[PR #308] [MERGED] Improve nginx config to support better streaming response from models #395

Closed
opened 2026-02-15 17:16:43 -05:00 by yindo · 0 comments
Owner

📋 Pull Request Information

Original PR: https://github.com/open-webui/docs/pull/308
Author: @diwakar-s-maurya
Created: 11/30/2024
Status: Merged
Merged: 12/9/2024
Merged by: @tjbck

Base: mainHead: main


📝 Commits (1)

  • 6eb55d4 Improve nginx config to support better streaming response from models

📊 Changes

2 files changed (+6 additions, -0 deletions)

View changed files

📝 docs/tutorials/integrations/tab-nginx/LetsEncrypt.md (+3 -0)
📝 docs/tutorials/integrations/tab-nginx/SelfSigned.md (+3 -0)

📄 Description

By default, proxy buffer is turned on in nginx with size of 4KB or 8KB. https://nginx.org/en/docs/http/ngx_http_proxy_module.html#proxy_buffers

If proxy buffer is turned on, nginx will buffer data received from openwebui and then relay it to user's browser in chunks of 4KB/8KB. This means that when model is returning a streaming response, on browser it is not visible as fluid as it should be. The model response is displayed in chunks and UX is not as beautiful. This only happens if nginx is introduced as a reverse-proxy in between. It does not happen when openwebui is used directly.

This can be easily fixed by turning proxy buffering off. The response received from openwebui is relayed to user's browser immediately, removing the jankiness and restoring the fluid stream of text on browser.


🔄 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/docs/pull/308 **Author:** [@diwakar-s-maurya](https://github.com/diwakar-s-maurya) **Created:** 11/30/2024 **Status:** ✅ Merged **Merged:** 12/9/2024 **Merged by:** [@tjbck](https://github.com/tjbck) **Base:** `main` ← **Head:** `main` --- ### 📝 Commits (1) - [`6eb55d4`](https://github.com/open-webui/docs/commit/6eb55d426c344e9671d28942f81645c2696361c5) Improve nginx config to support better streaming response from models ### 📊 Changes **2 files changed** (+6 additions, -0 deletions) <details> <summary>View changed files</summary> 📝 `docs/tutorials/integrations/tab-nginx/LetsEncrypt.md` (+3 -0) 📝 `docs/tutorials/integrations/tab-nginx/SelfSigned.md` (+3 -0) </details> ### 📄 Description By default, proxy buffer is turned on in nginx with size of 4KB or 8KB. https://nginx.org/en/docs/http/ngx_http_proxy_module.html#proxy_buffers If proxy buffer is turned on, nginx will buffer data received from openwebui and then relay it to user's browser in chunks of 4KB/8KB. This means that when model is returning a streaming response, on browser it is not visible as fluid as it should be. The model response is displayed in chunks and UX is not as beautiful. This only happens if nginx is introduced as a reverse-proxy in between. It does not happen when openwebui is used directly. This can be easily fixed by turning proxy buffering off. The response received from openwebui is relayed to user's browser immediately, removing the jankiness and restoring the fluid stream of text on browser. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>
yindo added the pull-request label 2026-02-15 17:16:43 -05:00
yindo closed this issue 2026-02-15 17:16:43 -05:00
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: open-webui/docs#395