VMware Cloud Community
bergerpjr
Contributor
Contributor

backup of vCenter in VDR 2 causing cluster failures

We're using VDR v2.0.0.1861 on vCenter v5.0.0.455964. Whenever we run a backup of the vCenter VM we get the error message below from each and every one of our ESXi hosts. The VDR backup runs fine and works -- however while quiescing, it times SQL out just long enough for vCenter to complain about itself.

We moved the vCenter's VDR backup to its own backup job during low periods of activity and no other VDR jobs running and it still happens.

VMware technical support poingted me to the VDR 2.0 Administration guide (page 40 - Enable Windows 2008 Virtual Machine Application Consistent Quiescing) where they want to make sure the UUID was set correctly for the 2008R2 VM -- which it was/is.

I am told that there is a way to use  "custom quiescing scripts" in "C:\Program Files\VMware\VMware\Tools\backupScripts.d\" to quiesce VSS-aware applications. I've read about quescing Exchange/SQL, etc. -- but is there a way for vCenter to quiesce its own SQL application without bringing down vCenter? If so -- what would be the syntax?

Or....if there isn't a way to have vCenter quiesce itself via VDR...is there a way to modify the vCenter alarm "Host connection failure" and "Host connection and power state" to increase the timeout value? This would give VDR more time to have vCenter quiesce itself before timing out and complaing about a cluster outage.

Any help/411 would be great. thanks -- Peter

> -----Original Message-----

> From: vCenter@foo.com [mailto:vCenter@foo.com]

> Sent: Wednesday, October 05, 2011 10:31 AM

> To: OnCall_Technical_Services

> Subject: [VMware vCenter - Alarm alarm.HostConnectionStateAlarm]

> alarm.HostConnectionStateAlarm changed status from Green to Red

>

> Target: ncb-vs-003.foo.com

> Previous Status: Green

> New Status: Red

>

> Alarm Definition:

> ([Red state Is equal to notResponding] AND [Red state Not equal to

> standBy])

>

> Current values for metric/state:

>  State = Not responding AND State = Unknown

>

> Description:

> Alarm 'Host connection and power state' on ncb-vs-003.foo.com changed

> from Green to Red

Reply
0 Kudos
2 Replies
bergerpjr
Contributor
Contributor

addendum: I saw vmware's KB #5962168 -- but the vCenter box doesn't even have this SYNC driver installed.

Reply
0 Kudos
bergerpjr
Contributor
Contributor

In working with VMware Technical support some more -- we changed the vCenter heartbeat timeout from 120 to 300 seconds in the "\\vcenter_server\C\ProgramData\VMware\VMware VirtualCenter\vpxd.cfg" file in the XML area:

                <vpxd><heartbeat><notRespondingTimeout>300</notRespondingTimeout></heartbeat></vpxd>

Changed the default time from 120 to 300. Restarted the VMware server services.

Ran the vCenter backup manually and let it run at the normal schedule and no time-out issues in the cluster. yeah. hope this helps others.

Reply
0 Kudos