After vcsa change Veeam job is killing Exchange server

[mosschops]

Hi All. We have a remote site that had its own virtual infrastructure inc. a veeam server onsite with storage and backups working well for a couple of years.
The VCSA at the site was decommissioned and the VMs were attached to a VCSA at our head office.
The Veeam server was reconfigured for the new VCSA and the backups jobs are running fine except for the Exchange job.
The guest CPU (Exchange) is at 100% for the duration of the job and it kills the Exchange server. We don't have the option of just letting the job run as it means the site will have no emails for hours.
Any ideas on how to get a backup of this server? If I can get a window of a couple of hours would it help if I powered the Exchange server down for the backup?

[wishr]

Hi Jason,

The behavior you are describing is not normal and most likely indicates an issue with the exchange server itself. I assume you are backing it up with application-aware processing (AAIP) enabled which utilizes Microsoft VSS technology and it causes the CPU spike for some reason.

I would recommend investigating it with our technical support engineers. If you decide to open a case, please share its ID here so we could track it.

Thanks

[mosschops]

Thanks for the reply. Yes application processing is enabled. I will log a case but at the moment i'm not a case admin for that licence. Should get that sorted today.

[Andreas Neufert]

Hi Jason,

not nice to hear. What Fedor indicated is that we are triggering VSS processing for consistency and the specific VSS writer from Exchange bring Exchange in a consistent state.
This is usually a CPU intensive task but usually do not "kill" the server and all customer things are still served. It is as well IO intensive on the storage side and I would really look into this first. (Check Datastore Latency within vcenter. Everything higher than 30ms is not within Exchange guildelines.

Let me please share as well that on the following conditions there can be additional overhead that could cause as well high CPU processing:
1) DNS is not configured correctly or work remotely. When you have migrated the system, are you sure that the DNS server setting is correct and a local AD server with GC role is selected as first DNS server (not working over WAN with the old AD Server on the ROBO office)?
2) If it is a Always On Cluster, please make sure that you have configured the Microsoft Failover Cluster correctly. (https://andyandthevms.com/exchange-dag- ... plication/) If you process multiple Failovers within a short time, the Exchange Index Service goes to 100% CPU load to keep up with indexing everything from scratch.
3) Check the Exchange Index Service and make sure that everything is configured correctly. As usually this service can cause high CPU spikes.
4) Check availability of your AD Forest Root domain (firewall ports). As the central groups/roles are defined there. If Exchange has no access you will end up in longer processing for a lot of tasks (waiting for timeouts). You should see as well some warnings issues in the Windows event log.