Running 6.3.4.38 . Cloudberry now is requiring checkpoints to be enabled in order to complete hyper-v backups. Ok fine, I enabled checkpoints. Now I have full tree's of unmerged checkpoints for all my virtual machines.
If cloudberry is going to require checkpoints for hyper-v backups, it should be able to merge (delete) the checkbpoint after the backup operation is completed.
A tree of checkpoints can reduce VM performance from a disk I/O perspective. Cloudberry needs to actually test the results of the updates they make for huge application changes such as this!
Merging the checkpoints should happen automatically. I had someone in Support test on Hyper-V and he was not able to reproduce this issue - all the checkpoints (both Production and Standard) were being merged right after the backup completed.
To see why you're running into this issue, we'll need to have you create a Support Ticket. You can open one up from the Tools | Diagnostic toolbar option or on the web site.
We too are running into this issue at multiple sites. Cloudberry support said its a known issue but that thought makes me cringe that this could crash our servers. My colleague can comment more on this issue and maybe provide a fix.