One of the things I like to do when testing out software, is to mimic customer behaviour and sometimes even deliberately act or do something not well thought. The latter is mostly based on true stories and/or events .
At the moment I'm further testing the capabilities and features of Veeam in combination with vCloud Director (and vSphere if I'm being honest), how it all interacts and which problems surface in a default setup, without any fancy configuration.
Well, I'm not even halfway and I've already gotten myself into situation where it seems that I can't get myself out of it. The situation (consists of two problems) is reproducible, so if anyone wants to try it, I've written the steps below. However, before I go any further, I would like to mention our environment:
- 1. Veeam version 184.108.40.2062
- 2. Veeam-environment installed on Windows Server 2016
- 3. Running vCloud Director 9.1
- 4. Running vSphere 6.5U1
Coming back to the situation; I've got the impression that Veeam tries to backup a non-existing VM (problem #1) and because of that the backup-job never ends with the status "SUCCESS". Because of that, Veeam starts the same job every-time as a FULL backup (problem #2)
The situation I'm dealing with can be reproduced by using the following steps:
- 1. Created a daily backup-job in VBR. I didn't do anything fancy in the "advanced settings" of the backup-job, except for a synthetic full each Sunday
- 2. Added a tenant to the job, so that every vApp and VM will be backed-up automatically.
- 3. Let the scheduler start things automatically. The first full backup and all incrementals afterwards = 100% SUCCES.
- 4. The day before the synthetic full, I did the following:
a. Renamed a VM "veeam" to "veeam-old"
b. Restored that VM ("veeam") to vCloud Director with the original name, namely "veeam"
c. At this point I've got two identical VMs sitting in the same vApp, the restored one called "veeam" and the original renamed one "veeam-old"
d. After checking that the restored VM was OK, I deleted the renamed ("veeam-old") VM usin vCloud Director. From this point on the renamed VM was deleted from disk.
- 5. Let the scheduler start the 'synthetic full' backup-job automatically. All VMs went OK, except for the fact that:
a. Veeam lists and tries to backup the old renamed and afterwards deleted VM "veeam-old". This results in a FAILED job.
b. As a cascaded effect, because the 'synthetic full' backup-job failed, Veeam is ever since starting the job as an active full.
Personally I think that Veeam stopped making incrementals, because it needs/wants a 100% successful backup job first. However, because Veeam tries to backup a non-existing VM, we seem to be in a continuous loop.
My question to all the Veeam-enthusiasts, how do I tell Veeam that it doesn't need to backup non-existing virtual machines? How do I fix this? Are my assumptions above correct?