SUMMARY
Backup of Two VMs in a DAG Configuration Fails
ISSUE
Back up two Exchange VMs in DAG configuration
Description
Create separate schedules for any Exchange VMware virtual machines in DAG configuration.
Cause
When backing up two Exchange VMware virtual machines in a DAG configuration at the same time with application aware backups, both of the backups fail because the DAG is active on only one of the VMs.
Resolution
Create two schedules for your VMs. One with one Exchange server, one with the other. Be sure the schedules run at different times. Any overlap may cause a failure.
Third-Party Sources
RESOLUTION
Though it is fully supported to run Exchange in a DAG configuration inside a VM, Microsoft as well as VMWare do not support the protection of an Exchange Server in DAG configuration, and such Unitrends does not either. The snapshot of a DAG node can result in unplanned cluster failover and potential data loss.
For older versions of Exchange: https://docs.microsoft.com/en-us/previous-versions/office/exchange-server-2010/jj126252(v=exchg.141)
For current versions: https://docs.microsoft.com/en-us/exchange/plan-and-deploy/virtualization?view=exchserver-2019
The ONLY safe way to protect an exchange server that is in a DAG configuration is via agent-based protection.
Configure protection of each node's OS and file system using standard asset backups, and configure a schedule to protect each passive DAG database using Exchange protection options.
Note: protecting Exchange databases using agent protection rewquires Unitrends Enterprise UB licenses, or, a Unitrends Physical Appliance.
Note: Anytime a passive node becomes active, a manual full backup will be required. For this reason Microsoft recommends DAG configurations have a minimum of 3 nodes and protect the lowest priority node that is least likely to become active.
Protection of the active node is supported but may result in performance issues or failure to successfully truncate logs. If a node failover occurs and does not fail back returning the prior protected node to a passive state, Unitrends recommends disabling the backup job for the prior passive node and configuring one for the new passive node to avoid performance issues, or, manually fail the node back to it;s original active state if possible. If you have a mix of active vs passive nodes on several servers, you may require multiple schedules to accomodate.
DO NOT back up the same database on more than 1 node!
NOTES
Microsoft does not provide APIs for 3rd party DAG cluster awareness as is provided for SQL. Backup vendors can support active or passive node backup, but, cannot automatically detect or react to DAG failover events. Manual remediation of jobs and manual full backups are required when a cluster configuration chaqnges or a failover event occurs that makes a passive database become active.