github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-11-10 15:02:13 -06:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #13230] ollama serve causes ollama ps to always return empty results #34507

New Issue
Closed
opened 2026-04-22 18:07:51 -05:00 by GiteaMirror · 8 comments
GiteaMirror commented 2026-04-22 18:07:51 -05:00
Owner
Copy Link
Copy Source

Originally created by @anderscui on GitHub (Nov 24, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/13230

What is the issue?

When ollama serve is running, the command ollama ps always returns an empty list, regardless of whether OLLAMA_KEEP_ALIVE is set. After stopping ollama serve, ollama ps immediately shows the expected running model instances again.

Issue 1: Without serve → model appears; With serve → model disappears

Steps to reproduce

  1. Do NOT run ollama serve.

  2. Run a generation request:

curl http://localhost:11434/api/generate -d '{
  "model": "deepseek-r1:1.5b",
  "prompt": "hello world"
}'
  1. ollama ps shows:
NAME                ID              SIZE      PROCESSOR    CONTEXT    UNTIL
deepseek-r1:1.5b    e0979632db5a    2.8 GB    100% GPU     32768      4 minutes from now
  1. Then run:
export OLLAMA_KEEP_ALIVE=30m
ollama serve

  1. Run ollama ps again → returns empty result

  2. Stop ollama serve → run ollama ps → the record reappears

Issue 2: With serve running, requests produce no visible instances.

Steps to reproduce

Similar to Issue 1: first run ollama serve, then run the above curl command, ollama ps shows empty result. Stop ollama serve → run ollama ps → the record appears.

Relevant log output

OS

macOS

GPU

Intel

CPU

Apple

Ollama version

0.13.0

Originally created by @anderscui on GitHub (Nov 24, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/13230 ### What is the issue? When `ollama serve` is running, the command `ollama ps` always returns an empty list, regardless of whether `OLLAMA_KEEP_ALIVE` is set. After stopping `ollama serve`, `ollama ps` immediately shows the expected running model instances again. ### Issue 1: Without serve → model appears; With serve → model disappears **Steps to reproduce** 1. Do NOT run `ollama serve`. 2. Run a generation request: ```bash curl http://localhost:11434/api/generate -d '{ "model": "deepseek-r1:1.5b", "prompt": "hello world" }' ``` 3. `ollama ps` shows: ```bash NAME ID SIZE PROCESSOR CONTEXT UNTIL deepseek-r1:1.5b e0979632db5a 2.8 GB 100% GPU 32768 4 minutes from now ``` 4. Then run: ```bash export OLLAMA_KEEP_ALIVE=30m ollama serve ``` 5. Run ollama ps again → returns empty result 6. Stop ollama serve → run ollama ps → the record reappears ### Issue 2: With serve running, requests produce no visible instances. **Steps to reproduce** Similar to Issue 1: first run `ollama serve`, then run the above curl command, `ollama ps` shows empty result. Stop ollama serve → run ollama ps → the record appears. ### Relevant log output ```shell ``` ### OS macOS ### GPU Intel ### CPU Apple ### Ollama version 0.13.0
GiteaMirror added the bug label 2026-04-22 18:07:51 -05:00
GiteaMirror commented 2026-04-22 18:07:55 -05:00
Author
Owner
Copy Link
Copy Source

@rick-github commented on GitHub (Nov 24, 2025):

What's the output of ps ax | grep -i [o]llama at the different steps of your test?

<!-- gh-comment-id:3571937955 --> @rick-github commented on GitHub (Nov 24, 2025): What's the output of `ps ax | grep -i [o]llama` at the different steps of your test?
GiteaMirror commented 2026-04-22 18:07:57 -05:00
Author
Owner
Copy Link
Copy Source

@anderscui commented on GitHub (Nov 24, 2025):

@rick-github

Before ollama serve:

 1840   ??  S      0:04.48 /Applications/Ollama.app/Contents/MacOS/Ollama hidden
12681   ??  S      0:00.89 /Applications/Ollama.app/Contents/Resources/ollama runner --model ~/.ollama/models/blobs/sha256-aabd4debf0c8f08881923f2c25fc0fdeed24435271c2b3e92c4af36704040dbc --port 50593
92012   ??  S      0:03.27 /Applications/Ollama.app/Contents/Resources/ollama serve

After:

 1840   ??  S      0:04.49 /Applications/Ollama.app/Contents/MacOS/Ollama hidden
12681   ??  S      0:00.89 /Applications/Ollama.app/Contents/Resources/ollama runner --model ~/.ollama/models/blobs/sha256-aabd4debf0c8f08881923f2c25fc0fdeed24435271c2b3e92c4af36704040dbc --port 50593
92012   ??  S      0:03.27 /Applications/Ollama.app/Contents/Resources/ollama serve
13142 s004  S+     0:00.04 ollama serve
<!-- gh-comment-id:3571997218 --> @anderscui commented on GitHub (Nov 24, 2025): @rick-github Before `ollama serve`: ```bash 1840 ?? S 0:04.48 /Applications/Ollama.app/Contents/MacOS/Ollama hidden 12681 ?? S 0:00.89 /Applications/Ollama.app/Contents/Resources/ollama runner --model ~/.ollama/models/blobs/sha256-aabd4debf0c8f08881923f2c25fc0fdeed24435271c2b3e92c4af36704040dbc --port 50593 92012 ?? S 0:03.27 /Applications/Ollama.app/Contents/Resources/ollama serve ``` After: ```bash 1840 ?? S 0:04.49 /Applications/Ollama.app/Contents/MacOS/Ollama hidden 12681 ?? S 0:00.89 /Applications/Ollama.app/Contents/Resources/ollama runner --model ~/.ollama/models/blobs/sha256-aabd4debf0c8f08881923f2c25fc0fdeed24435271c2b3e92c4af36704040dbc --port 50593 92012 ?? S 0:03.27 /Applications/Ollama.app/Contents/Resources/ollama serve 13142 s004 S+ 0:00.04 ollama serve ```
GiteaMirror commented 2026-04-22 18:07:57 -05:00
Author
Owner
Copy Link
Copy Source

@rick-github commented on GitHub (Nov 24, 2025):

You are running multiple servers. When you start a new one, it becomes the one that the client connects to. Since it's not running a model, ollama ps returns no results. When you stop it, ollama ps is connecting to the previously started server which has a loaded model and so ollama ps shows that model running.

<!-- gh-comment-id:3572157834 --> @rick-github commented on GitHub (Nov 24, 2025): You are running multiple servers. When you start a new one, it becomes the one that the client connects to. Since it's not running a model, `ollama ps` returns no results. When you stop it, `ollama ps` is connecting to the previously started server which has a loaded model and so `ollama ps` shows that model running.
GiteaMirror commented 2026-04-22 18:07:58 -05:00
Author
Owner
Copy Link
Copy Source

@pdevine commented on GitHub (Nov 25, 2025):

I'll go ahead and close the issue as answered (ty @rick-github !).

<!-- gh-comment-id:3573405722 --> @pdevine commented on GitHub (Nov 25, 2025): I'll go ahead and close the issue as answered (ty @rick-github !).
GiteaMirror commented 2026-04-22 18:07:59 -05:00
Author
Owner
Copy Link
Copy Source

@anderscui commented on GitHub (Nov 25, 2025):

@rick-github So how to set a global OLLAMA_KEEP_ALIVE, and how to use ollama serve correctly? The first serve is started automatically.

<!-- gh-comment-id:3573427799 --> @anderscui commented on GitHub (Nov 25, 2025): @rick-github So how to set a global OLLAMA_KEEP_ALIVE, and how to use `ollama serve` correctly? The first `serve` is started automatically.
GiteaMirror commented 2026-04-22 18:08:00 -05:00
Author
Owner
Copy Link
Copy Source

@anderscui commented on GitHub (Nov 25, 2025):

@rick-github , maybe we can use the following methods:

Method 1:

stop all ollama related processes, including ollama GUI app, then

export OLLAMA_KEEP_ALIVE=60m
ollama ps

This will start the GUI app and a new /Applications/Ollama.app/Contents/Resources/ollama serve process, and its keep_alive value is also set, but ollama serve is not usable, as this issue says.

Method 2:

stop all ollama related processes, including ollama GUI app, then

export OLLAMA_KEEP_ALIVE=60m
ollama serve

This will start ollama serve(which is useful for debugging), and its keep_alive value is also set correctly, but the GUI app is not useable.

Conclusion

Now I understand the design, but it seems not very intuitive.

<!-- gh-comment-id:3575439003 --> @anderscui commented on GitHub (Nov 25, 2025): @rick-github , maybe we can use the following methods: ### Method 1: stop all ollama related processes, including ollama GUI app, then ```bash export OLLAMA_KEEP_ALIVE=60m ollama ps ``` This will start the GUI app and a new `/Applications/Ollama.app/Contents/Resources/ollama serve` process, and its keep_alive value is also set, but `ollama serve` is not usable, as this issue says. ### Method 2: stop all ollama related processes, including ollama GUI app, then ```bash export OLLAMA_KEEP_ALIVE=60m ollama serve ``` This will start `ollama serve`(which is useful for debugging), and its keep_alive value is also set correctly, but the GUI app is not useable. ### Conclusion Now I understand the design, but it seems not very intuitive.
GiteaMirror commented 2026-04-22 18:08:02 -05:00
Author
Owner
Copy Link
Copy Source

@rick-github commented on GitHub (Nov 25, 2025):

https://github.com/ollama/ollama/blob/main/docs/faq.mdx#setting-environment-variables-on-mac

<!-- gh-comment-id:3575449741 --> @rick-github commented on GitHub (Nov 25, 2025): https://github.com/ollama/ollama/blob/main/docs/faq.mdx#setting-environment-variables-on-mac
GiteaMirror commented 2026-04-22 18:08:03 -05:00
Author
Owner
Copy Link
Copy Source

@anderscui commented on GitHub (Nov 25, 2025):

@rick-github Thanks a lot.

<!-- gh-comment-id:3575468771 --> @anderscui commented on GitHub (Nov 25, 2025): @rick-github Thanks a lot.
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 labels bug
Milestone
No items
No Milestone
Projects
Clear projects
No projects
Assignees
Clear assignees
GiteaMirror ninjasurge (Kaehlen Rog)
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#34507