github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-06 08:02:14 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #11974] 🤔❓ 0.11.5 does not appear to respect OLLAMA_MAX_LOADED_MODELS like 0.11.4 did. #7950

New Issue
Closed
opened 2026-04-12 20:07:13 -05:00 by GiteaMirror · 5 comments
GiteaMirror commented 2026-04-12 20:07:13 -05:00
Owner
Copy Link

Originally created by @FieldMouse-AI on GitHub (Aug 20, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/11974

What is the issue?

I have OLLAMA_MAX_LOADED_MODELS=8 set in my Docker environment. I also set the timeout for all of my models to 30m.

Under v0.11.4, I easily host all of the models that I wanted to remain in memory without issue.

But, under v0;11.5, ollama server is behaving as if OLLAMA_MAX_LOADED_MODELS=1, even though it is still set to 8.

The result of this is masively slow performance due to models being forced to relead from disk storage for every request.

It would seem that this is a bit of a bug.

If you have any questions or comments, please feel free to ask. 🤗

Relevant log output

OS

Linux, Docker

GPU

No response

CPU

AMD

Ollama version

v0.11.5

Originally created by @FieldMouse-AI on GitHub (Aug 20, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/11974 ### What is the issue? I have `OLLAMA_MAX_LOADED_MODELS=8` set in my Docker environment. I also set the timeout for all of my models to `30m`. Under v0.11.4, I easily host all of the models that I wanted to remain in memory without issue. But, under v0;11.5, `ollama server` is behaving as if `OLLAMA_MAX_LOADED_MODELS=1`, even though it is still set to `8`. The result of this is masively slow performance due to models being forced to relead from disk storage for every request. It would seem that this is a bit of a bug. If you have any questions or comments, please feel free to ask. 🤗 ### Relevant log output ```shell ``` ### OS Linux, Docker ### GPU _No response_ ### CPU AMD ### Ollama version v0.11.5
GiteaMirror added the bug label 2026-04-12 20:07:13 -05:00
GiteaMirror commented 2026-04-12 20:07:15 -05:00
Author
Owner
Copy Link

@jessegross commented on GitHub (Aug 20, 2025):

Can you please post the server logs? Ideally from both 0.11.5 and 0.11.4.

<!-- gh-comment-id:3204006234 --> @jessegross commented on GitHub (Aug 20, 2025): Can you please post the [server logs](https://github.com/ollama/ollama/blob/main/docs/troubleshooting.md#how-to-troubleshoot-issues)? Ideally from both 0.11.5 and 0.11.4.
GiteaMirror commented 2026-04-12 20:07:15 -05:00
Author
Owner
Copy Link

@FieldMouse-AI commented on GitHub (Aug 20, 2025):

Can you please post the server logs? Ideally from both 0.11.5 and 0.11.4.

I've attached the server logs.

Somnething else you may notice as well: 0.11.4 is just hands down much faster than 0.11.5 -- even for cold start time.

In both logs I ran the exact same request 3 times after a total restart of the server to give everytihng a fresh start. 😊

I hope these logs are helpful. 😊

ollama-0.11.4.log
ollama-0.11.5.log

<!-- gh-comment-id:3204164811 --> @FieldMouse-AI commented on GitHub (Aug 20, 2025): > Can you please post the [server logs](https://github.com/ollama/ollama/blob/main/docs/troubleshooting.md#how-to-troubleshoot-issues)? Ideally from both 0.11.5 and 0.11.4. I've attached the server logs. Somnething else you may notice as well: **0.11.4** is just hands down **much faster than 0.11.5** -- even for cold start time. In both logs I ran the exact same request 3 times after a total restart of the server to give everytihng a fresh start. 😊 I hope these logs are helpful. 😊 [ollama-0.11.4.log](https://github.com/user-attachments/files/21881167/ollama-0.11.4.log) [ollama-0.11.5.log](https://github.com/user-attachments/files/21881168/ollama-0.11.5.log)
GiteaMirror commented 2026-04-12 20:07:17 -05:00
Author
Owner
Copy Link

@nfsecurity commented on GitHub (Aug 20, 2025):

I had to go back from 0.11.5 to 0.11.4 because the environment variables that I configured (via systemctl edit ollama.service) didn't work in the new version 0.11.5. Basically, I have 3 variables:

Environment="OLLAMA_KEEP_ALIVE=-1"
Environment="OLLAMA_FLASH_ATTENTION=1"
Environment="OLLAMA_NUM_PARALLEL=4"

But none of them were applied after my upgrade:

# curl https://ollama.ai/install.sh | sh
# systemctl edit ollama.service
# systemctl daemon-reload
# systemctl restart ollama
<!-- gh-comment-id:3207954642 --> @nfsecurity commented on GitHub (Aug 20, 2025): I had to go back from 0.11.5 to 0.11.4 because the environment variables that I configured (via systemctl edit ollama.service) didn't work in the new version 0.11.5. Basically, I have 3 variables: ``` Environment="OLLAMA_KEEP_ALIVE=-1" Environment="OLLAMA_FLASH_ATTENTION=1" Environment="OLLAMA_NUM_PARALLEL=4" ``` But none of them were applied after my upgrade: ``` # curl https://ollama.ai/install.sh | sh # systemctl edit ollama.service # systemctl daemon-reload # systemctl restart ollama ```
GiteaMirror commented 2026-04-12 20:07:18 -05:00
Author
Owner
Copy Link

@jessegross commented on GitHub (Aug 20, 2025):

@nfsecurity If none of those took effect, it's probably not related. Please file a new bug and post logs.

<!-- gh-comment-id:3207965529 --> @jessegross commented on GitHub (Aug 20, 2025): @nfsecurity If none of those took effect, it's probably not related. Please file a new bug and post logs.
GiteaMirror commented 2026-04-12 20:07:18 -05:00
Author
Owner
Copy Link

@Adearizqin commented on GitHub (Aug 20, 2025):

AA

On Thu, 21 Aug 2025, 04:32 ADE, @.***> wrote:

On Thu, 21 Aug 2025, 04:18 Jesse Gross, @.***> wrote:

jessegross left a comment (ollama/ollama#11974)
https://github.com/ollama/ollama/issues/11974#issuecomment-3207965529

@nfsecurity https://github.com/nfsecurity If none of those took
effect, it's probably not related. Please file a new bug and post logs.

—
Reply to this email directly, view it on GitHub
https://github.com/ollama/ollama/issues/11974#issuecomment-3207965529,
or unsubscribe
https://github.com/notifications/unsubscribe-auth/BA6WBHYXSL4NHR4VK4LNEJD3OTJY3AVCNFSM6AAAAACEKBIFPKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTEMBXHE3DKNJSHE
.
You are receiving this because you are subscribed to this thread.Message
ID: @.***>

<!-- gh-comment-id:3208006559 --> @Adearizqin commented on GitHub (Aug 20, 2025): AA On Thu, 21 Aug 2025, 04:32 ADE, ***@***.***> wrote: > > On Thu, 21 Aug 2025, 04:18 Jesse Gross, ***@***.***> wrote: > >> *jessegross* left a comment (ollama/ollama#11974) >> <https://github.com/ollama/ollama/issues/11974#issuecomment-3207965529> >> >> @nfsecurity <https://github.com/nfsecurity> If none of those took >> effect, it's probably not related. Please file a new bug and post logs. >> >> — >> Reply to this email directly, view it on GitHub >> <https://github.com/ollama/ollama/issues/11974#issuecomment-3207965529>, >> or unsubscribe >> <https://github.com/notifications/unsubscribe-auth/BA6WBHYXSL4NHR4VK4LNEJD3OTJY3AVCNFSM6AAAAACEKBIFPKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTEMBXHE3DKNJSHE> >> . >> You are receiving this because you are subscribed to this thread.Message >> ID: ***@***.***> >> >
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label bug
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/ollama#7950
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?