CRC errors when backing up one VMDK

[FireExit]

Hi,
We had a support case 06320318 for a backup job which was failing with a CRC error. I'm not asking what it means or how to fix it, I'm curious - for a CRC check there must be a stored checksum value, where is that stored? To have a checksum for 77kb request size(?) some 36GB offset into the disk suggests VMDK files are filled with lots of checksums all through them? Are they?

Which side (VMware / Veeam) is doing the checksum check?

The logs came out with:

--deleted by moderator--

Always the same requestsize and offset ([requestsize = 77824] [offset = 37569167360]), I think.

[PetrM]

Hello,

It's calculated by Veeam on fly by the source Data Mover and repeated by the target one on retrieval. The algorithm is applied to detect silent data corruption in transmission. Usually, when we see different offsets in CRC errors, it means that there is a CRC mismatch due to a network-related issue: blocks are corrupted on a way from one Data Mover to another. In your case, most likely there is a problem with .vmdk itself and we simply cannot compute the input block checksum in the first place.

You may try storage vMotion and run an active full after it but the best option is to follow the recommendations of our support engineer.

Also, I deleted log snippets above as per forum rules displayed when you clicked New Topic.

Thanks!