Created attachment 1144689 [details] engine and vdsm log Description of problem: Update VM NUMA pinning via host menu, when VM run will result to VM failed to start on next run Version-Release number of selected component (if applicable): rhevm-3.6.5.1-0.1.el6.noarch How reproducible: Always Steps to Reproduce: 1. Create VM with one NUMA node 2. Pin VNUMA to PNUMA node 3. Start VM 4. Update VNUMA pinning via host "NUMA support" menu(drag VNUMA node to another PNUMA node) 5. Poweroff VM 6. Start VM Actual results: VM failed to start Expected results: VM succeed to start Additional info:
2016-04-07 12:34:28,030 ERROR [org.ovirt.engine.core.dal.dbbroker.auditloghandling.AuditLogDirector] (DefaultQuartzScheduler_Worker-70) [74a8d9d2] Correlation ID: null, Call Stack: null, Custom Event ID: -1, Message: VM test_numa is down with error. Exit message: invalid argument: Failed to parse bitmap ''. The VM configuration sent from the engine is: numaTune={nodeset=0, mode=interleave}, guestNumaNodes=[{memory=2048, cpus=, nodeIndex=0}], Can you have a look?
Moving from 4.0 alpha to 4.0 beta since 4.0 alpha has been already released and bug is not ON_QA.
oVirt 4.0 beta has been released, moving to RC milestone.
Artyom and I can confirm that while the NUMA pinning still gets lost when editing the VM through the "NUMA Support" dialogue on the host, the VM starts fine. @rgolan I think we can remove the blocker flag.
But this means that all the current numa vms will fail to start
@rgolan we should do a backport to 3.6
Verified on rhevm-4.0.2-0.2.rc1.el7ev.noarch