Hi
Hope someone can help with a problem that we have been experiencing.
We have recently started supporting a company that has VMware cluster with two physical nodes managed by a vCenter server. Datastores are provisioned from a Dell MD3000. Some of the VMs are being backed up using Veeam.
Some work was recently done to upgrade the memory in the physical hosts and as part of this the VMs were "balanced" across the two physical nodes using vMotion. After this work several of the Veeam backups started failing and we realised that it was the VMs that were originally on node 1 that had been migrated to node 2 that were failing. Veeam was complaining that the NFC server was busy and if the VMs were moved back to node 1 the backups were successful.
After trying various things we have come to the conclusion that the problem we are experiencing is the one described in the following KB:
http://kb.vmware.com/kb/1019286
It appears that trying to download a VMX file or running a Veeam backup using vCenter will always go through node 1 regardless of which node is actually hosting the VM. If the Veeam backup is configured using the physical node rather than the vCenter server the backup works.
I would like to be in a postition to have the Veeam backups using the vCenter server so that if any of the VMs are migrated the backup will not need to be reconfigured but the article above doesn't give me enough information to work out how to achieve this or even if it is possible.
I'm assuming the problem is with vCenter and/or node 2 but I don't know where to start in trying to resolve this, especially as I have had no part in the original configuration of this environment. As a side note, which may or may not be related, using vMotion to migrate VMs from node 1 to node 2 seems to take a lot longer than going from node 2 to node 1. I'm wondering if node 2 has some fundamental issue but what it is or how to diagnose it is beyond me at the moment, so any help appreciated.
Cheers
Rob