Invalid snapshot configuration - howto fix it without losing

[rosanaduran]

Hi,

Case ID 01818944

Suddenly last monday replicas stopped working just in one task. Veeam made the replica at 9.00am but it failed at 10.00am. The code: invalid snapshot configuration. In Vcenter, the vm_replica had an alert "Disk consolidation needed" and when I forced the consolidation I obtained an error "Snapshot not found"

I know that I can remove all snapshots, all replica disks, and start from scratch with the replica again, but this is a critical machine and I had this issue more than once.

If I edit settings on the vm_replica disks, the two of them are pointing to replica disks, no base disks:

Example:

HARD DISK 1 - [DATASTORE]vm_replica_folder/vm-000030.vmdk - provisioned size 0 MB
HARD DISK 2 - [DATASTORE]vm_replica_folder/vm_1 -000030.vmdk - provisioned size 1,73 MB

If I browse the datastore folder I have the base disks, vm_1-000001.vmdk to vm_1-00030.vmdk, but only vm-000001.vmdk to vm-00029.vmdk. Then many snapshots, but with a different number sequence.

If I tried to restore a file from the replica points it works, so I asked the technical support to remove just the last snapshot or to point the vm_replica disks to the 00029 instead of 00030 (which are faulty), but they told me that the proper way of doing it is removing any replica. Removing partial information can be inconsistent.

I don't know how it internally works, because if I see the replica disks browing the datastore folders, all of them have the same date, so with the replica disks are not totally independent among them, the last replica modified them all. At first I thought that, as everything was working properly at 9.00am there should be a way to go back to that point at resume from there. If everytime I have an issue I have to lose all the restore points and replicas, then the replica system is not enough and I need a second method to store and save the data of the company.

I'm also looking for the cause, because this issue or similar issues happened three times in the last 6 months. The problem is that removing everything doesn't prevent the system from failing again.
Is there anything else I can do or test before deteting everything?

Thank you in advance,

[larry]

I have had this happen, a ESX host has the file locked. Find the host locking the file and reboot it. Most of the times the host locking it is the host it is on. Also be sure a proxy does not have the drive mounted, if it is mount then just remove the mount (not the disk:) ) I have seen this a couple times since upgrading to VMware 6 but mostly on backups, there is a VMware knowledge base with the whole fix. This has fixed it every time for me.

[rosanaduran]

Thank you for your answer Larry. I answered to your comment last week, but I can't see it right now.
I was trying to find a locked file as you said, but I can't find any. It is weird, because in this case browing the datastore or through ESXi commands, that file doesn't exist.
Finally I removed all snapshots and previous replication points and started from scratch.

Now I'm trying to find some recommendations about replicas and backups, I think I'm depending on the replicas too much, and when they fail, I lose everything. I understand that using backups, removing newer backups don't need to remove the previous consolidated full backups.