Caution: VCHA + Snapshots

I’ve been playing around with VCHA and had been taking snapshots and have encountered some issues using snapshots on a VCHA enabled VCSA 6.5

You need to exercise caution when using snapshots on a VCSA that is enabled for VCHA.

Important: This is only when snapshot tasks are being done to the VCSA 6.5 VM. For all other VMs in your inventory there is no issue.

This includes taking snapshots of the current Active, reverting to snapshots of the current Active and deleting snapshots of the current Active.

For any and all of those tasks, ensure that the VCHA Cluster is in “Disabled” mode.

Taking and deleting snapshots will stun the VM which can trigger a VCHA Failover.

Reverting to an earlier snapshot will result in the Active going “back in time” but the Passive will not (due to the replication) and this will also trigger a failover and result in the PostgreSQL replication to fail.

Placing the VCHA cluster into “Maintenance” mode is not enough. It must be placed into “Disabled” mode to turn off automatic failover and replication.

Do not leave “Disabled” mode until you all done with snapshot operations and have deleted all snapshots on the VCSA.

Once you have deleted all snapshots, you can safely place the VCHA Cluster back to “Enabled”

How this impacts Image Level backups of a VCHA enabled VCSA 6.5 I do not know at this time.

Update: VMware has released a vCenter HA FAQ KB Article outlining that snapshot tasks against the VCHA Nodes is not supported.