CBT on first (full) backup, how?

[TrK]

Yesterday i made clean reinstall of my Veeam B&R v5 to NEW computer, create a NEW job and select NEW destination file.
On first (full) run some VM shows "Backup mode: NBD without changed block tracking" (and it`s ok, AFAIK), but some shows "NBD with changed block tracking".

How is it possible? And can it be a problem on future restore?

[Vitaliy S.]

Hello Dmitry,

I'm fairly sure that the full run was performed without Changed Block Tracking (based on your descripition). However, could you please clarify if you've seen this message while processing VMDK disks?

Thank you.

[Gostev]

Changed block tracking can be used during full run in a special mode that only returns VMDK blocks occupied by data. So what you are seeing is normal. Thanks.

[TrK]

could you please clarify if you've seen this message while processing VMDK disks?

Yes, i`ve seen this while processing and after that.

[TrK]

Changed block tracking can be used during full run in a special mode that only returns VMDK blocks occupied by data.

Ok, then why other VMs was processed without CBT?
Almost forgot. I have four ESXi - two 4.0u1 and two 4.1.Could this be the cause of this issue?

[Gostev]

The real cause for this issue is that this so-called "initial" CBT does not work reliably with VMware. For some VMs it works, for other it does not. By "does not work" I mean not that it fails to return only occupied blocks, so our engine has to go through all blocks including unoccupied (which is still very fast). Thus, these failures do not really cause any issues to our product.

We have put some effort into researching this last year, but did not find any patterns on why it would not work in some cases. We did find that in most cases, this can be "fixed" by disabling/enabling CTK in VM settings paired by VM poweroff after each change (can't remember exact sequence, multi-step process). So most likely just a bug on VMware side. But since working initial CBT is not too big of a deal for a full backup, we stopped pursuing this (this issue does not affect "incremental" change tracking, this one is flawless).

[joergr]

I have also done quite some research on this and found no pattern, it happens even on the very same esxi 4.1 machine with freshly 1:1 deployed windows servers which are absolutely the same. It is exactly the way like Anton explained to you, some are doing CBT from the very first backup, some are not. But what is very important to know here: Your Backup will be 100% OK, neither it can do CBT on first run nor it can´t.

Best regards,
Joerg