CDP: normal/full sync vs "quick" failback

[FrenchBlue]

Hello,

Now that cdp replication is working fine in V12, I've started doing some failover/failback tests. I'm particularly interested in the "quick rollback" option which synchronizes only changed blocks during the failover. This could be very useful when doing DRP exercises (in case of real DRP, the failback time is not very important and doing a full sync is not an issue).
The problem is that when using this option, the process starts with a full target (original production site) disk scan, which takes roughly as long as the full resync of the VM. I tested with a small 80GB VM:
- "Quick rollback option": the re-sync only took a couple of minutes but before this, the source disk scan/hash took around 20 minutes
- Full resync: also about 20 minutes

Is there a way to optimize the quick rollback disk scan? Currently this option seems quite useless, and there's nothing better than a full resync if there is any doubt about the status of the source VM.

Thanks.

[HannesK]

Hello,
that does not look expected to me... did you already check with support and could we please get the support case number to check?

Best regards,
Hannes

[FrenchBlue]

Hello,

I didn't create a support case yet, because it doesn't <seem> to be a bug so I wanted to ask here first if it can be some parameters or tuning issue. But I can try.

[FrenchBlue]

Case opened. Te be continued.

[HannesK]

if you post the case number, then I would look into it and double check with QA

[FrenchBlue]

Hello,

Case #06023805

Thanks.

[FrenchBlue]

So for follow up, answer was in the documentation, a classic RTFM case

"Due to changes in VMware vSphere 7.0 and newer, the replica failback operation forces digest recalculation for both source and target VMs. That is why the Quick rollback option is ignored for ESXi hosts starting from version 7.0."

Since 6.x is not supported anymore by VMware, it means that the feature is not usable at all.