github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-07-16 05:05:37 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #7283] Ollama Fails to Start On Ubuntu Server OS (Headless) when using a GPU #66683

New Issue
Closed
opened 2026-05-04 07:47:41 -05:00 by GiteaMirror · 2 comments
GiteaMirror commented 2026-05-04 07:47:41 -05:00
Owner
Copy Link

Originally created by @F1zzyD on GitHub (Oct 21, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/7283

What is the issue?

Ollama fails to load using docker-compose on a headless Ubuntu server. I have installed, purged, reinstalled, purged, and re-reinstalled drivers, docker, docker-compose, etc. and nothing allows Ollama to boot. Here is my compose.yml:
services:

  ollama:
    container_name: ollama
    ports:
      - "11434:11434"
    volumes:
      - /home/admin/services/ollama:/root/.ollama
    restart: unless-stopped
    image: ollama/ollama
    deploy:
      resources:
        reservations:
          devices:
            - driver: ${OLLAMA_GPU_DRIVER-nvidia}
              count: all
              capabilities:
                - gpu

When using docker compose up -d I get this output:

Failed to deploy a stack: Network ollamatest_default Creating Network ollamatest_default Created Container ollama Creating Container ollama Created Container ollama Starting Error response from daemon: failed to create task for container: failed to create shim task: OCI runtime create failed: runc create failed: unable to start container process: error during container init: error running hook #0: error running hook: exit status 1, stdout: , stderr: Auto-detected mode as 'legacy' nvidia-container-cli: initialization error: load library failed: libnvidia-ml.so.1: cannot open shared object file: no such file or directory: unknown

I have used Ollama on the desktop version of Ubuntu with zero issues, but it seems that Ollama does not work on a headless version of Ubuntu, which is rather silly.

OS

Linux

GPU

Nvidia

CPU

AMD

Ollama version

latest

Originally created by @F1zzyD on GitHub (Oct 21, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/7283 ### What is the issue? Ollama fails to load using docker-compose on a headless Ubuntu server. I have installed, purged, reinstalled, purged, and re-reinstalled drivers, docker, docker-compose, etc. and nothing allows Ollama to boot. Here is my compose.yml: services: ``` ollama: container_name: ollama ports: - "11434:11434" volumes: - /home/admin/services/ollama:/root/.ollama restart: unless-stopped image: ollama/ollama deploy: resources: reservations: devices: - driver: ${OLLAMA_GPU_DRIVER-nvidia} count: all capabilities: - gpu ``` When using `docker compose up -d` I get this output: ``` Failed to deploy a stack: Network ollamatest_default Creating Network ollamatest_default Created Container ollama Creating Container ollama Created Container ollama Starting Error response from daemon: failed to create task for container: failed to create shim task: OCI runtime create failed: runc create failed: unable to start container process: error during container init: error running hook #0: error running hook: exit status 1, stdout: , stderr: Auto-detected mode as 'legacy' nvidia-container-cli: initialization error: load library failed: libnvidia-ml.so.1: cannot open shared object file: no such file or directory: unknown ``` I have used Ollama on the desktop version of Ubuntu with zero issues, but it seems that Ollama does not work on a headless version of Ubuntu, which is rather silly. ### OS Linux ### GPU Nvidia ### CPU AMD ### Ollama version latest
GiteaMirror added the bug label 2026-05-04 07:47:41 -05:00
GiteaMirror commented 2026-05-04 07:47:45 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Oct 21, 2024):

 nvidia-container-cli: initialization error: load library failed: libnvidia-ml.so.1: 
  cannot open shared object file: no such file or directory: unknown 

$ apt-file search libnvidia-ml.so.1
libnvidia-compute-390: /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1
libnvidia-compute-418-server: /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1
libnvidia-compute-450-server: /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1
libnvidia-compute-470: /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1
libnvidia-compute-470-server: /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1
libnvidia-compute-510: /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1
libnvidia-compute-510-server: /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1
libnvidia-compute-535: /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1
libnvidia-compute-535-server: /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1
libnvidia-compute-545: /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1
libnvidia-compute-550: /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1
libnvidia-compute-550-server: /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1

Headless Ubuntu probably doesn't install nvidia drivers by default because headless. Even though you are using the GPU just for inference, you need to install the drivers to allow ollama to communicate with the GPU.

<!-- gh-comment-id:2427402976 --> @rick-github commented on GitHub (Oct 21, 2024): ``` nvidia-container-cli: initialization error: load library failed: libnvidia-ml.so.1: cannot open shared object file: no such file or directory: unknown ``` ``` $ apt-file search libnvidia-ml.so.1 libnvidia-compute-390: /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1 libnvidia-compute-418-server: /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1 libnvidia-compute-450-server: /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1 libnvidia-compute-470: /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1 libnvidia-compute-470-server: /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1 libnvidia-compute-510: /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1 libnvidia-compute-510-server: /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1 libnvidia-compute-535: /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1 libnvidia-compute-535-server: /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1 libnvidia-compute-545: /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1 libnvidia-compute-550: /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1 libnvidia-compute-550-server: /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1 ``` Headless Ubuntu probably doesn't install nvidia drivers by default because headless. Even though you are using the GPU just for inference, you need to install the drivers to allow ollama to communicate with the GPU.
GiteaMirror commented 2026-05-04 07:47:46 -05:00
Author
Owner
Copy Link

@F1zzyD commented on GitHub (Nov 8, 2024):

I switched back to desktop Ubuntu as I was having too many issues running it in headless for my purposes. However, I wanted to give an update. When I posted this bug, I had already installed NVIDIA drivers. Running 'nvidia-smi' showed my GPU and related stats. Ollama still failed to find an initialize the GPU and refused to start.

Since I've moved back to desktop I'll just chop this up as Ollama not working on headless Ubuntu

<!-- gh-comment-id:2464783854 --> @F1zzyD commented on GitHub (Nov 8, 2024): I switched back to desktop Ubuntu as I was having too many issues running it in headless for my purposes. However, I wanted to give an update. When I posted this bug, I had already installed NVIDIA drivers. Running 'nvidia-smi' showed my GPU and related stats. Ollama still failed to find an initialize the GPU and refused to start. Since I've moved back to desktop I'll just chop this up as Ollama not working on headless Ubuntu
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#66683
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?