github-starred/open-webui
2
0
Fork 0
mirror of https://github.com/open-webui/open-webui.git synced 2026-05-06 10:58:17 -05:00
Code Issues 1.2k Packages Projects Releases 126 Wiki Activity

[PR #23600] [CLOSED] feat: per-group token usage limits with enforcement #42902

New Issue
Closed
opened 2026-04-25 14:40:22 -05:00 by GiteaMirror · 0 comments
GiteaMirror commented 2026-04-25 14:40:22 -05:00
Owner
Copy Link

📋 Pull Request Information

Original PR: https://github.com/open-webui/open-webui/pull/23600
Author: @smorello87
Created: 4/11/2026
Status: ❌ Closed

Base: dev ← Head: feat/usage-limits

📝 Commits (1)

  • 3e4197b feat: per-group token usage limits with enforcement

📊 Changes

13 files changed (+760 additions, -12 deletions)

View changed files

📝 backend/open_webui/main.py (+2 -0)
➕ backend/open_webui/migrations/versions/c3d4e5f6a7b8_add_usage_ledger_table.py (+102 -0)
📝 backend/open_webui/models/chat_messages.py (+1 -1)
📝 backend/open_webui/models/chats.py (+22 -8)
📝 backend/open_webui/models/groups.py (+20 -0)
➕ backend/open_webui/models/usage_ledger.py (+144 -0)
➕ backend/open_webui/routers/usage_limits.py (+151 -0)
📝 backend/open_webui/utils/chat.py (+17 -1)
📝 backend/open_webui/utils/middleware.py (+60 -0)
➕ backend/open_webui/utils/usage_limits.py (+153 -0)
📝 src/lib/components/admin/Users/Groups/EditGroupModal.svelte (+2 -1)
📝 src/lib/components/admin/Users/Groups/Permissions.svelte (+81 -0)
📝 src/lib/components/chat/Chat.svelte (+5 -1)

📄 Description

  • Target branch: dev
  • Description: See below
  • Testing: Extensively tested on ECS Fargate deployment with OpenRouter and AWS Bedrock
  • Code review: Self-reviewed
  • Title Prefix: feat:

Changelog Entry

Added

  • Per-group token usage limits with configurable daily/monthly enforcement
  • Admin UI for setting token limits, soft limits, and priority per group
  • Soft limit warning toast and hard limit 429 block
  • Usage tracking for direct API calls and temporary chats
  • Usage data preserved when chats are deleted (content stripped, usage metadata retained)

Description

Adds per-group token usage limits that enforce daily or monthly caps on non-admin users. Full design discussion in #23558.

How it works:

  • Enforcement: Pre-request check in generate_chat_completion queries chat_message for the user's token usage in the current period. If over the group's limit, returns 429. Background tasks (title/tag/follow-up generation) are skipped via metadata.task check.
  • Usage data preservation: Chat deletion strips message content but preserves usage metadata (user_id, model_id, tokens, timestamps). Prevents users from resetting usage by deleting chats.
  • Temp chats & API calls: Since chat_message has a FK to chat (which doesn't exist for these), usage is written to a small usage_ledger table. Enforcement sums both tables.
  • Configuration: Stored in the existing group permissions JSON field. Admin UI added to the Permissions tab in group settings.
  • Soft/hard limits: Configurable soft limit shows a warning toast; hard limit blocks with 429.
  • Priority: When a user belongs to multiple groups, the limit from the highest-priority group (lowest priority number) applies.

Testing

Tested on an ECS Fargate deployment (CUNY AI Lab, PostgreSQL 16.3) with:

  • OpenRouter (Gemma 4 31B) and AWS Bedrock (Nova Pro, Nova Premier)
  • Token counts verified to match analytics dashboard
  • Background tasks (title/tag/follow-up gen) excluded from counting and enforcement — verified via CloudWatch logs
  • Chat deletion confirmed to preserve usage counts (FK changed to SET NULL, verified via pg_constraint query)
  • Soft limit warning toasts and hard limit 429 blocks working
  • Admin users bypass all limits
  • Direct API calls (POST /api/chat/completions with JWT, no chat_id, stream: false) — both enforced and tracked
  • Alembic migration tested on PostgreSQL (uses raw SQL for FK alteration, batch mode for SQLite)
  • Automated code review via pr-validator-bot — both findings addressed and retested

Testing

Tested on an ECS Fargate deployment (CUNY AI Lab, PostgreSQL 16.3) with:

  • OpenRouter (Gemma 4 31B) and AWS Bedrock (Nova Pro, Nova Premier)
  • Token counts verified to match analytics dashboard
  • Background tasks (title/tag/follow-up gen) excluded from counting and enforcement — verified via CloudWatch logs
  • Chat deletion confirmed to preserve usage counts (FK changed to SET NULL, verified via pg_constraint query)
  • Soft limit warning toasts and hard limit 429 blocks working
  • Admin users bypass all limits
  • Direct API calls (POST /api/chat/completions with JWT, no chat_id, stream: false) — both enforced and tracked
  • Alembic migration tested on PostgreSQL (uses raw SQL for FK alteration, batch mode for SQLite)
  • Automated code review via pr-validator-bot — both findings addressed and retested

Testing

Tested on an ECS Fargate deployment (CUNY AI Lab, PostgreSQL 16.3) with:

  • OpenRouter (Gemma 4 31B) and AWS Bedrock (Nova Pro, Nova Premier)
  • Token counts verified to match analytics dashboard
  • Background tasks (title/tag/follow-up gen) excluded from counting and enforcement — verified via CloudWatch logs
  • Chat deletion confirmed to preserve usage counts (FK changed to SET NULL, verified via pg_constraint query)
  • Soft limit warning toasts and hard limit 429 blocks working
  • Admin users bypass all limits
  • Direct API calls (POST /api/chat/completions with JWT, no chat_id, stream: false) — both enforced and tracked
  • Alembic migration tested on PostgreSQL (uses raw SQL for FK alteration, batch mode for SQLite)
  • Automated code review via pr-validator-bot — both findings addressed and retested

Testing

Tested on an ECS Fargate deployment (CUNY AI Lab, PostgreSQL 16.3) with:

  • OpenRouter (Gemma 4 31B) and AWS Bedrock (Nova Pro, Nova Premier)
  • Token counts verified to match analytics dashboard
  • Background tasks (title/tag/follow-up gen) excluded from counting and enforcement — verified via CloudWatch logs
  • Chat deletion confirmed to preserve usage counts (FK changed to SET NULL, verified via pg_constraint query)
  • Soft limit warning toasts and hard limit 429 blocks working
  • Admin users bypass all limits
  • Direct API calls (POST /api/chat/completions with JWT, no chat_id, stream: false) — both enforced and tracked
  • Alembic migration tested on PostgreSQL (uses raw SQL for FK alteration, batch mode for SQLite)
  • Automated code review via pr-validator-bot — both findings addressed and retested

Testing

Tested on an ECS Fargate deployment (CUNY AI Lab, PostgreSQL 16.3) with:

  • OpenRouter (Gemma 4 31B) and AWS Bedrock (Nova Pro, Nova Premier)
  • Token counts verified to match analytics dashboard
  • Background tasks (title/tag/follow-up gen) excluded from counting and enforcement — verified via CloudWatch logs
  • Chat deletion confirmed to preserve usage counts (FK changed to SET NULL, verified via pg_constraint query)
  • Soft limit warning toasts and hard limit 429 blocks working
  • Admin users bypass all limits
  • Direct API calls (POST /api/chat/completions with JWT, no chat_id, stream: false) — both enforced and tracked
  • Alembic migration tested on PostgreSQL (uses raw SQL for FK alteration, batch mode for SQLite)
  • Automated code review via pr-validator-bot — both findings addressed and retested

Testing

Tested on an ECS Fargate deployment (CUNY AI Lab, PostgreSQL 16.3) with:

  • OpenRouter (Gemma 4 31B) and AWS Bedrock (Nova Pro, Nova Premier)
  • Token counts verified to match analytics dashboard
  • Background tasks (title/tag/follow-up gen) excluded from counting and enforcement — verified via CloudWatch logs
  • Chat deletion confirmed to preserve usage counts (FK changed to SET NULL, verified via pg_constraint query)
  • Soft limit warning toasts and hard limit 429 blocks working
  • Admin users bypass all limits
  • Direct API calls (POST /api/chat/completions with JWT, no chat_id, stream: false) — both enforced and tracked
  • Alembic migration tested on PostgreSQL (uses raw SQL for FK alteration, batch mode for SQLite)
  • Automated code review via pr-validator-bot — both findings addressed and retested

Testing

Tested on an ECS Fargate deployment (CUNY AI Lab, PostgreSQL 16.3) with:

  • OpenRouter (Gemma 4 31B) and AWS Bedrock (Nova Pro, Nova Premier)
  • Token counts verified to match analytics dashboard
  • Background tasks (title/tag/follow-up gen) excluded from counting and enforcement — verified via CloudWatch logs
  • Chat deletion confirmed to preserve usage counts (FK changed to SET NULL, verified via pg_constraint query)
  • Soft limit warning toasts and hard limit 429 blocks working
  • Admin users bypass all limits
  • Direct API calls (POST /api/chat/completions with JWT, no chat_id, stream: false) — both enforced and tracked
  • Alembic migration tested on PostgreSQL (uses raw SQL for FK alteration, batch mode for SQLite)
  • Automated code review via pr-validator-bot — both findings addressed and retested

Testing

Tested on an ECS Fargate deployment (CUNY AI Lab, PostgreSQL 16.3) with:

  • OpenRouter (Gemma 4 31B) and AWS Bedrock (Nova Pro, Nova Premier)
  • Token counts verified to match analytics dashboard
  • Background tasks (title/tag/follow-up gen) excluded from counting and enforcement — verified via CloudWatch logs
  • Chat deletion confirmed to preserve usage counts (FK changed to SET NULL, verified via pg_constraint query)
  • Soft limit warning toasts and hard limit 429 blocks working
  • Admin users bypass all limits
  • Direct API calls (POST /api/chat/completions with JWT, no chat_id, stream: false) — both enforced and tracked
  • Alembic migration tested on PostgreSQL (uses raw SQL for FK alteration, batch mode for SQLite)
  • Automated code review via pr-validator-bot — both findings addressed and retested

Testing

Tested on an ECS Fargate deployment (CUNY AI Lab, PostgreSQL 16.3) with:

  • OpenRouter (Gemma 4 31B) and AWS Bedrock (Nova Pro, Nova Premier)
  • Token counts verified to match analytics dashboard
  • Background tasks (title/tag/follow-up gen) excluded from counting and enforcement — verified via CloudWatch logs
  • Chat deletion confirmed to preserve usage counts (FK changed to SET NULL, verified via pg_constraint query)
  • Soft limit warning toasts and hard limit 429 blocks working
  • Admin users bypass all limits
  • Direct API calls (POST /api/chat/completions with JWT, no chat_id, stream: false) — both enforced and tracked
  • Alembic migration tested on PostgreSQL (uses raw SQL for FK alteration, batch mode for SQLite)
  • Automated code review via pr-validator-bot — both findings addressed and retested

Testing

Tested on an ECS Fargate deployment (CUNY AI Lab, PostgreSQL 16.3) with:

  • OpenRouter (Gemma 4 31B) and AWS Bedrock (Nova Pro, Nova Premier)
  • Token counts verified to match analytics dashboard
  • Background tasks (title/tag/follow-up gen) excluded from counting and enforcement — verified via CloudWatch logs
  • Chat deletion confirmed to preserve usage counts (FK changed to SET NULL, verified via pg_constraint query)
  • Soft limit warning toasts and hard limit 429 blocks working
  • Admin users bypass all limits
  • Direct API calls (POST /api/chat/completions with JWT, no chat_id, stream: false) — both enforced and tracked
  • Alembic migration tested on PostgreSQL (uses raw SQL for FK alteration, batch mode for SQLite)
  • Automated code review via pr-validator-bot — both findings addressed and retested

Testing

Tested on an ECS Fargate deployment (CUNY AI Lab, PostgreSQL 16.3) with:

  • OpenRouter (Gemma 4 31B) and AWS Bedrock (Nova Pro, Nova Premier)
  • Token counts verified to match analytics dashboard
  • Background tasks (title/tag/follow-up gen) excluded from counting and enforcement — verified via CloudWatch logs
  • Chat deletion confirmed to preserve usage counts (FK changed to SET NULL, verified via pg_constraint query)
  • Soft limit warning toasts and hard limit 429 blocks working
  • Admin users bypass all limits
  • Direct API calls (POST /api/chat/completions with JWT, no chat_id, stream: false) — both enforced and tracked
  • Alembic migration tested on PostgreSQL (uses raw SQL for FK alteration, batch mode for SQLite)
  • Automated code review via pr-validator-bot — both findings addressed and retested

Breaking Changes

  • Chat message content is now stripped (set to NULL) on deletion instead of row deletion. Usage metadata (user_id, model_id, input_tokens, output_tokens, timestamps) is preserved. This changes the behavior of delete_chat_by_id, delete_chat_by_id_and_user_id, delete_chats_by_user_id, and delete_chats_by_user_id_and_folder_id.

Additional Information

Contributor License Agreement

🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.

## 📋 Pull Request Information **Original PR:** https://github.com/open-webui/open-webui/pull/23600 **Author:** [@smorello87](https://github.com/smorello87) **Created:** 4/11/2026 **Status:** ❌ Closed **Base:** `dev` ← **Head:** `feat/usage-limits` --- ### 📝 Commits (1) - [`3e4197b`](https://github.com/open-webui/open-webui/commit/3e4197b15484425e777ce106440799d80113f9a4) feat: per-group token usage limits with enforcement ### 📊 Changes **13 files changed** (+760 additions, -12 deletions) <details> <summary>View changed files</summary> 📝 `backend/open_webui/main.py` (+2 -0) ➕ `backend/open_webui/migrations/versions/c3d4e5f6a7b8_add_usage_ledger_table.py` (+102 -0) 📝 `backend/open_webui/models/chat_messages.py` (+1 -1) 📝 `backend/open_webui/models/chats.py` (+22 -8) 📝 `backend/open_webui/models/groups.py` (+20 -0) ➕ `backend/open_webui/models/usage_ledger.py` (+144 -0) ➕ `backend/open_webui/routers/usage_limits.py` (+151 -0) 📝 `backend/open_webui/utils/chat.py` (+17 -1) 📝 `backend/open_webui/utils/middleware.py` (+60 -0) ➕ `backend/open_webui/utils/usage_limits.py` (+153 -0) 📝 `src/lib/components/admin/Users/Groups/EditGroupModal.svelte` (+2 -1) 📝 `src/lib/components/admin/Users/Groups/Permissions.svelte` (+81 -0) 📝 `src/lib/components/chat/Chat.svelte` (+5 -1) </details> ### 📄 Description <!-- ⚠️ CRITICAL CHECKS FOR CONTRIBUTORS (READ, DON'T DELETE) ⚠️ 1. Target the `dev` branch. PRs targeting `main` will be automatically closed. 2. Do NOT delete the CLA section at the bottom. It is required for the bot to accept your PR. --> - [x] **Target branch:** `dev` - [x] **Description:** See below - [x] **Testing:** Extensively tested on ECS Fargate deployment with OpenRouter and AWS Bedrock - [x] **Code review:** Self-reviewed - [x] **Title Prefix:** `feat:` # Changelog Entry ### Added - Per-group token usage limits with configurable daily/monthly enforcement - Admin UI for setting token limits, soft limits, and priority per group - Soft limit warning toast and hard limit 429 block - Usage tracking for direct API calls and temporary chats - Usage data preserved when chats are deleted (content stripped, usage metadata retained) --- ### Description Adds per-group token usage limits that enforce daily or monthly caps on non-admin users. Full design discussion in #23558. **How it works:** - **Enforcement**: Pre-request check in `generate_chat_completion` queries `chat_message` for the user's token usage in the current period. If over the group's limit, returns 429. Background tasks (title/tag/follow-up generation) are skipped via `metadata.task` check. - **Usage data preservation**: Chat deletion strips message content but preserves usage metadata (user_id, model_id, tokens, timestamps). Prevents users from resetting usage by deleting chats. - **Temp chats & API calls**: Since `chat_message` has a FK to `chat` (which doesn't exist for these), usage is written to a small `usage_ledger` table. Enforcement sums both tables. - **Configuration**: Stored in the existing group `permissions` JSON field. Admin UI added to the Permissions tab in group settings. - **Soft/hard limits**: Configurable soft limit shows a warning toast; hard limit blocks with 429. - **Priority**: When a user belongs to multiple groups, the limit from the highest-priority group (lowest priority number) applies. ### Testing Tested on an ECS Fargate deployment (CUNY AI Lab, PostgreSQL 16.3) with: - OpenRouter (Gemma 4 31B) and AWS Bedrock (Nova Pro, Nova Premier) - Token counts verified to match analytics dashboard - Background tasks (title/tag/follow-up gen) excluded from counting and enforcement — verified via CloudWatch logs - Chat deletion confirmed to preserve usage counts (FK changed to SET NULL, verified via `pg_constraint` query) - Soft limit warning toasts and hard limit 429 blocks working - Admin users bypass all limits - Direct API calls (`POST /api/chat/completions` with JWT, no `chat_id`, `stream: false`) — both enforced and tracked - Alembic migration tested on PostgreSQL (uses raw SQL for FK alteration, batch mode for SQLite) - Automated code review via pr-validator-bot — both findings addressed and retested ### Testing Tested on an ECS Fargate deployment (CUNY AI Lab, PostgreSQL 16.3) with: - OpenRouter (Gemma 4 31B) and AWS Bedrock (Nova Pro, Nova Premier) - Token counts verified to match analytics dashboard - Background tasks (title/tag/follow-up gen) excluded from counting and enforcement — verified via CloudWatch logs - Chat deletion confirmed to preserve usage counts (FK changed to SET NULL, verified via `pg_constraint` query) - Soft limit warning toasts and hard limit 429 blocks working - Admin users bypass all limits - Direct API calls (`POST /api/chat/completions` with JWT, no `chat_id`, `stream: false`) — both enforced and tracked - Alembic migration tested on PostgreSQL (uses raw SQL for FK alteration, batch mode for SQLite) - Automated code review via pr-validator-bot — both findings addressed and retested ### Testing Tested on an ECS Fargate deployment (CUNY AI Lab, PostgreSQL 16.3) with: - OpenRouter (Gemma 4 31B) and AWS Bedrock (Nova Pro, Nova Premier) - Token counts verified to match analytics dashboard - Background tasks (title/tag/follow-up gen) excluded from counting and enforcement — verified via CloudWatch logs - Chat deletion confirmed to preserve usage counts (FK changed to SET NULL, verified via `pg_constraint` query) - Soft limit warning toasts and hard limit 429 blocks working - Admin users bypass all limits - Direct API calls (`POST /api/chat/completions` with JWT, no `chat_id`, `stream: false`) — both enforced and tracked - Alembic migration tested on PostgreSQL (uses raw SQL for FK alteration, batch mode for SQLite) - Automated code review via pr-validator-bot — both findings addressed and retested ### Testing Tested on an ECS Fargate deployment (CUNY AI Lab, PostgreSQL 16.3) with: - OpenRouter (Gemma 4 31B) and AWS Bedrock (Nova Pro, Nova Premier) - Token counts verified to match analytics dashboard - Background tasks (title/tag/follow-up gen) excluded from counting and enforcement — verified via CloudWatch logs - Chat deletion confirmed to preserve usage counts (FK changed to SET NULL, verified via `pg_constraint` query) - Soft limit warning toasts and hard limit 429 blocks working - Admin users bypass all limits - Direct API calls (`POST /api/chat/completions` with JWT, no `chat_id`, `stream: false`) — both enforced and tracked - Alembic migration tested on PostgreSQL (uses raw SQL for FK alteration, batch mode for SQLite) - Automated code review via pr-validator-bot — both findings addressed and retested ### Testing Tested on an ECS Fargate deployment (CUNY AI Lab, PostgreSQL 16.3) with: - OpenRouter (Gemma 4 31B) and AWS Bedrock (Nova Pro, Nova Premier) - Token counts verified to match analytics dashboard - Background tasks (title/tag/follow-up gen) excluded from counting and enforcement — verified via CloudWatch logs - Chat deletion confirmed to preserve usage counts (FK changed to SET NULL, verified via `pg_constraint` query) - Soft limit warning toasts and hard limit 429 blocks working - Admin users bypass all limits - Direct API calls (`POST /api/chat/completions` with JWT, no `chat_id`, `stream: false`) — both enforced and tracked - Alembic migration tested on PostgreSQL (uses raw SQL for FK alteration, batch mode for SQLite) - Automated code review via pr-validator-bot — both findings addressed and retested ### Testing Tested on an ECS Fargate deployment (CUNY AI Lab, PostgreSQL 16.3) with: - OpenRouter (Gemma 4 31B) and AWS Bedrock (Nova Pro, Nova Premier) - Token counts verified to match analytics dashboard - Background tasks (title/tag/follow-up gen) excluded from counting and enforcement — verified via CloudWatch logs - Chat deletion confirmed to preserve usage counts (FK changed to SET NULL, verified via `pg_constraint` query) - Soft limit warning toasts and hard limit 429 blocks working - Admin users bypass all limits - Direct API calls (`POST /api/chat/completions` with JWT, no `chat_id`, `stream: false`) — both enforced and tracked - Alembic migration tested on PostgreSQL (uses raw SQL for FK alteration, batch mode for SQLite) - Automated code review via pr-validator-bot — both findings addressed and retested ### Testing Tested on an ECS Fargate deployment (CUNY AI Lab, PostgreSQL 16.3) with: - OpenRouter (Gemma 4 31B) and AWS Bedrock (Nova Pro, Nova Premier) - Token counts verified to match analytics dashboard - Background tasks (title/tag/follow-up gen) excluded from counting and enforcement — verified via CloudWatch logs - Chat deletion confirmed to preserve usage counts (FK changed to SET NULL, verified via `pg_constraint` query) - Soft limit warning toasts and hard limit 429 blocks working - Admin users bypass all limits - Direct API calls (`POST /api/chat/completions` with JWT, no `chat_id`, `stream: false`) — both enforced and tracked - Alembic migration tested on PostgreSQL (uses raw SQL for FK alteration, batch mode for SQLite) - Automated code review via pr-validator-bot — both findings addressed and retested ### Testing Tested on an ECS Fargate deployment (CUNY AI Lab, PostgreSQL 16.3) with: - OpenRouter (Gemma 4 31B) and AWS Bedrock (Nova Pro, Nova Premier) - Token counts verified to match analytics dashboard - Background tasks (title/tag/follow-up gen) excluded from counting and enforcement — verified via CloudWatch logs - Chat deletion confirmed to preserve usage counts (FK changed to SET NULL, verified via `pg_constraint` query) - Soft limit warning toasts and hard limit 429 blocks working - Admin users bypass all limits - Direct API calls (`POST /api/chat/completions` with JWT, no `chat_id`, `stream: false`) — both enforced and tracked - Alembic migration tested on PostgreSQL (uses raw SQL for FK alteration, batch mode for SQLite) - Automated code review via pr-validator-bot — both findings addressed and retested ### Testing Tested on an ECS Fargate deployment (CUNY AI Lab, PostgreSQL 16.3) with: - OpenRouter (Gemma 4 31B) and AWS Bedrock (Nova Pro, Nova Premier) - Token counts verified to match analytics dashboard - Background tasks (title/tag/follow-up gen) excluded from counting and enforcement — verified via CloudWatch logs - Chat deletion confirmed to preserve usage counts (FK changed to SET NULL, verified via `pg_constraint` query) - Soft limit warning toasts and hard limit 429 blocks working - Admin users bypass all limits - Direct API calls (`POST /api/chat/completions` with JWT, no `chat_id`, `stream: false`) — both enforced and tracked - Alembic migration tested on PostgreSQL (uses raw SQL for FK alteration, batch mode for SQLite) - Automated code review via pr-validator-bot — both findings addressed and retested ### Testing Tested on an ECS Fargate deployment (CUNY AI Lab, PostgreSQL 16.3) with: - OpenRouter (Gemma 4 31B) and AWS Bedrock (Nova Pro, Nova Premier) - Token counts verified to match analytics dashboard - Background tasks (title/tag/follow-up gen) excluded from counting and enforcement — verified via CloudWatch logs - Chat deletion confirmed to preserve usage counts (FK changed to SET NULL, verified via `pg_constraint` query) - Soft limit warning toasts and hard limit 429 blocks working - Admin users bypass all limits - Direct API calls (`POST /api/chat/completions` with JWT, no `chat_id`, `stream: false`) — both enforced and tracked - Alembic migration tested on PostgreSQL (uses raw SQL for FK alteration, batch mode for SQLite) - Automated code review via pr-validator-bot — both findings addressed and retested ### Testing Tested on an ECS Fargate deployment (CUNY AI Lab, PostgreSQL 16.3) with: - OpenRouter (Gemma 4 31B) and AWS Bedrock (Nova Pro, Nova Premier) - Token counts verified to match analytics dashboard - Background tasks (title/tag/follow-up gen) excluded from counting and enforcement — verified via CloudWatch logs - Chat deletion confirmed to preserve usage counts (FK changed to SET NULL, verified via `pg_constraint` query) - Soft limit warning toasts and hard limit 429 blocks working - Admin users bypass all limits - Direct API calls (`POST /api/chat/completions` with JWT, no `chat_id`, `stream: false`) — both enforced and tracked - Alembic migration tested on PostgreSQL (uses raw SQL for FK alteration, batch mode for SQLite) - Automated code review via pr-validator-bot — both findings addressed and retested ### Breaking Changes - Chat message content is now stripped (set to NULL) on deletion instead of row deletion. Usage metadata (user_id, model_id, input_tokens, output_tokens, timestamps) is preserved. This changes the behavior of `delete_chat_by_id`, `delete_chat_by_id_and_user_id`, `delete_chats_by_user_id`, and `delete_chats_by_user_id_and_folder_id`. ### Additional Information - Discussion: #23558 - Related: #21675 (analytics enhancements), #23323 (original feature request) - New Alembic migration adds `usage_ledger` table (revision `c3d4e5f6a7b8`, depends on `b7c8d9e0f1a2`) ### Contributor License Agreement - [x] By submitting this pull request, I confirm that I have read and fully agree to the [Contributor License Agreement (CLA)](https://github.com/open-webui/open-webui/blob/main/CONTRIBUTOR_LICENSE_AGREEMENT), and I am providing my contributions under its terms. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>
GiteaMirror added the pull-request label 2026-04-25 14:40:22 -05:00
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
bug
confirmed
confirmed issue
core
documentation
enhancement
good first issue
help wanted
non-core
pull-request
Mirrored from GitHub Pull Request
 python
question
testing wanted
No Label pull-request
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/open-webui#42902
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?