Comment 4 for bug 1643911

Revision history for this message
Jordan Pittier (jordan-pittier) wrote :

I can confirm that this happens mainly (human check, based on analysing 6 gate jobs) with Tempest tests in test_images.py. And also this happens always during VM deletion (destroy).

What's special with these Tempest tests is that they try to delete the same delete server twice in 2 very short/successive API call (first time using the addCleanup() mechanism, the other attempts come from the class method clear_servers() in the BaseV2ComputeTest class. So this looks like a race condition in libvirtd.