Create utils function for nvidia-smi check #22
Conversation
Signed-off-by: laraPPr <[email protected]>
laraPPr
changed the title
|
A sync with |
|
I think you can have a couple of tests for this, there's mocking for |
Signed-off-by: laraPPr <[email protected]>
…ayer-scripts into utils_function_nvidia
| BUILD_STEP_ARGS+=("--nvidia" "all") | ||
| elif [ ${ec} -eq 1 ]; then | ||
| BUILD_STEP_ARGS+=("--nvidia" "install") | ||
| elif [ ${ec} -eq 2 ]; then | ||
| BUILD_STEP_ARGS+=("--nvidia" "install") |
There was a problem hiding this comment.
This is now exactly mimicking the behavior but I'm not sure this is correct.
This is now set in this case No 'nvidia-smi' found, no available GPU but allowing overriding this check.
There was a problem hiding this comment.
It is indeed now always set. I do not think this is what we want, right. exerpt of logs from EESSI/software-layer#1143
No 'nvidia-smi' found, no available GPU but allowing overriding this check
Executing command to build software:
/project/60006/SHARED/jobs/2025.08/pr_1107/event_cfecebd0-6eb2-11f0-9bd7-28ae4dcb0ae2/run_000/linux_x86_64_intel_sapphirerapids/eessi.io-2023.06-software/software-layer-scripts/eessi_container.sh
--verbose --access rw --mode run --container docker://ghcr.io/eessi/build-node:debian12
--repository eessi.io-2023.06-software --extra-bind-paths /project/60006/SHARED/jobs/2025.08/pr_1107/event_cfecebd0-6eb2-11f0-9bd7-28ae4dcb0ae2/run_000/linux_x86_64_intel_sapphirerapids/eessi.io-2023.06-software/software-layer-scripts,/dev
--pass-through --contain --save /project/60006/SHARED/jobs/2025.08/pr_1107/event_cfecebd0-6eb2-11f0-9bd7-28ae4dcb0ae2/run_000/linux_x86_64_intel_sapphirerapids/eessi.io-2023.06-software/previous_tmp/build_step
--storage /tmp/bot/EESSI/eessi_job.N27hQTIAbE --nvidia install
--host-injections /project/def-users/bot/shared/host-injections
There was a problem hiding this comment.
In our scripts this is fine, it is a once only thing, once things are installed they are not reinstalled.
Co-authored-by: ocaisa <[email protected]>
Signed-off-by: laraPPr <[email protected]>
Signed-off-by: laraPPr <[email protected]>
Signed-off-by: laraPPr <[email protected]>
There was a problem hiding this comment.
This looks good, a CI check would be great. You can do it in https://github.com/EESSI/software-layer-scripts/blob/main/.github/workflows/tests_link_nvidia_host_libraries.yml, check once before nvidia-smi is mocked that it returns non-zero, and once after it has been mocked
…ies.yml Signed-off-by: laraPPr <[email protected]>
|
Implemented the CI see https://github.com/EESSI/software-layer-scripts/actions/runs/16718197861/job/47316223515?pr=22 and I think it does what is expected? |
|
This looks good, can you sync it with |
No description provided.