Cannot find dirty block information in the previous restore point, CBT will not be used

[JeroenL]

We use NBD to HotADD replication.
We are seeing comparable issues after upgrading from 12.1.2.172 to 12.2.0.334.
We first updated the Secondary location with a Veeam server only responsible for backups. The only thing that is shared between the primary and secondary locations is the VMware infrastructure. No Veeam roles are "shared".

What we see is "Cannot find dirty block information in the previous restore point, CBT will not be used". This is causing replication to take an enormous amount of extra time to complete. The issue does not resolve after a replication. Stranger is that some sessions complete fine for a specific VM, although they too seem slower and other sessions for the same VM show the CBT notification and take forever.

We will now try to add a Windows Proxy on the primary site so replication will not use NBD, but use HotAdd to HotAdd for replication.
This might indicate NBD is the prime suspect.

[Gostev]

What we see is "Cannot find dirty block information in the previous restore point, CBT will not be used".

Then this is completely unrelated to the default NBD buffer settings changed discussed in the topic you have originally posted this in, so I'm splitting this into a new discussion. When you get a chance please share the support case ID for this issue, in order to keep this new thread going.

[JeroenL]

Sorry for hijacking a toppic. It seemed related at the time.

We found the issue. We thought the only change that was performed was the upgrade of Veeam, but.....

Someone else upgraded vSphere components a new patch version, without telling the Backup administrators.
What I heard is that some of the ESXi hosts were updated, but others were not. vCenter servers were also not upgraded.
After upgrading everything to the same versions the errors were resolved.