Repository navigation
nvidia-gpu-tests: optionally fail the play on a failed DCGM diagnostic - #1436
Merged
Merged
Conversation
Every test in the role runs with ignore_errors, so a failing GPU never fails the play. Add gpu_test_dcgm_fail_on_error (off by default); when set, a non-zero exit from dcgmi diag fails the play. Signed-off-by: Saqib Khan <11544614+saqibkh@users.noreply.github.com>
Honor string false/no values supplied through Ansible extra vars. Exercise the actual diagnostic task with a local stub across default, boolean and string settings, with successful and failing diagnostics. Co-Authored-By: Paperclip <noreply@paperclip.ing> Signed-off-by: Doug Holt <dholt@nvidia.com>
dholt
approved these changes
Oct 27, 2026
dholt
left a comment
dholt
left a comment
Contributor
There was a problem hiding this comment.
Reviewed the opt-in diagnostic failure handling and the follow-up boolean normalization. The default still ignores diagnostic failures; explicit true values fail the play on a non-zero diagnostic exit, while CLI string false/no values preserve the default behavior.
The 14-case local regression passes on Ansible 2.17 and 2.19. Independent verification of the final head passed, including explicit lint of the affected role. Role lint, playbook syntax checks, and all current-head GitHub checks pass. The original contributor commit is preserved.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
As suggested in #1434. Every test in
roles/nvidia-gpu-testsruns withignore_errors, so a failing GPU never fails the play. This addsgpu_test_dcgm_fail_on_error, off by default. When it is set, a non-zero exit fromdcgmi diagfails the play. The default behaviour doesn't change, and the other tests keepignore_errors. The README line for the DCGM step mentions the option.I tested it with ansible-core 2.21, running the role's DCGM task with a stub
dcgmion the PATH: with the option off a failing diagnostic is ignored and the play continues; with it on and a passing diagnostic the play continues; with it on and a failing one (the stub exits 226) the play fails.ansible-lint --offline roles/nvidia-gpu-testspasses.Signed off with
git commit -s.