Issues after replacing ESXi hosts

[FrancWest]

Case #05779390 & #05779983

This week, we replaced all ESXi hosts in our vCenter cluster with new hardware. Using vmotion this worked perfectly. However, Veeam B&R has issues with this:

- The virtual lab is broken. Although the VM is still present, we can't use it anymore. First it was not visible within Veeam and when I tried to manually connect to it, it reported that it can't add the virtual lab since it already existed in the database. Probably because the host it was residing on doesn't exist anymore. I had to remove all surebackup jobs and then was able to remove the virtual lab using powereshell. However, when connecting it manually again it now reported that the configuration was corrupt. According to support I have to re-create the lab, but I don't know the settings anymore. Support is able to pull these settings from the sure backup job log they say so I can re-create it a bit easier. However, I still have to recreate the sure backup jobs.

- When opening the properties of the replication jobs, they all error out with a 'object reference not set to an instance of an object'. After that all resources (hosts, datastore, VM folder, etc) on the destination tab are empty. I guess this is because the replication job still points to the old ESXi host which doesn't exist anymore.

- Jobs failed at first since the new hosts could not be licensed from Veeam because the licenses were still being used by the old, non-existing, hosts. I had to revoke the licenses from the old hosts and after that the backups started working again and the new hosts were licensed. Why doesn't Veeam remove licenses for hosts automatically when it detects that a hosts doesn't exist anymore in vCenter?

All in all it appears that Veeam doesn't like it when hosts are being removed from vCenter. It would be nice to have some remapping feature and also much clearer error messages.

[HannesK]

Hello,

1) correct, VLABs must stay on the same host. That's a limitation.

2) correct, if somebody points to a host, that does not exist anymore, then this causes issues. Pointing to a cluster would prevent that.

3) correct, if you are on socket licensing, then you need to take care of the sockets (revoke license from non-existing host). Removing something automatically could also be disliked by other customers. The scenario is "well documented" and easy to solve. Alternatively, the Universal License would also solve it, because it's bound to the workload instead of the host.

Except for scenario 1, the software is totally fine with removing hosts.

Best regards,
Hannes

[FrancWest]

Hi,

2) can we point to a cluster even if we have vSphere standard?

And I wouldn’t says the software is ‘totally fine’ with removing hosts. An object reference error is not an error that shows the software is expecting a totally fine situation

[HannesK]

Hello,
2) clusters are mentioned in the product edition comparison for all editions, so I see no reason why it would not work. Did you see problems?

An object reference error is not an error that shows the software is expecting a totally fine situation

that error shows, that somebody removed the host, which the software expects. Not sure, how a software could handle it better.

Similar example: If one replaces my favorite beer with a bad tasting beer, then I would also throw an error like that. The new beer is still beer, but not the same

Best regards,
Hannes

[siranee.ja]

Hi FrancWest,
My case is Case # 06091941
One of my esxi cluster not responding from vcenter and I had to end with remove and re-add the esxi to my vcenter.
I have the same problem with you as your posted:
"I have to re-create the lab, but I don't know the settings anymore."
"Support is able to pull these settings from the sure backup job log they say so I can re-create it a bit easier. However, I still have to recreate the sure backup jobs."

Could you share with me? My support always say that it is impossible to get virtual lab configuration.

Best Regards,
Siranee Jaraswachirakul

[TWuser]

This is likely too late, but we recently had a host die on us, causing our datalabs to break. I wasn't 100% confident I had printscreens of all settings, and it took a fair amount of time to set them up correctly.
Support helped me edit the VMX file of the datalab VM, and change the HostRef and VMref to match the new.
After that, I was able to edit the datalab in Orchestrator. The lab still wouldn't start, but at least I was able to edit and get printscreens.

Note - the VMX file seems to contain most, if not ALL of the datalab settings. It's hard to read, so I put it into a spreadsheet so I could get massage it into columns.

After that, I manually rebuilt my datalab in the VBR server. Then in Orchestrator I used reports from previous Datalab runs in order to help me remember all the Plan Steps I had on each Datalab VM (I had 2 DC's, a load balancer, and a test VM).

I am going to start backing up the Datalab VM itself, hoping I can recover it using Veeam next time.