docker pull jamesdolezal/slideflow:latest-torchdocker run -it --shm-size=2g --gpus all jamesdolezal/slideflow:latest-torch
Output:
latest-torch: Pulling from jamesdolezal/slideflow
cf06a7c31611: Pull complete
41acec2bfcb9: Pull complete
f2531a2e2fb3: Pull complete
491f1d30a6d5: Pull complete
b47137a77b34: Pull complete
20f3d07a7d65: Pull complete
2861d6217cee: Pull complete
0ce2375d834b: Pull complete
c994d0811ab5: Pull complete
d40823bdf444: Pull complete
843f2129020c: Pull complete
3523444eb9cd: Pull complete
f832a00b3ac0: Pull complete
Digest: sha256:7ec6a62bee473387fa110e569b38e756f9be836090c50fc120bca636758e0b9c
Status: Downloaded newer image for jamesdolezal/slideflow:latest-torch
docker.io/jamesdolezal/slideflow:latest-torch
docker: 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.
ERRO[0000] error waiting for container:
Expected behavior
The container should start without errors.
Environment:
Docker version 24.0.5
Slideflow Version (e.g., 1.0): The container should do this.
OS (e.g., Ubuntu): Ubuntu 22.04 LTS
How you installed Slideflow (pip, source): The container should do this.
Python version: The container should do this.
CUDA/cuDNN version: 12.2
GPU models and configuration: NVIDIA GeForce RTX 4070 Ti
Any other relevant information: I also tried this on a windows machine with wsl2 and got:
docker: Error response from daemon: could not select device driver "" with capabilities: [gpu]
After running: docker run -it --shm-size=2g --gpus all jamesdolezal/slideflow:latest-torch
Additional context
I also tried this on a windows machine with wsl2 and got:
docker: Error response from daemon: could not select device driver "" with capabilities: [gpu]
After running: docker run -it --shm-size=2g --gpus all jamesdolezal/slideflow:latest-torch
Description
While following https://slideflow.dev/installation/#run-a-docker-container for the torch backend container I encountered a problem with starting the container.
To Reproduce
docker pull jamesdolezal/slideflow:latest-torchdocker run -it --shm-size=2g --gpus all jamesdolezal/slideflow:latest-torchOutput: latest-torch: Pulling from jamesdolezal/slideflow cf06a7c31611: Pull complete 41acec2bfcb9: Pull complete f2531a2e2fb3: Pull complete 491f1d30a6d5: Pull complete b47137a77b34: Pull complete 20f3d07a7d65: Pull complete 2861d6217cee: Pull complete 0ce2375d834b: Pull complete c994d0811ab5: Pull complete d40823bdf444: Pull complete 843f2129020c: Pull complete 3523444eb9cd: Pull complete f832a00b3ac0: Pull complete Digest: sha256:7ec6a62bee473387fa110e569b38e756f9be836090c50fc120bca636758e0b9c Status: Downloaded newer image for jamesdolezal/slideflow:latest-torch docker.io/jamesdolezal/slideflow:latest-torch docker: 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. ERRO[0000] error waiting for container:Expected behavior
The container should start without errors.
Environment:
pip, source): The container should do this.After running:
docker run -it --shm-size=2g --gpus all jamesdolezal/slideflow:latest-torchAdditional context
I also tried this on a windows machine with wsl2 and got: docker: Error response from daemon: could not select device driver "" with capabilities: [gpu]
After running:
docker run -it --shm-size=2g --gpus all jamesdolezal/slideflow:latest-torch