Description of problem: engine must propogate its knowledge of problematic VDSs quick and wide to avoid using them for VM scheduling or for SPM actions. see also bug 767222 Version-Release number of selected component (if applicable): How reproducible: 100% Steps to Reproduce: 1.create a cluster with 3 hosts - A (SPM) B and C 2. run several vm's on spm host 3. block storage connectivity with iptables to A and B Actual results: - engine will pick A and B as problematic - meanwhile will try to migrate the VM to B. - migration will fail obviously (B doesn't see storage) - the VM will return to status Paused and won't re-run (engine doesn't support migrating paused VM see bug 767222) - C is not used Expected results: - engine should pick C as the destination host for migration and neglect B till is ok Additional info:
Created attachment 654415 [details] logs for roy
Closing old bugs. If this issue is still relevant/important in current version, please re-open the bug.