Instant recovery of rhel 8 server to vmware duplicated vmdks

[sfaits]

I have a linux rhel 8.7 server that we have been backing up using the veeam linux agent. The VM has three vmdk files one that is a few hundred gigs and two that are 1terabyte or more.
Recently it became necessary to restore the entire VM from backup. We used the Instant recovery and when we started to migrate it to production storage we found it had doubled in size. Since we had to have the VM we went a head and completed the migration hoping this process would sort out the issue. Sadly it did not. The server is live and powers on fine. The VM is still showing 6 vmdk, 3 original and 3 dupes. I have no idea what happened or how to identify the ones that should not be there so I can remove them.

Has anyone else seen this with agent-based instant restores?
Any thoughts on how to identify the unneeded vmdks?

Thank you for any input you can offer on this.

[chris.childerhose]

The easiest way in vCenter would be to edit the VM and check each disk, as it will tell you the datastore and the VMDK name it uses. Then you can see where it is pointing and hopefully remove the other three VMDK files that are not used.

[sfaits]

They are all on the same datastore and the disk names does not help me identify which ones are actually in use on the VM. Thank you though.

I looked in the vmx file and of course just lists the disks in scsi_id order.

In a windows os I usually just go in computer manager, disk management, then look at each disk's propertie location and can see the Bus number and Target ID which matches the Disk SCSI node information on each disk (when looking at edit settings on VM and going into a disk.

[chris.childerhose]

If you edit the settings of the VM and expand each disk, it will show you the path to the VMDK file, including the datastore. This is what I meant to help know which ones are used and which are not.

[sfaits]

I guess I am not completely understanding, thank you for your patiance and willingness to explain futher Chris
I can go into the Edit settings and expand the disks and it shows all 6 of them, the 3 original (not sure which they are) and the 3 duplicates.
Its like the instant restore duplicated the original 3 disks and updated the vmx file to include them even though the RHEL OS is not using them.

[chris.childerhose]

Oh ok. Then that is an issue for sure. It might be best to open a support case with Veeam to get to the bottom of it. I thought in the settings; it was only three disks, not all six.

[sfaits]

I have a case open but no real response yet.

I did find a command lsscsi that I think gives me what I need to identify which disk is which.
Just need to do a little more digging before I pull the trigger.

Thank you again for your input Chris

[chris.childerhose]

Not a problem. Post back if you get it fixed and what the main issue was.

[PTide]

Hello,

I've taken this issue to a higher level. Cannot promise any ETA right now though.

Thank you!

[rovshan.pashayev]

I have a case open but no real response yet.

I did find a command lsscsi that I think gives me what I need to identify which disk is which.
Just need to do a little more digging before I pull the trigger.

Thank you again for your input Chris

Good day,
I wonder if you managed to successfully remove “duplicated” disks. Following is basic action to do on Linux VM for SCSI vDisk removal:
# echo 1 > /sys/block/sdX/device/delete
Where sdX is the SCSI device you want to remove/detach, after identifying with lsscsi command. Of course, you must be 100% sure that disk is not being used first.

Rovshan

[sfaits]

Hello Rovshan,
I have only recently been able to start exploring this issue again as other fires were burning
I had another smaller linux vm (rhel that was restored in the same fashion and had duplicate vmdks as well so this gives me something I can test with out affecting production data.
So far I am not having much luck. Removing what I thought was the unneeded disk in vcenter and the VM booted into emergency mode until it was added back. I then removed the other disk and the VM acted as if there was no OS.

I have not attempted your command yet. Just to make sure I am understanding. You are saying that within the OS identify the duplicate disk by the SCSI id (such as sda, sdb, etc) and then use the command "echo 1 > /sys/block/sdc/device/delete" Assuming in this case the dupelicate disk is sdc.

Thank you for your input on this I will give it run and update as soon as I am able.

[rovshan.pashayev]

Hello,

Please share output of following commands, from your "new" test RHEL machine:

Code: Select all

df -hT; lsblk; lsscsi

Rovshan