Bug 1729812

Summary: [HP3Par-8400] Instance creation failed.
Product: Red Hat OpenStack Reporter: Nilesh <nchandek>
Component: openstack-cinderAssignee: Cinder Bugs List <cinder-bugs>
Status: NEW --- QA Contact: Tzach Shefi <tshefi>
Severity: urgent Docs Contact: Chuck Copello <ccopello>
Priority: medium    
Version: 13.0 (Queens)CC: abishop
Target Milestone: ---Keywords: Triaged
Target Release: ---   
Hardware: All   
OS: All   
Whiteboard:
Fixed In Version: Doc Type: If docs needed, set a value
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:

Comment 3 Alan Bishop 2019-07-16 19:40:09 UTC
Once again, the logs were not collected with DEBUG enabled. Please enable DEBUG and collect the cinder-volume logs.

My understanding is the os-brick package was a hotfix. Was the hotfix installed by customizing the cinder-volume container image on the director, followed by having the director deploying the hotfixed c-vol image? Or was the os-brick RPM patched into the running c-vol container? I ask because I'd like to confirm the c-vol process was restarted after the os-brick RPM was updated.

Comment 6 Alan Bishop 2019-07-23 20:15:31 UTC
The symptoms match a known problem with the 3par driver [1]. The LP bug
describes a situation where the FC zoning doesn't allow the controller to
access all target ports on the 3par backend, and the 3par driver happens
to pick a port that isn't accessible.

[1] https://bugs.launchpad.net/cinder/+bug/1809249

Per LP bug, a workaround is to set use_multipath_for_image_xfer=True, which 
will force the driver to scan all ports.

If setting use_multipath_for_image_xfer=True doesn't correct the problem, then
we may need to get HPE involved. If that's necessary, then please set
hpe3par_debug=True and capture more cinder-volume logs.