CI runner randomly crashing in cta-tape-label

Just to track what happens on this annoying bit...

Our runners have been crashing on tape label part randomly for in the beginning of October.

Randomly as in the 2 runners, in various virtual libraries.

We are looking in various ways to improve MHVTL (operations#873) and CI workflows (operations##632 (closed) and #102 (closed))as a necessary step to better control the CI environment.

But this is still a mystery.

As we are facing some gitlab issues with VMs and the openstack team is cold migrating plenty of VMs to new flavors, these issues could be related to the underlying hypervisors.

I have opened this ticket to see if something happened recently to our runners from the openstack side INC3255212.

added Continuous Integration label

assigned to @jcamarer

added workflowIn Progress label

created branch 171-ci-runner-randomly-crashing-in-cta-tape-label to address this issue

Looks like me and the concerned SE can see the ticket...

Default ticket visibility change is biting us again. Please request ticket visibility if you want to see the content...

mentioned in merge request !65 (merged)

I have closed the openstack SNOW ticket.

The problem was created because the tests for the non native tape format was writing data into the tapes, but it wasn´t deleted. Because in the CI, the drives and tapes are outside of kubernetes, they are in the host machine. So these resources are shared. And when cta-tape-label was used on top of not empty tapes, then it was failing. So, the solution consist in to force it, label the tape even if it's not and empty tape. Of course the force command it's only for CI, in production we shouldn't use it.

closed with merge request !65 (merged)