[PR #13031] [MERGED] feat: Add support for configuring FastAPI/AnyIO Thread Pool #23078

New Issue

GiteaMirror · 2026-04-20T04:37:00-05:00

GiteaMirror commented

2026-04-20 04:37:00 -05:00

📋 Pull Request Information

Original PR: https://github.com/open-webui/open-webui/pull/13031
Author: @gaby
Created: 4/18/2025
Status: ✅ Merged
Merged: 4/19/2025
Merged by: @tjbck

Base: dev ← Head: anyio-workers

📝 Commits (4)

19b3d88 Add support for configuring FastAPI/AnyIO Thread Pool Size
e19554e Simplify import
f77f6cd Update imports to match anyio docs
62f87c2 Use default AnyIO pool size

📊 Changes

4 files changed (+21 additions, -5 deletions)

View changed files

📝 backend/open_webui/config.py (+3 -0)
📝 backend/open_webui/main.py (+8 -0)
📝 backend/open_webui/routers/knowledge.py (+2 -1)
📝 backend/open_webui/routers/users.py (+8 -4)

📄 Description

Description

FastAPI uses AnyIO which by default has a limit of 40 threads for the whole application. These threads are used by any non-async or threaded function. This PR adds support for configuring this value via ENV, ~~it also bumps the default value to 64 threads.~~
This should help the server from getting blocked when handling web-searches or any other non-async calls.

Implementation Source: https://github.com/Kludex/fastapi-tips#2-be-careful-with-non-async-functions
AnyIO Docs: https://www.starlette.io/threadpool/#concurrency-limitations

Related: https://github.com/open-webui/open-webui/pull/12958

_{🔄 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/13031 **Author:** [@gaby](https://github.com/gaby) **Created:** 4/18/2025 **Status:** ✅ Merged **Merged:** 4/19/2025 **Merged by:** [@tjbck](https://github.com/tjbck) **Base:** `dev` ← **Head:** `anyio-workers` --- ### 📝 Commits (4) - [`19b3d88`](https://github.com/open-webui/open-webui/commit/19b3d88737e49eb24a37f27fde52c6bf4efc9a59) Add support for configuring FastAPI/AnyIO Thread Pool Size - [`e19554e`](https://github.com/open-webui/open-webui/commit/e19554e9a8c9c2259527e44660eb9826557070be) Simplify import - [`f77f6cd`](https://github.com/open-webui/open-webui/commit/f77f6cd1e24b3fc1e6ba99c1d16bbdeb330e1cb3) Update imports to match anyio docs - [`62f87c2`](https://github.com/open-webui/open-webui/commit/62f87c2e38ebd5d1ee785bb10c76aaef6ffd4223) Use default AnyIO pool size ### 📊 Changes **4 files changed** (+21 additions, -5 deletions) <details> <summary>View changed files</summary> 📝 `backend/open_webui/config.py` (+3 -0) 📝 `backend/open_webui/main.py` (+8 -0) 📝 `backend/open_webui/routers/knowledge.py` (+2 -1) 📝 `backend/open_webui/routers/users.py` (+8 -4) </details> ### 📄 Description ### Description - FastAPI uses AnyIO which by default has a limit of `40` threads for the whole application. These threads are used by any non-async or threaded function. This PR adds support for configuring this value via `ENV`, ~it also bumps the default value to 64 threads.~ - This should help the server from getting blocked when handling web-searches or any other `non-async` calls. Implementation Source: https://github.com/Kludex/fastapi-tips#2-be-careful-with-non-async-functions AnyIO Docs: https://www.starlette.io/threadpool/#concurrency-limitations Related: https://github.com/open-webui/open-webui/pull/12958 --- <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 2026-04-20 04:37:00 -05:00

GiteaMirror closed this issue

2026-04-20 04:37:00 -05:00

GiteaMirror referenced this issue

2026-04-20 06:49:06 -05:00

[PR #23078] [CLOSED] fix(terminals): proxy admin terminal verification through backend #27012

GiteaMirror referenced this issue

2026-04-25 14:28:01 -05:00

[PR #23078] [CLOSED] fix(terminals): proxy admin terminal verification through backend #42642

GiteaMirror referenced this issue

2026-04-30 02:33:34 -05:00

[PR #23078] [CLOSED] fix(terminals): proxy admin terminal verification through backend #50060