDS[0])
-    assert (
-        _VALID_SALESFORCE_IDS[40] not in acme_children
-    ), "Old relationship not removed"
-
-    # Verify new relationship is created
-    globex_children = get_child_ids(_VALID_SALESFORCE_IDS[1])
-    assert _VALID_SALESFORCE_IDS[40] in globex_children, "New relationship not created"
-
-    print("All relationship update tests passed successfully!")
-
-
-def test_get_affected_parent_ids() -> None:
-    """
-    Tests get_affected_parent_ids functionality by verifying:
-    1. IDs that are directly in the parent_types list are included
-    2. IDs that have children in the updated_ids list are included
-    3. IDs that are neither of the above are not included
-    """
-    # Create test data with relationships
-    test_data = {
-        "Account": [
-            {
-                "Id": _VALID_SALESFORCE_IDS[0],
-                "Name": "Parent Account 1",
-            },
-            {
-                "Id": _VALID_SALESFORCE_IDS[1],
-                "Name": "Parent Account 2",
-            },
-            {
-                "Id": _VALID_SALESFORCE_IDS[2],
-                "Name": "Not Affected Account",
-            },
-        ],
-        "Contact": [
-            {
-                "Id": _VALID_SALESFORCE_IDS[40],
-                "AccountId": _VALID_SALESFORCE_IDS[0],
-                "FirstName": "Child",
-                "LastName": "Contact",
-            }
-        ],
-    }
-
-    # Create and update CSV files for test data
-    for object_type, records in test_data.items():
-        create_csv_file(object_type, records)
-
-    # Test Case 1: Account directly in updated_ids and parent_types
-    updated_ids = {_VALID_SALESFORCE_IDS[1]}  # Parent Account 2
-    parent_types = ["Account"]
-    affected_ids = get_affected_parent_ids_by_type(updated_ids, parent_types)
-    assert _VALID_SALESFORCE_IDS[1] in affected_ids, "Direct parent ID not included"
-
-    # Test Case 2: Account with child in updated_ids
-    updated_ids = {_VALID_SALESFORCE_IDS[40]}  # Child Contact
-    parent_types = ["Account"]
-    affected_ids = get_affected_parent_ids_by_type(updated_ids, parent_types)
-    assert (
-        _VALID_SALESFORCE_IDS[0] in affected_ids
-    ), "Parent of updated child not included"
-
-    # Test Case 3: Both direct and indirect affects
-    updated_ids = {_VALID_SALESFORCE_IDS[1], _VALID_SALESFORCE_IDS[40]}  # Both cases
-    parent_types = ["Account"]
-    affected_ids = get_affected_parent_ids_by_type(updated_ids, parent_types)
-    assert len(affected_ids) == 2, "Expected exactly two affected parent IDs"
-    assert _VALID_SALESFORCE_IDS[0] in affected_ids, "Parent of child not included"
-    assert _VALID_SALESFORCE_IDS[1] in affected_ids, "Direct parent ID not included"
-    assert (
-        _VALID_SALESFORCE_IDS[2] not in affected_ids
-    ), "Unaffected ID incorrectly included"
-
-    # Test Case 4: No matches
-    updated_ids = {_VALID_SALESFORCE_IDS[40]}  # Child Contact
-    parent_types = ["Opportunity"]  # Wrong type
-    affected_ids = get_affected_parent_ids_by_type(updated_ids, parent_types)
-    assert len(affected_ids) == 0, "Should return empty list when no matches"
-
-    print("All get_affected_parent_ids tests passed successfully!")
-
-
-def main_build() -> None:
-    clear_sf_db()
-    create_csv_with_example_data()
-    test_query()
-    test_upsert()
-    test_relationships()
-    test_account_with_children()
-    test_relationship_updates()
-    test_get_affected_parent_ids()
-
-
-if __name__ == "__main__":
-    main_build()

backend/onyx/connectors/salesforce/sqlite_functions.py

@@ -40,20 +40,20 @@ def get_db_connection(
 
 def init_db() -> None:
     """Initialize the SQLite database with required tables if they don't exist."""
+    if os.path.exists(get_sqlite_db_path()):
+        return
+
     # Create database directory if it doesn't exist
     os.makedirs(os.path.dirname(get_sqlite_db_path()), exist_ok=True)
 
     with get_db_connection("EXCLUSIVE") as conn:
         cursor = conn.cursor()
 
-        db_exists = os.path.exists(get_sqlite_db_path())
-
-        if not db_exists:
-            # Enable WAL mode for better concurrent access and write performance
-            cursor.execute("PRAGMA journal_mode=WAL")
-            cursor.execute("PRAGMA synchronous=NORMAL")
-            cursor.execute("PRAGMA temp_store=MEMORY")
-            cursor.execute("PRAGMA cache_size=-2000000")  # Use 2GB memory for cache
+        # Enable WAL mode for better concurrent access and write performance
+        cursor.execute("PRAGMA journal_mode=WAL")
+        cursor.execute("PRAGMA synchronous=NORMAL")
+        cursor.execute("PRAGMA temp_store=MEMORY")
+        cursor.execute("PRAGMA cache_size=-2000000")  # Use 2GB memory for cache
 
         # Main table for storing Salesforce objects
         cursor.execute(
@@ -90,69 +90,49 @@ def init_db() -> None:
         """
         )
 
-        # Create a table for User email to ID mapping if it doesn't exist
+        # Always recreate indexes to ensure they exist
+        cursor.execute("DROP INDEX IF EXISTS idx_object_type")
+        cursor.execute("DROP INDEX IF EXISTS idx_parent_id")
+        cursor.execute("DROP INDEX IF EXISTS idx_child_parent")
+        cursor.execute("DROP INDEX IF EXISTS idx_object_type_id")
+        cursor.execute("DROP INDEX IF EXISTS idx_relationship_types_lookup")
+
+        # Create covering indexes for common queries
         cursor.execute(
-            """
-            CREATE TABLE IF NOT EXISTS user_email_map (
-                email TEXT PRIMARY KEY,
-                user_id TEXT,  -- Nullable to allow for users without IDs
-                FOREIGN KEY (user_id) REFERENCES salesforce_objects(id)
-            ) WITHOUT ROWID
-        """
-        )
-
-        # Create indexes if they don't exist (SQLite ignores IF NOT EXISTS for indexes)
-        def create_index_if_not_exists(index_name: str, create_statement: str) -> None:
-            cursor.execute(
-                f"SELECT name FROM sqlite_master WHERE type='index' AND name='{index_name}'"
-            )
-            if not cursor.fetchone():
-                cursor.execute(create_statement)
-
-        create_index_if_not_exists(
-            "idx_object_type",
             """
             CREATE INDEX idx_object_type
             ON salesforce_objects(object_type, id)
             WHERE object_type IS NOT NULL
-            """,
+        """
         )
 
-        create_index_if_not_exists(
-            "idx_parent_id",
+        cursor.execute(
             """
             CREATE INDEX idx_parent_id
             ON relationships(parent_id, child_id)
-            """,
+        """
         )
 
-        create_index_if_not_exists(
-            "idx_child_parent",
+        cursor.execute(
             """
             CREATE INDEX idx_child_parent
             ON relationships(child_id)
             WHERE child_id IS NOT NULL
-            """,
+        """
         )
 
-        create_index_if_not_exists(
-            "idx_relationship_types_lookup",
+        # New composite index for fast parent type lookups
+        cursor.execute(
             """
             CREATE INDEX idx_relationship_types_lookup
             ON relationship_types(parent_type, child_id, parent_id)
-            """,
+        """
         )
 
         # Analyze tables to help query planner
         cursor.execute("ANALYZE relationships")
         cursor.execute("ANALYZE salesforce_objects")
         cursor.execute("ANALYZE relationship_types")
-        cursor.execute("ANALYZE user_email_map")
-
-        # If database already existed but user_email_map needs to be populated
-        cursor.execute("SELECT COUNT(*) FROM user_email_map")
-        if cursor.fetchone()[0] == 0:
-            _update_user_email_map(conn)
 
         conn.commit()
 
@@ -223,27 +203,7 @@ def _update_relationship_tables(
         raise
 
 
-def _update_user_email_map(conn: sqlite3.Connection) -> None:
-    """Update the user_email_map table with current User objects.
-    Called internally by update_sf_db_with_csv when User objects are updated.
-    """
-    cursor = conn.cursor()
-    cursor.execute(
-        """
-        INSERT OR REPLACE INTO user_email_map (email, user_id)
-        SELECT json_extract(data, '$.Email'), id
-        FROM salesforce_objects
-        WHERE object_type = 'User'
-        AND json_extract(data, '$.Email') IS NOT NULL
-        """
-    )
-
-
-def update_sf_db_with_csv(
-    object_type: str,
-    csv_download_path: str,
-    delete_csv_after_use: bool = True,
-) -> list[str]:
+def update_sf_db_with_csv(object_type: str, csv_download_path: str) -> list[str]:
     """Update the SF DB with a CSV file using SQLite storage."""
     updated_ids = []
 
@@ -289,17 +249,8 @@ def update_sf_db_with_csv(
                 _update_relationship_tables(conn, id, parent_ids)
                 updated_ids.append(id)
 
-        # If we're updating User objects, update the email map
-        if object_type == "User":
-            _update_user_email_map(conn)
-
         conn.commit()
 
-    if delete_csv_after_use:
-        # Remove the csv file after it has been used
-        # to successfully update the db
-        os.remove(csv_download_path)
-
     return updated_ids
 
 
@@ -378,9 +329,6 @@ def get_affected_parent_ids_by_type(
         cursor = conn.cursor()
 
         for batch_ids in updated_ids_batches:
-            batch_ids = list(set(batch_ids) - updated_parent_ids)
-            if not batch_ids:
-                continue
             id_placeholders = ",".join(["?" for _ in batch_ids])
 
             for parent_type in parent_types:
@@ -436,40 +384,3 @@ def has_at_least_one_object_of_type(object_type: str) -> bool:
         )
         count = cursor.fetchone()[0]
         return count > 0
-
-
-# NULL_ID_STRING is used to indicate that the user ID was queried but not found
-# As opposed to None because it has yet to be queried at all
-NULL_ID_STRING = "N/A"
-
-
-def get_user_id_by_email(email: str) -> str | None:
-    """Get the Salesforce User ID for a given email address.
-
-    Args:
-        email: The email address to look up
-
-    Returns:
-        A tuple of (was_found, user_id):
-            - was_found: True if the email exists in the table, False if not found
-            - user_id: The Salesforce User ID if exists, None otherwise
-    """
-    with get_db_connection() as conn:
-        cursor = conn.cursor()
-        cursor.execute("SELECT user_id FROM user_email_map WHERE email = ?", (email,))
-        result = cursor.fetchone()
-        if result is None:
-            return None
-        return result[0]
-
-
-def update_email_to_id_table(email: str, id: str | None) -> None:
-    """Update the email to ID map table with a new email and ID."""
-    id_to_use = id or NULL_ID_STRING
-    with get_db_connection() as conn:
-        cursor = conn.cursor()
-        cursor.execute(
-            "INSERT OR REPLACE INTO user_email_map (email, user_id) VALUES (?, ?)",
-            (email, id_to_use),
-        )
-        conn.commit()

backend/onyx/context/search/pipeline.py

@@ -37,7 +37,6 @@ from onyx.utils.logger import setup_logger
 from onyx.utils.threadpool_concurrency import FunctionCall
 from onyx.utils.threadpool_concurrency import run_functions_in_parallel
 from onyx.utils.timing import log_function_time
-from onyx.utils.variable_functionality import fetch_ee_implementation_or_noop
 
 logger = setup_logger()
 
@@ -164,17 +163,6 @@ class SearchPipeline:
         # These chunks are ordered, deduped, and contain no large chunks
         retrieved_chunks = self._get_chunks()
 
-        # If ee is enabled, censor the chunk sections based on user access
-        # Otherwise, return the retrieved chunks
-        censored_chunks = fetch_ee_implementation_or_noop(
-            "onyx.external_permissions.post_query_censoring",
-            "_post_query_chunk_censoring",
-            retrieved_chunks,
-        )(
-            chunks=retrieved_chunks,
-            user=self.user,
-        )
-
         above = self.search_query.chunks_above
         below = self.search_query.chunks_below
 
@@ -187,7 +175,7 @@ class SearchPipeline:
             seen_document_ids = set()
 
             # This preserves the ordering since the chunks are retrieved in score order
-            for chunk in censored_chunks:
+            for chunk in retrieved_chunks:
                 if chunk.document_id not in seen_document_ids:
                     seen_document_ids.add(chunk.document_id)
                     chunk_requests.append(
@@ -237,7 +225,7 @@ class SearchPipeline:
         #   This maintains the original chunks ordering. Note, we cannot simply sort by score here
         #   as reranking flow may wipe the scores for a lot of the chunks.
         doc_chunk_ranges_map = defaultdict(list)
-        for chunk in censored_chunks:
+        for chunk in retrieved_chunks:
             # The list of ranges for each document is ordered by score
             doc_chunk_ranges_map[chunk.document_id].append(
                 ChunkRange(
@@ -286,11 +274,11 @@ class SearchPipeline:
 
         # In case of failed parallel calls to Vespa, at least we should have the initial retrieved chunks
         doc_chunk_ind_to_chunk.update(
-            {(chunk.document_id, chunk.chunk_id): chunk for chunk in censored_chunks}
+            {(chunk.document_id, chunk.chunk_id): chunk for chunk in retrieved_chunks}
         )
 
         # Build the surroundings for all of the initial retrieved chunks
-        for chunk in censored_chunks:
+        for chunk in retrieved_chunks:
             start_ind = max(0, chunk.chunk_id - above)
             end_ind = chunk.chunk_id + below