Looking closer at this, there are two things here we may be able to address:
1. The cleanup script fails to un-mount a filesystem that is already not
mounted. We can probably easily fix that [1].
2. The reason the cleanup script was trying to cleanup stuff in the 1st place
was because a major mess was left around on the node by the vdsm
check_merged job that ran on it prior to this job [2]. The check_merged job
failed in such a way that made the cleanup script not run at all. I'm still
not sure what it the root cause of that, so we'll need to further
investigate.
I will make a quick fix for #1 so that #2 failures do not cascade into
other jobs until we can figure out why is it happening.
[1]:
https://ovirt-jira.atlassian.net/browse/OVIRT-937
[2]:
https://ovirt-jira.atlassian.net/browse/OVIRT-938
--
Barak Korren
bkorren(a)redhat.com
RHCE, RHCi, RHV-DevOps Team
https://ifireball.wordpress.com/