Thursday, January 23, 2014

Event ID 5120 Hyper V 2012 Failover Clustering and DPM 2012 Backup Behavior

In this article I would recommend installing the KB 2879635 update for Windows Server 2012 based failover clusters that improves resiliency. This update should prevent the "notorious" event 5120 with description:
Cluster Shared Volume 'Volume1' ('name’) is no longer available on this node because of 'STATUS_IO_TIMEOUT(c00000b5)'. All I/O will temporarily be queued until a path to the volume is reestablished.
from happening during backup of VMs from Hyper V host located on CSV volume. Note that after installing this hotfix on Hyper V hosts, you should update integration components on Windows Server 2012 based guest virtual machines running on those hosts.

Also, If you're using DPM 2012 SP1 as you're backup solution I would recommend installing the latest rollup 3 KB2836751 and hotfix KB2886362 .

From my point of view, before installing these updates and rollups I was experiencing strange behavior when I was doing Hyper V host based backup of Windows Server 2012 virtual machines using DPM 2012 SP1. After initial replica creation of vm with installed Windows Server 2012 which is basically transferring the whole vhd(x) file to the DPM 2012 SP1 server, all other scheduled recovery points were transferring again the same size of data to the DPM server. This is an example of backup of one Windows Server 2012 vm:

 
After installing all the rollups and updates, the situation has changed, and the DPM scheduled replica creation task was transferring significantly smaller amount of data, and naturally the backup task was completing much faster.
Also, you will eliminate memory spike on node which is owning the CSV resource, during backup of VMs located on that volume fixed with KB2813630 which is included in KB 2879635.
 

No comments:

Post a Comment