[GH-ISSUE #10807] Performance degradation with external databases #16038

New Issue

GiteaMirror · 2026-04-19T22:05:04-05:00

GiteaMirror commented

2026-04-19 22:05:04 -05:00

Originally created by @TobiasGoerke on GitHub (Feb 26, 2025).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/10807

Description

When using external PostgreSQL instances (Azure Flexible Server/Google Cloud SQL), Open WebUI becomes severely sluggish – simple actions like opening a chat take 10-15 seconds.

The core issue stems from:

Query amplification: 3+ auth queries (get_user_by_id, update_user_last_active_by_id) + 2+ endpoint-specific queries per request
Latency multiplier: 150-250ms average database response time
Cumulative effect: A window reload triggers ~20 API calls → 60-100 database queries. This is about 15 seconds for me.
Note: Pooling has helped me to cut the response times in half. Still, the issue persists.

Example Breakdown

For get_chat_tags_by_id:

Authentication: 3 queries (200ms × 3 = 600ms)
Endpoint logic: 2+ queries (200ms × 2 = 400ms)
Total: ~1 second baseline before business logic

Key Questions

Are others running Open WebUI successfully with managed databases at this latency tier?
Can we:

Reduce auth query count (3→1?)
Reduce redundant user activity updates or at least update asynchronously
Parallelize queries in endpoint chains

What caching strategies work for session data without sacrificing security?

Query Log

Here's all queries launched for one page reload (61 queries):
queries.log

Thanks!

Originally created by @TobiasGoerke on GitHub (Feb 26, 2025). Original GitHub issue: https://github.com/open-webui/open-webui/issues/10807 ## Description When using external PostgreSQL instances (Azure Flexible Server/Google Cloud SQL), Open WebUI becomes severely sluggish – simple actions like opening a chat take 10-15 seconds. The core issue stems from: * Query amplification: 3+ auth queries (`get_user_by_id`, `update_user_last_active_by_id`) + 2+ endpoint-specific queries per request * Latency multiplier: 150-250ms average database response time * Cumulative effect: A window reload triggers ~20 API calls → 60-100 database queries. This is about 15 seconds for me. * Note: Pooling has helped me to cut the response times in half. Still, the issue persists. ## Example Breakdown For `get_chat_tags_by_id`: * Authentication: 3 queries (200ms × 3 = 600ms) * Endpoint logic: 2+ queries (200ms × 2 = 400ms) * Total: ~1 second baseline before business logic ## Key Questions 1. Are others running Open WebUI successfully with managed databases at this latency tier? 2. Can we: * Reduce auth query count (3→1?) * Reduce redundant user activity updates or at least update asynchronously * Parallelize queries in endpoint chains 3. What caching strategies work for session data without sacrificing security? ## Query Log Here's all queries launched for one page reload (61 queries): [queries.log](https://github.com/user-attachments/files/18981257/queries.log) Thanks!

GiteaMirror closed this issue

2026-04-19 22:05:04 -05:00

GiteaMirror referenced this issue

2026-04-20 05:11:05 -05:00

[PR #16038] [CLOSED] feat: support audio language selector for deepgram API, add fallback for both openai/deepgram #24012

GiteaMirror referenced this issue

2026-04-25 12:09:20 -05:00

[PR #16038] [CLOSED] feat: support audio language selector for deepgram API, add fallback for both openai/deepgram #39642

GiteaMirror referenced this issue

2026-04-29 22:06:46 -05:00

[PR #16038] [CLOSED] feat: support audio language selector for deepgram API, add fallback for both openai/deepgram #47060

GiteaMirror referenced this issue