Fix partition parse bug for internal vLab - Githubissues

lava-nc / lava

A Software Framework for Neuromorphic Computing

https://lava-nc.org

Other

555 stars 143 forks source link

Fix partition parse bug for internal vLab #741

Closed tim-shea closed 1 year ago

tim-shea commented 1 year ago

Issue Number: #739

Objective of pull request: Fixing a bug from earlier merge due to invalid partition rows in sinfo.

Pull request checklist

Your PR fulfills the following requirements:

[X] Issue created that explains the change and why it's needed
[X] Tests are part of the PR (for bug fixes / features)
[X] Docs reviewed and added / updated if needed (for bug fixes / features)
[X] PR conforms to Coding Conventions
[X] PR applys BSD 3-clause or LGPL2.1+ Licenses to all code files
[X] Lint (flakeheaven lint src/lava tests/) and (bandit -r src/lava/.) pass locally
[X] Build tests (pytest) passes locally

Pull request type

[X] Bugfix

What is the current behavior?

lava.utils.slurm produces errors when run on internal vLab due to missing nodelist field in sinfo.

What is the new behavior?

No error will be produced, a default value (empty string) will be returned in PartitionInfo.

Does this introduce a breaking change?

[X] No

harryliu-intel commented 1 year ago

Not specifically about the new change... But a partition could appear more than once in sinfo if some nodes are up, some are done, e.g.:

og_gdc up 30:00 1 down* ncl-gdc-og-07 og_gdc up 30:00 3 idle ncl-gdc-og-[02,04,08]

It seems like both will be parsed by get_partitions, but get_partition_info returns only the first one. So if not all boards are down in a partition, the code might complain the partition is down when it's still possible to run.