VBR v10a does not respect usage of proxy while backing up that proxy VM

[asc4]

After last weeks upgrade from VBR v9.4a to v10a one of our replication jobs got broken with error message "Deleting helper snapshot Error: An error occurred while reverting to a snapshot: Invalid change tracker error code."

Investigation showed that snapshot config of replica VM indeed was messed up, with snapshot vmdk's missing for some disks. And a lot of them were still hot add mounted to the proxy VM transporting data for the replication update.
The replicated VM is our VCSA, so there are unusual 13 disks attached to the VM.

After deleting the replica and rebuilding from backup (what worked well) the next incremental update again broke the replica VM.

Suspecting hot add as the source of confusion led my to try replication via nbd transport instead of hot add, so I moved the proxy VM away from ESXi host, and success: The replication job worked well again.

While gathering evidence for support I realized what is going wrong: Due to our job schedule the replication job is running in parallel to the backup job that processes the proxy VM.

For years, VBR v9 honored the usage of the proxy and delayed it's backup until the proxy got unused.
v10 seems to have dropped this feature. Is that an intentional optimization or should that been considered as bug?

Support Case is #04511406.

Cheers Axel

[veremin]

The parallel processing mechanism has not changed in v10, so, what you've seen (if confirmed) is unexpected behaviour. So, let's wait till our support engineer investigate the debug logs and track the root cause of the issue. Thanks!

[asc4]

Support engineer was able to reproduce conflict between replication job and backup of hot-add proxy VM in Veeam lab both on V9.4 and V10.
So it seems we all the time had been in luck that your jobs didn't crash for years due to snaphot of proxy backup interfering with mounted volumed of replica VM.