VM has disks with multi-writer option enabled, skipping

[vminsf]

No change to the jobs, but they started failing after the upgrade. Support has no awareness of any change to Veeam.

Message: "...has disks with multi-writer option enabled, skipping"

[PTide]

Hi,

Apparently you have disks that have multi-writer flag set. If so, then are those disks configured as independent? Please also share you support case ID for future reference.

Thanks!

[Gostev]

Also, what do you mean by "jobs failing"? They should be simply skipping such VMs (because VMware does not support snapshotting them), and moving along to other VMs - instead of failing? In fact, just a few days ago I've received a "thank you" email for implementing this feature along with a screenshot of a 100% successful job run showing those VMs automatically skipped

Overall, this change I am well aware of - and it has nothing to deal with CBT disabled. Before U4, we would attempt to process multi-writer VMs with shared disk not marked as independent (which is a best practice) normally. So, creating a VM snapshot would cause issues to a VM and required its reboot. Because of that, our customers had to manually exclude each such VM from every job, which was a tedious on-going process - and we had many requests to skip such VMs automatically. This is exactly what was added in U4.

[stephen.loera]

Hello,

If a VM has a system disk that does not have the multi-writer flag enabled, and a disk that does have the multi-writer flag enabled, is there a way to backup the system disk only? Does the multi-writer disk need to have the disk mode set as "independent persistent"?

[PTide]

Hi,

Veeam issues the Snapshot API call, we will attempt to snapshot the entire VM. VMware explicitly says that "Snapshots of VMs with independent-persistent disks" is supported for vSphere 5.1 U2 and later. That is, making the multi-write-enabled disk independent-persistent should allow you to backup only the other disk, however I haven't tested that myself. Meanwhile, you can try it by yourself, or resort to an alternative solution - Veeam Agent for Windows.

Btw, what kind of an application do you use with the multi-write disk?

Thanks!

[stephen.loera]

Hello PTide,

The use case is a a Windows File Server Cluster with 2 Nodes (VM) that access a shared VMDK. I am currently using the Veeam Agent, which is cluster aware, to backup this File Server cluster locally. Then a backup copy job sends that backup to a VCSP.

What I want to accomplish is replicating the 2 VM nodes offsite to a VCSP, and in the event of a DR, we can use a recovery plan to spin up the 2 VM nodes, and then restore the shared VMDK from the agent backup.

I hope that makes sense.

Long story short I want to replicate the OS system disks of a VM that also has multi-writer disks.

[Regnor]

Hi,

Veeam issues the Snapshot API call, we will attempt to snapshot the entire VM. VMware explicitly says that "Snapshots of VMs with independent-persistent disks" is supported for vSphere 5.1 U2 and later. That is, making the multi-write-enabled disk independent-persistent should allow you to backup only the other disk, however I haven't tested that myself. Meanwhile, you can try it by yourself, or resort to an alternative solution - Veeam Agent for Windows.

Btw, what kind of an application do you use with the multi-write disk?

Thanks!

Hi,
we have this case and Veeam skips the complete VM although the mult-writer disk is "independet-persistent".
Before U4 the backup worked successfully, ignoring the multi-writer disk.
Is there a way to get it back working?

@Anton: Now I do understand why you don't like changing existing behaviours in Veeam

[Gostev]

@Regnor feel my pain! Your scenario should be fixed in Update 4a though, as just a few days ago I had to proofread the following message:

Code: Select all

VM {0} has dependent mode disks with multi-writer option enabled, skipping (set multi-writer disks mode to independent)

[Regnor]

I forgot to report back; U4a fixes the issue so everything is fine again.

We've now noticed a different 'problem':
Does Veeam (re-)-enable CBT for independent and/or excluded virtual disks if the option is ticked in the storage section?

[foggy]

Could you please describe the problem in a bit more details?

[Regnor]

We had to disable CBT for the multi writer disk in the VMX. Although the disk is independent and excluded from the job Veeam reenables CBT. To solve this we unchecked the CBT option in the storage section of the job settings.

[foggy]

Does enabling CBT affect anything negatively?

[Regnor]

If you need concurrent access from different VMs which run on different hosts you'll have to disable CBT. Else the second VM will fail to power on.
https://kb.vmware.com/s/article/2110452

[foggy]

Got it, thanks for the clarification. Veeam B&R enables CBT for the entire VM anyway, so disabling it in the job seems to be the only workaround.

[Regnor]

We initially thought that Veeam wouldn't touch independent or excluded disks.
It's only a smaller issue and probably rather rare so I don't think anything needs to be changed.