[ovirt-users] Why Node was rebooting?

[Charles Kozler]

Did you setup fencing?

I've also seen this behavior with stressed CPU and NMI watch dog in BIOS
rebooting a server but that was on freebsd. Have not seen it on Linux

On Nov 25, 2017 2:07 PM, "Jonathan Baecker" <jonbae77 at gmail.com> wrote:

> Hello community,
>
> yesterday evening one of our nodes was rebooted, but I have not found out
> why. The engine only reports this:
>
> 24.11.2017 22:01:43 Storage Pool Manager runs on Host onode-1 (Address:
> onode-1.worknet.lan).
> 24.11.2017 21:58:50 Failed to verify Host onode-1 power management.
> 24.11.2017 21:58:50 Status of host onode-1 was set to Up.
> 24.11.2017 21:58:41 Successfully refreshed the capabilities of host
> onode-1.
> 24.11.2017 21:58:37 VDSM onode-1 command GetCapabilitiesVDS failed: Client
> close
> 24.11.2017 21:58:37 VDSM onode-1 command HSMGetAllTasksStatusesVDS failed:
> Not SPM: ()
> 24.11.2017 21:58:22 Host onode-1 is rebooting.
> 24.11.2017 21:58:22 Kdump flow is not in progress on host onode-1.
> 24.11.2017 21:57:51 Host onode-1 is non responsive.
> 24.11.2017 21:57:51 VM playout was set to the Unknown status.
> 24.11.2017 21:57:51 VM gogs was set to the Unknown status.
> 24.11.2017 21:57:51 VM Windows2008 was set to the Unknown status.
> [...]
>
> There is no crash report, and no relevant errors in dmesg.
>
> Does the engine send a reboot command to the node, when it gets no
> responds? Is there any other way to found out why the node was rebooting?
> The node hangs on a usv and all other servers was running well...
>
> In the time, when the reboot was happen, I had a bigger video compression
> job in one of the VMs, so maybe the CPUs got a bit stressed, but they are
> not over committed.
>
>
> Regards
>
> Jonathan
>
> _______________________________________________
> Users mailing list
> Users at ovirt.org
> http://lists.ovirt.org/mailman/listinfo/users
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ovirt.org/pipermail/users/attachments/20171125/bed68098/attachment.html>