Solved

Unable to restore full virtual machines for VMware client

  • 24 August 2022
  • 5 replies
  • 1534 views

Userlevel 2
Badge +9

Hello! I’m trying to do a “full virtual machine” restore of a few VMs, and I think that something is majorly wrong with the system. Disclosure that full VMs are not something we have to do very often at all.

About 6 months ago we switched from using a Dell SAN that had intellisnap support in commvault via Dell Storage Manager (DSM) to a cheaper SAN (a powervault) that doesn’t have intellisnap support. So at the recommendation of our Commvault local partner that we buy our licenses through, we switched our VSA backup type from NBD/Auto to SAN.

Now, for the first-ever attempt of a “full virtual machine” the restore job starts, then I immediately get an error like below.

Error Code: [91:81]
Description: Unable to write data to the virtual disk [[iSCSI (ME2)] VM4_testrestore_delete/VM4_testrestore_delete.vmdk] because it is read-only. Please ensure that the proxy has write access to the disk. For SAN mode it may be necessary to use DiskPart to clear the Read-Only attribute.
Source: mediaagent01, Process: vsrst

I then attempted a second, different virtual machine and got the exact same error.

Both machines are Linux, so I’m not sure what the error means by “use diskpart” in this case.

I confirmed that nothing has changed in terms of the service account - which is what it means by the proxy, right? (DOMAIN\CommvaultSVC user account) - this account is a VMware administrator and has full “administrator” level rights to all datastores in vcenter. vCenter shows successful events where CommvaultSVC ‘creates’ all the files required to do the restore in the datastore, but it seems like these files are just empty/dummies and it cannot actually fill the files with the data stored in commvault.

I also tried a different type of restore, the “guest files and folders” option to another VM in the same subclient, and this works just fine.

icon

Best answer by ZachHeise 24 August 2022, 22:20

View original

5 replies

Userlevel 7
Badge +17

Your policy is probably preventing write access to the SAN lun.

I will check for the necessary commands, brb

Userlevel 7
Badge +17

Here is the probable solution: https://kb.commvault.com/article/60448

  1. Open a command prompt.
  2. Run diskpart.
  3. List and select the SAN shared disks.
  4. Type attribute disk clear readonly.

For example:

C:\Users\Administrator> diskpart
DISKPART> list disk

Disk ### Status Size Free Dyn Gpt
-------- ---------- ------- ------- --- ---
Disk 0 Online 64 GB 0 B
Disk 1 Offline 78 GB 2048 KB <--- The share disk is usually reported as Offline

DISKPART> select disk 1
Disk 1 is now the selected disk.

DISKPART> detail disk

ROCKET IMAGEFILE SCSI Disk Device Disk ID: 02872505 Type : iSCSI Bus : 0 Target : 0 LUN ID : 0 Read-only : Yes <--- Make sure of read-only mode Boot Disk : No Pagefile Disk : No Hibernation File Disk : No Crashdump Disk : No

There are no volumes.

DISKPART> attribute disk clear readonly

Disk attributes cleared successfully.

 

Edit: Whoops you have Linux, my bad 😶
I am assuming you were referring to the media agents which have the SAN disks.

Use this instead:

  1. Open a shell.
  2. Run this command: sudo /sbin/blockdev --setrw /dev/sdxx

In place of /dev/sdxx, use the name of the device or devices that were listed in the error message.

Userlevel 7
Badge +17

Question though, are your VM's thick or thin imaged?

Reason for asking, there are some general guidelines when restoring:

  • SAN mode is generally better for Thick Eager Zero restores
  • NBD or HotAdd is generally better for Thin and Thick lazy zero restores

If you perform Thin and Thick lazy zero restores via SAN you could have slower restore performance then via NBD or a HotAdd VSA proxy.

Userlevel 2
Badge +9

Actually, you mentioning the LUN not having permission jogged my memory, Jos. The answer that ended up working for me was completely unrelated (I’ll be honest, I’m not sure how diskpart would help me here, also that link you shared - https://kb.commvault.com/article/60448 - goes to a blank page for me; does it work for you?)

What solved it was logging into my powervault’s web GUI console, going to mediaagent01’s mapped LUNs, and noticing, haha - I had set them to ‘read-only’ when initially we set up this new SAN. Changing the mapping from read-only to read-write for the relevant volume (VMware iSCSI 2 in my case) completely solved the problem and I was able to restore the linux VM without an issue.

 

Userlevel 7
Badge +17

My reply was indeed based on the OS being the blocking factor.
I have actual scenario's where this was fixed by setting the LUN's attributes.
Didn't think about LUN mapping, but glad I at least gave you a nudge into the right direction 😀
Good to hear it's working now 👍

Reply