[GH-ISSUE #1643] [BUG]: Preloaded model gets reloaded #1071

Closed
opened 2026-02-22 18:22:59 -05:00 by yindo · 2 comments
Owner

Originally created by @PaoloC68 on GitHub (Jun 9, 2024).
Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/1643

How are you running AnythingLLM?

Docker (remote machine)

What happened?

I preload codestral, then I start or continue a chat and even if a model is loaded it gets reloaded with 5 min keepalive

Are there known steps to reproduce?

I try to run anythingLLM with codestral preloaded permanently on ollama:
ollama run codestral "" --keep-alive -1s

:~$ ollama ps
NAME                ID              SIZE     PROCESSOR    UNTIL
codestral:latest    fcc0019dcee9    13 GB    100% GPU     Forever

But when I start a chat with anythingLLM the model gets reloaded:

:~$ ollama ps
NAME                       ID              SIZE     PROCESSOR    UNTIL
codestral:22b-v0.1-q4_0    fcc0019dcee9    13 GB    100% GPU     4 minutes from now
Originally created by @PaoloC68 on GitHub (Jun 9, 2024). Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/1643 ### How are you running AnythingLLM? Docker (remote machine) ### What happened? I preload codestral, then I start or continue a chat and even if a model is loaded it gets reloaded with 5 min keepalive ### Are there known steps to reproduce? I try to run anythingLLM with codestral preloaded permanently on ollama: `ollama run codestral "" --keep-alive -1s` ``` :~$ ollama ps NAME ID SIZE PROCESSOR UNTIL codestral:latest fcc0019dcee9 13 GB 100% GPU Forever ``` But when I start a chat with anythingLLM the model gets reloaded: ``` :~$ ollama ps NAME ID SIZE PROCESSOR UNTIL codestral:22b-v0.1-q4_0 fcc0019dcee9 13 GB 100% GPU 4 minutes from now ```
yindo added the possible bug label 2026-02-22 18:22:59 -05:00
yindo closed this issue 2026-02-22 18:22:59 -05:00
Author
Owner

@PaoloC68 commented on GitHub (Jun 9, 2024):

I know that there was a similar bug before but the bug is still present:

https://github.com/Mintplex-Labs/anything-llm/issues/1011

@PaoloC68 commented on GitHub (Jun 9, 2024): I know that there was a similar bug before but the bug is still present: https://github.com/Mintplex-Labs/anything-llm/issues/1011
Author
Owner

@timothycarambat commented on GitHub (Jun 10, 2024):

Ah yes, being tracked in https://github.com/Mintplex-Labs/anything-llm/issues/1585

That issue was for mlock, but not for prevention of unloading. We need to close that chore issue and will set it to something higher since 5 min is quite low.

@timothycarambat commented on GitHub (Jun 10, 2024): Ah yes, being tracked in https://github.com/Mintplex-Labs/anything-llm/issues/1585 That issue was for `mlock`, but not for prevention of unloading. We need to close that chore issue and will set it to something higher since 5 min is quite low.
yindo changed title from [BUG]: Preloaded model gets reloaded to [GH-ISSUE #1643] [BUG]: Preloaded model gets reloaded 2026-06-05 14:38:49 -04:00
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: Mintplex-Labs/anything-llm#1071