replica job doesn't delete removed disk

[mcz]

Hi veeam-guys,

had a strange situation last week... It all begun when I added a second disk to one of my windows servers to do an upgrade. When the upgrade was done, I deleted the disk again, but during that period obviously a replication job took place. So far so good, nothing wrong here.

Last week I did a planned failover, everything went well but during failback the job failed and told me that there's a disk missing on the original vm. I was surprised that the second disk on my replica didn't get removed automatically and so I was stuck. The trick was to add an empty disk to the original disk, let the failback continue and remove it afterwards.

Is this (somehow) expected or a bug? I'm currently using v 10 (all updates included).

Thanks!

[Egor Yakovlev]

Hi Michael,

From the first glance it does look odd to me. Investigating the reasons...
Update: with Restore(Failback) to Original VM we do not overwrite existing VM(aka .vmx) but push disks data only. That is why Replica config .vmx(with existing disk in its configuration) is not populated back to original VM, so we give error about the non-existent disk. It is a by design behavior and your workaround is the way to go.

/Cheers!

[mcz]

Thanks Egor! Yeah I'd agree on the failback but maybe not on the replication part: If I delete a disk on my source-vm, shouldn't this change be replicated to the target-replica-VM? I'd say the disk should get deleted there as well!

Please let me know your thoughts - thanks!

[foggy]

The VMX file should be updated with the new one during the next replication job run. Any chance the job just didn't run since it grabbed the temp disk and prior to a failover?

[mcz]

Hi foggy/Alexander,

no, there were many (successful) replication runs in between. I know that for sure because I've almost forgotten that disk and was a bit confused that it still was there (after some weeks and 3 replica passes per day)...

[Egor Yakovlev]

After a walk with test labs and discussion with QA, this appears to be a currently expected behavior.
There are close to none requests to alter it, so the change has very low priority, however I will raise the question and see if we can alter this case, as it does contradict with some general data replication concepts.

/Thanks!

[mcz]

Hi Egor, thanks for your message & the research. I understand but at the same time I'm confused as well. With the current approach, the target VM would start to look and behave different then the source VM... Also it can lead to troubles/confusion on the user-side when an old disk re-appears that shouldn't be there (like it was in my case)...

Thanks!

[dbeerts]

Hello,

I have an active replication of a VM and was wondering what will happen when I change the source VM hardware settings. Will these changes replicate to the destination VM?
In my specific case I removed the 'SATA Controller AHCI' from the source VM but at the destination VM it keeps present. Is this normal behavior?
This would mean that hardware config is replicated once at initial run? If you change it afterwards you would have to change it manually or recreate the target VM?

Thanks in advance!
Dries

[veremin]

Currently this is a normal behaviour. Manual update of the configuration file won't help, so starting replica anew is a way to go in such cases. Thanks!

[foggy]

To clarify (as it might seem that Veeam B&R doesn't replicate hardware changes at all), hardware additions and changes (except for the disk size change) are replicated to the target VM, however, if something gets deleted on the source VM, this change will not get replicated. For example, the target VM will keep the removed disk in its configuration, it just will not be synced anymore.

[BrianS_WincVA]

I have a production VM that is being replicated. After replication was configured and working, a second disk (.vmdk) was created and added to the VM for a short-term requirement. It is now no longer needed and has been removed from the production VM (and deleted from the datastore). The replication job is set to retain 4 replicas, so after the 5th replication, I was expecting the second disk to disappear, but it has not. Veeam is correctly replicating the VM otherwise, the job shows only one disk being replicated and there are no errors.

So now the question is how can I remove the vmdk from the replica without breaking the snapshot / replication chain? I know if I delete all replicas from disk and re-run the job, it will fix the issue but I want to avoid having to send multiple TB of data again which is already at the DR site anyway.

Anyone ran into this scenario before? Thanks.

[Mildur]

Hello Brian

I merged your request with this existing topic.
Unfortunately, starting a new replica of this VM from scratch is required.

Two workarounds to not have to replicate all TBs again:
If you have backups of that VM in your DR site, you may use replica seeding.

Or you can try to remove the replica from the configuration, remove all snapshots on the replica VM, remove the VMDK from the replica and then start a new replica and use mapping option with existing replica VM. First run will probably take more time to calculate the difference between source and replica VM.
https://helpcenter.veeam.com/docs/backu ... ml?ver=120

Best,
Fabian

[BrianS_WincVA]

Fabian,

Thanks for your reply. I just tried your solution in my lab and it works well. This is also helpful to know in case I ever need to move a VM to a different replication job.