Description of problem: I have a VM that should start on host_1, but is started on host_2 instead. In log I could find that it was started there because host_1 was not accessible, but I can migrate the VM there afterwards with no problem. Version-Release number of selected component (if applicable): 4.1.1-1 How reproducible: always Steps to Reproduce: 1. set a VM to run on host-1 2. have other VMs running on host-1, but leave enough resources for the VM that is pinned there 3. start the VM Actual results: VM is started on different host Expected results: VM should be started on host it's pinned to if there are enough resources for it Additional info:
> Steps to Reproduce: > 1. set a VM to run on host-1 Can you elaborate? What exactly did you do?
This happens when you set host in 'start on host' in host edit dialogue and have 'migration possible' or 'only manual migration possible' set in migration policy and start the VM. You can fix this by setting higher weight in scheduling policy or by setting the the host in run once dialogue or choose 'no migration possible' in migration policy.
This bug report has Keywords: Regression or TestBlocker. Since no regressions or test blockers are allowed between releases, it is also being identified as a blocker for this release. Please resolve ASAP.
*** Bug 1426740 has been marked as a duplicate of this bug. ***
Fails for ovirt-engine-4.1.1.2-0.1.el7.noarch
(In reply to Andrei Stepanov from comment #6) > Fails for ovirt-engine-4.1.1.2-0.1.el7.noarch - Can you check whether the cluster policy shows 99 next to the PreferredHosts cluster policy? - Does the VM start on your host when you use Run Once and select the host? - Can you attach the log with DEBUG enabled according to the testing section of: http://www.ovirt.org/develop/release-management/features/sla/scheduling-weight-normalization/
Created attachment 1258683 [details] Host assign
The status of this bug is supposed to be MODIFIED only. The patch was merged and bug moved to modified after tagging for 4.1.1.2 but before the CI moved all modified bugs to ON_QA (including this one). The fix will be part of the build tomorrow.
Verified on rhevm-4.1.1.4-0.1.el7.noarch "Preferred Hosts" - weight module now has the factor equals to 99 for all default policies.