When I set local maintenance on the self-hosted engine hypervisor host through the manager GUI it becomes stuck in state "Preparing for maintenance". The hosted engine virtual guest is automatically migrated because the HA broker is successfully set to maintenance mode=local. The host returns to state "Up" in the manager if I press button "Activate", but the column "Virtual Machines" continues to count the migrated VM. If I then reboot the host it reverts to state "Up" and the VM count no longer includes the vacated hosted engine, so I'm guessing it never *really* achieved state "Maintenance" even though the event log claims so immediately after the button is pressed. The condition persists after the reboot if the then-current hosted engine hypervisor host is again assigned to mode "Maintenance". I'm running 4.0.1, and I've tried restarting services and rebooting the hosts and VM involved. I suspect it has something to do with my other thread about the locked storage domain.
On 31 July 2016 at 02:24, Kenneth Bingham <w@qrk.us> wrote:
When I set local maintenance on the self-hosted engine hypervisor host through the manager GUI it becomes stuck in state "Preparing for maintenance". The
hosted engine virtual guest is automatically migrated because the HA broker is successfully set to maintenance mode=local. The host returns to state "Up" in the manager if I press button "Activate", but the column "Virtual Machines" continues to count the migrated VM. If I then reboot the host it reverts to state "Up" and the VM count no longer includes the vacated hosted engine, so I'm guessing it never *really* achieved state "Maintenance" even though the event log claims so immediately after the button is pressed. The condition persists after the reboot if the then-current hosted engine hypervisor host is again assigned to mode "Maintenance". I'm running 4.0.1, and I've tried restarting services and rebooting the hosts and VM involved. I suspect it has something to do with my other thread about the locked storage domain.
how much time did you wait? The Host is in preparing for maintenance as long as it's VM count is > 0. So if the HE VM migrated to another host, the engine will detect that and will update the count accordingly. This transition is fast but not instant, cause the engine is pulling the info from all the hosts. If the host is not running any vm and still can't be move to maintenance then lets see the logs VDSM of src and dst + engine.log _______________________________________________
Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
Thanks, Roy. It really seemed to be stuck because I waited more than 8 hours. I've since nuked the self-hosted engine VM, reinitialized the lockspace, and redployed the appliance with hosted-engine. I'm no longer able to reproduce this cycle and the main difference I see is that the import of the /engine storage domain succeeded this time and the hosted engine VM now appears in the "Virtual Machines" tab. It was always visible in 'virsh --readonly list'. On Sun, Jul 31, 2016 at 3:20 AM Roy Golan <rgolan@redhat.com> wrote:
On 31 July 2016 at 02:24, Kenneth Bingham <w@qrk.us> wrote:
When I set local maintenance on the self-hosted engine hypervisor host through the manager GUI it becomes stuck in state "Preparing for maintenance". The hosted engine virtual guest is automatically migrated because the HA broker is successfully set to maintenance mode=local. The host returns to state "Up" in the manager if I press button "Activate", but the column "Virtual Machines" continues to count the migrated VM. If I then reboot the host it reverts to state "Up" and the VM count no longer includes the vacated hosted engine, so I'm guessing it never *really* achieved state "Maintenance" even though the event log claims so immediately after the button is pressed. The condition persists after the reboot if the then-current hosted engine hypervisor host is again assigned to mode "Maintenance". I'm running 4.0.1, and I've tried restarting services and rebooting the hosts and VM involved. I suspect it has something to do with my other thread about the locked storage domain.
how much time did you wait?
The Host is in preparing for maintenance as long as it's VM count is > 0. So if the HE VM migrated to another host, the engine will detect that and will update the count accordingly. This transition is fast but not instant, cause the engine is pulling the info from all the hosts. If the host is not running any vm and still can't be move to maintenance then lets see the logs VDSM of src and dst + engine.log
_______________________________________________
Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
participants (2)
-
Kenneth Bingham -
Roy Golan