[openstack-dev] [cinder][qa] Volume attachment intermittently failing
guilherme.moro at gmail.com
Thu Aug 4 12:12:17 UTC 2016
I'm running on Kilo (by customer demands) and I have a problem where I get
random failures while deploying a stack, the stack is a series of DB
instances, each with their own volume. I can reproduce it in pretty much
every run, usually in a different instance each time.
The problem is that the attached device doesn't show up during the boot
time, but is available; after a rescan of the PCI bus ('echo "1" >
/sys/bus/pci/rescan') the device is detected and functional.
I found some references to it, and it seems that there's a test in tempest
that triggers the bug:
Old e-mail thread discussing the test that triggered the bug:
The actual bug that is marked in the test as being the one blocking the
So you can get the background, the test is still disabled in the HEAD of
tempest, so I would believe that the bug never got fixed, or someone just
never got around to update the tickets in case, and re-enabling the test.
If anybody has any info, I would appreciate :)
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the OpenStack-dev