github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-08 00:51:34 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #12939] ollama rocm > v0.12.6-rc0 (docker; 18d) crashes gfx1030 GPU and entire Linux system! #55091

New Issue
Open
opened 2026-04-29 08:18:52 -05:00 by GiteaMirror · 5 comments
GiteaMirror commented 2026-04-29 08:18:52 -05:00
Owner
Copy Link

Originally created by @ksaadDE on GitHub (Nov 4, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/12939

What is the issue?

Hello,

I worked all day through the ollama rocm docker image versions (after wondering why after a docker pull my stack aint working anymore!) and I could locate the issue beginning after 0.12.5 and 0.12.6-rc0

Thus,
ollama rocm > v0.12.6-rc0 (docker; 18d) crashes gfx1030 GPU and entire Linux system!

Title describes it. No logging shows. Furthermore, after a crash a full reboot is needed, no killall, kill -9 or restarting the docker service helps!

I will provide kernel version and other details later. My pc crashed one again, so I used another device to quickly file the bug report.

Relevant log output

none

OS

Linux

GPU

AMD

CPU

AMD

Ollama version

v0.12.6-rc1, v0.12.6-rc0

Originally created by @ksaadDE on GitHub (Nov 4, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/12939 ### What is the issue? Hello, I worked all day through the ollama rocm docker image versions (after wondering why after a docker pull my stack aint working anymore!) and I could locate the issue beginning after 0.12.5 and 0.12.6-rc0 Thus, ollama rocm > v0.12.6-rc0 (docker; 18d) crashes gfx1030 GPU and entire Linux system! Title describes it. No logging shows. Furthermore, after a crash a full reboot is needed, no killall, kill -9 or restarting the docker service helps! I will provide kernel version and other details later. My pc crashed one again, so I used another device to quickly file the bug report. ### Relevant log output ```shell none ``` ### OS Linux ### GPU AMD ### CPU AMD ### Ollama version v0.12.6-rc1, v0.12.6-rc0
GiteaMirror added the amdbuglinux labels 2026-04-29 08:18:53 -05:00
GiteaMirror commented 2026-04-29 08:18:54 -05:00
Author
Owner
Copy Link

@ksaadDE commented on GitHub (Nov 4, 2025):

General Sys Info

  • uname -r
    • 6.17.6-zen1-1-zen on Arch alike (host)
  • docker --version
    • Docker version 28.5.1, build e180ab8ab8
  • docker exec -it llm-ollama-1 /usr/bin/ollama --version
    • ollama version is 0.12.6-rc0 (working! until random crash)

Random Crash info:

Occuring after some time the LLMs are loaded and everything was working fine!

level=INFO source=runner.go:548 msg="failure during GPU discovery" OLLAMA_LIBRARY_PATH="[/usr/lib/ollama /usr/lib/ollama/rocm]" extra_envs=[] error="failed to finish discovery before timeout"
<!-- gh-comment-id:3485214643 --> @ksaadDE commented on GitHub (Nov 4, 2025): ## General Sys Info - **uname -r** - `6.17.6-zen1-1-zen on Arch alike (host)` - **docker --version** - `Docker version 28.5.1, build e180ab8ab8` - **docker exec -it llm-ollama-1 /usr/bin/ollama --version** - `ollama version is 0.12.6-rc0` (working! until random crash) ## Random Crash info: Occuring after some time the LLMs are loaded and everything was working fine! ``` level=INFO source=runner.go:548 msg="failure during GPU discovery" OLLAMA_LIBRARY_PATH="[/usr/lib/ollama /usr/lib/ollama/rocm]" extra_envs=[] error="failed to finish discovery before timeout" ```
GiteaMirror commented 2026-04-29 08:18:55 -05:00
Author
Owner
Copy Link

@ksaadDE commented on GitHub (Nov 4, 2025):

0.12.6-rc0 has for previously working models a very long loading time, resulting in:

level=INFO source=server.go:1306 msg="waiting for server to become available" status="llm server not responding"

and "load failed",

as well as error="timed out waiting for llama runner to start: context canceled"

<!-- gh-comment-id:3485609095 --> @ksaadDE commented on GitHub (Nov 4, 2025): 0.12.6-rc0 has for previously working models a very long loading time, resulting in: ``` level=INFO source=server.go:1306 msg="waiting for server to become available" status="llm server not responding" ``` and "load failed", as well as `error="timed out waiting for llama runner to start: context canceled"`
GiteaMirror commented 2026-04-29 08:18:55 -05:00
Author
Owner
Copy Link

@ganakee commented on GitHub (Nov 6, 2025):

I too experience this and have since 0.12.6.
Linux Ubuntu 25.10
AMD GPU 6650
gfx1030

The system fails entirely to use the AMD GPU. Ollama worked fine until 0.12.6.

A temp workaround is reverting (for me) to 0.12.3. I tested through 0.12.10-rc1

This may be related (and I mentioned before) #12754 .

<!-- gh-comment-id:3498043049 --> @ganakee commented on GitHub (Nov 6, 2025): I too experience this and have since 0.12.6. Linux Ubuntu 25.10 AMD GPU 6650 gfx1030 The system fails entirely to use the AMD GPU. Ollama worked fine until 0.12.6. A temp workaround is reverting (for me) to 0.12.3. I tested through 0.12.10-rc1 This may be related (and I mentioned before)[ #12754 ](https://github.com/ollama/ollama/issues/12754).
GiteaMirror commented 2026-04-29 08:18:55 -05:00
Author
Owner
Copy Link

@dhiltgen commented on GitHub (Nov 14, 2025):

If the entire Linux system is crashing, this sound like it's more likely to be a driver bug or possibly hardware fault. You can try downgrading (or upgrading) your amdgpu driver to see if that helps.

<!-- gh-comment-id:3530357478 --> @dhiltgen commented on GitHub (Nov 14, 2025): If the entire Linux system is crashing, this sound like it's more likely to be a driver bug or possibly hardware fault. You can try downgrading (or upgrading) your amdgpu driver to see if that helps.
GiteaMirror commented 2026-04-29 08:18:56 -05:00
Author
Owner
Copy Link

@ksaadDE commented on GitHub (Nov 16, 2025):

If the entire Linux system is crashing, this sound like it's more likely to be a driver bug or possibly hardware fault. You can try downgrading (or upgrading) your amdgpu driver to see if that helps.

I tried that, its likely the kernel version. However, since last driver update it crashses "less". But I also kept the older version to avoid the issue. It only crashed when ollama was run within a docker stack.

I will try in a few weeks or months again, perhaps all that changes. For now the older version works.

<!-- gh-comment-id:3537216459 --> @ksaadDE commented on GitHub (Nov 16, 2025): > If the entire Linux system is crashing, this sound like it's more likely to be a driver bug or possibly hardware fault. You can try downgrading (or upgrading) your amdgpu driver to see if that helps. I tried that, its likely the kernel version. However, since last driver update it crashses "less". But I also kept the older version to avoid the issue. It only crashed when ollama was run within a docker stack. I will try in a few weeks or months again, perhaps all that changes. For now the older version works.
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 amd bug linux
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#55091
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?