Error Importing .VBM Manifest on Repository Rescan

[ericjamesWL]

Open case ID: 01802506

My issue started when I noticed that GFS was not properly rolling off old restore points in one of my backup copy jobs. The GFS flags for each point in the chain looked something like this from newest to oldest:

(no flag)
(no flag)
W
W
W
M
R
R
W
R
MQ

I opened a case because I was concerned that the “R” points were not being removed, even after several fulls had occurred. The support engineer found this concerning as well and said this shouldn’t happen. He advised to manually delete the “R” points because they were no longer needed since there had subsequently been multiple fulls. He walked me through manually deleting them using the Veeam FILE browser pane. After this we did a repository rescan, then went into the properties of the GFS restore point chain and removed the deleted points from the configuration.

That night the backup copy job successfully copied the latest restore point but threw an error upon completion: “Failed to generate points Error: Sequence contains no matching element.” So it seems like the job had trouble tying the newly copied restore point to the GFS chain. Perhaps those “R” points were still somehow being used in the chain.

The engineer then advised running an active full, which I did, but the same behavior was observed: restore point was copied but job returned the error above. He then wanted me to try to disable GFS on the job for one night, let the job run, then re-enable it. Unfortunately, the job had failed enough times that it was automatically disabled by Veeam, and I could not re-enable it without a repository rescan. I don’t remember the exact error here. I think it said something about the repository being out of sync with the database. In any case, after running a rescan of the repository it failed with the error: “Failed to import backup path /home/ExagridVeeamDM/VEEAM/foldername/filename.vbm Details: The given key was not present in the dictionary.”

At this point I removed the GFS chain of restore points from the configuration and tried a rescan but received the same error. So I renamed the folder and let the backup copy job run that night to create a new folder with an entirely new GFS chain. The next day I ran another rescan to try to import the old chain into the config but received the same error. The engineer suggested deleting the .vib files from the folder, which I did (there were 3 .vib points for each server in the job). I ran a rescan and it still failed with the same error.

Has anyone else seen this error ("The given key was not present in the dictionary")? Anyone have any suggestions?

[foggy]

Eric, this looks like a VBM file corruption/going out of sync with the database. I suggest to let the new job do its job and keep older GFS points somewhere aside (once you need to restore from them) until the new job creates the required number of restore points and you could manually delete the older ones.

[bjorn.patterson]

Hi Eric, Foggy,

Have you managed to get around/fix this error?

I'm also having the same problem when moving backups to different repositories, importing backups etc. It just seems to be a few vms/backups that have the problem, others are fine.

Is there anyway to fully resync the database so the key become "present in the dictionary"?

Thanks,
Bjorn

[foggy]

Hi Bjorn, from Eric's case I can see that the issue was resolved, however, it doesn't contain any details of resolution. So you can PM him directly and contact support in the meanwhile.