rivosinc/prometheus-slurm-exporter

[cli-fallback] job failed to parse mem string 0

Closed this issue · 2 comments

Apparently mem can be a integer or at least 0. We should change the default behavior to Inc the error metric but also continue for line by line json parsing instead of returning no data at the first error

Error log:

time=2023-11-15T08:24:13.056-08:00 level=ERROR msg="job failed to parse with \"mem string 0 doesn't match regex ^(?P<num>([0-9]*[.])?[0-9]+)(?P<memunit>G|M|T)>

I second this error, newest slurm version 23.02

got the bug fixed, still should probably change the logic to line by line erroring, but this works for now.