system cannot boot after creating a snapshot --The redo log of 'HomeServer-000001.vmdk' is corrupted

[liusirjiayou]

The storage is on MSATA, and system cannot boot after creating snapshot. The issue disappears after deleting the snapshot.

Here is the error I got:

The redo log of 'HomeServer-000001.vmdk' is corrupted. If the problem persists, discard the redo log.

[asajm]

Hi liusirjiayou​

Look this kb VMware Knowledge Base

[liusirjiayou]

More info:

The virtual system is on MSATA and the MSATA has no bad sector.

It works well if the vmdk on the MSATA is moved to SATA and creates snapshot, and yet error occurs as long as creating snapshot on MSATA.

[NathanosBlightc]

Is that a large snapshot file ? I mean your delta disk that is generated after snapshot creation.

Is there enough space on your datastore to redo action? Is that large enough more than your initial VMDK (for committing/consolidation actions in snapshots management) ?

[continuum]

Connect via ssh.

Go to the directory where that delta is stored.

Run the command:

vmkfstools -p 0 HomeServer-000001-delta.vmdk > homelab-delta-map.txt

(adjust accordingly if you have a sesparse delta file.)

Post the textfile if possible or report results if the command fails.

[liusirjiayou]

Thanks for the reply. Here is the file content:

Mapping for file HomeServer-000002-sesparse.vmdk (71303168 bytes in size):

[           0:     1048576] --> [VMFS -- LVID:5ce69e20-bb446faa-7df8-0c54a5517aa6/5ce69e1c-5306ddcc-8bf1-0c54a5517aa6/1:( 67009249280 -->  67010297856)]

[     1048576:    70254592] --> [NOMP -- 😞          0 -->     70254592)]

----------------------------------------

[Snify]

I have the same problem with my (M)SATA Controller... Running VMs was no Problem until the first Snapshot was taken.

The snapshot works as you create it but as soon as ESXi (actually the guest system) tries to continue to write to the snapshot file (to the virtualized drive), I get the same error and my VM crashes.

A workaround, which I did, was to take out my SSD drive and mount it as a USB drive. Now it works. Such a shame that only snapshots are the issue here, otherwise everything works just fine without any snapshots.

Would love to see a fix on this .

[liusirjiayou]

@TimMann would it be possible provide some clue to help me fix this issue? haven't found a workable solution. Thanks a lot!

[huw02]

Hello,

I had the same error from ESXi version 6.5 including 6.7u3! What is weird, is that it only happen on my ESXi with a local SATA attached Samsung 840 pro 500GB SSD. On my other ESXi host (also 6.7u3) with MSATA Intel 120GB and a Samsung 850 pro 1TB SSD disk the issue never happens.

I have found a good article about VMFS and SEsparse changes in those ESXi versions. If I understand it right, the bug has been fixed in the update 1 of ESXi 6.7. But for me it was not the case unfortunately.

The only solution that is working for me was to delete the partition table and create a new datastore in the old VMFS5 format (instead of VMFS6).

With this change I can now make snapshot again without any redo log issue.

One thing with this workaround is that I can not have vmdk bigger than 2TB, otherwise SEsparse would be used with VMFS5. And SEsparse is what seems to be related to this issue.

I hope this will help you and also that vmware will re-open this issue that should have been fixed with the update01.