Hi I have installed server 2012 standard in our production network as a 4 nodes cluster. Also installed Microsoft system center DPM 2012 SP1 as the backup solution.
I have been trying to take backup of our VMs on the CSVs and noticed a very strange behavior while i'm taking backup.
The CSV owner cluster node RAM is consumed rapidly and if the backup is for a large sized VM and taking a long time the RAM of the CSV owner may get 100% consumption.
This doesn't affect the backup process itself the backup is done successfully but for any reason any other process is running on this host gets affected badly.
Yesterday one of our exchange team was working on upgrading one of the servers which happens to be a VM that's running on a node with the above case, the upgrade required the VM to restart. The VM shutdown but never started because the needed RAM was consumed
by the backup process. This host services was stopped and the guest VMs was moved to other servers in the cluster but the restarted machine was not started until I stopped the physical server itself.
I don't know if this is a backup problem or the server problem but we were using this server to backup 2008 R2 cluster without this problem to appear, so I guess it's a server 2012 Cluster problem.
I hope that someone can help if he faced a similar issue and if there is a hotfix for it or so, as I can't just tell everyone to stop working in the backup time, and what if there is a problem in one node during the backup time and the VMs has to be migrated to one of the nodes that is affected by this backup issue. I think its a very dangerous situation on a production environment.