NAS backup: Failed to transform metadata

[jeff.richeson]

Case # 04568008

My configuration is simple: I have a single Veeam server (build 10.0.1.4854 P20201202) with all the roles and an additional Linux repository using NFS on the same subnet. I have been using this configuration for several months but have recently tried to use the NAS backup

I am trying to use the NAS backup to backup about 3TB of files to the Linux repository over NFS. The job appears to transfer all the data but fails every time at the very end as part of the copy/transform metadata step. The error is "Agent: Failed to process method {NasMaster.ReplicateMeta}: NFS4 status code: [0]". The tech points to network issues since logs show a connection issue to the NFS repository. The same issue happens every time the job is run, including when doing a health check on the failed job. I have a hard time believing that the same network issue occurs every time this job is run. There seem to be no way to recovery as no matter what I do with the job, it continues to fail. If I make a new job with a small subset of the NAS files, the job seems to complete fine. If I have the job use the local drive repository the job seems to complete fine. (I can't afford to have the these backups use my expensive SAN disk.) Has anyone run across this issue?

.

[HannesK]

Hello,
from the logfiles, support summarized you have a network issue or not enough resources. I did not see in the case how much RAM & CPUs your components have.

The question is, whether that whole setup could be simplified and maybe solved indirectly. As you say that you have a Linux repository that has an NFS share mounted... what do you think about just removing that Linux server and write to the NFS share directly? https://helpcenter.veeam.com/docs/backu ... ml?ver=100

If I make a new job with a small subset of the NAS files, the job seems to complete fine.

that sounds like a resource issue.

Best regards,
Hannes

[jeff.richeson]

The Veeam server is virtualized and has 2 vCPUs and 20GB of memory. I have a small infrastructure so I don't want to dedicate too much of it to backups.

I likely misstated how I was using the Linux repository. That Linux machine is the NFS server and I am writing directly to NFS. Just to try things, I added the Linux machine as a managed server and created a new repository on the same disk but a different path. This job just actually completed successfully. I guess I can have this one job use the Linux repository and all the other jobs use the NFS directly but it seems like I should be able to have everything go directly over NFS.

[HannesK]

2 vCPUs is undersized for everything... if you count 1-2 CPU for Windows, then you have literally nothing left for backup server, SQL, proxy (repository)... please see the system requirements: https://helpcenter.veeam.com/docs/backu ... ml?ver=100

I likely misstated how I was using the Linux repository. That Linux machine is the NFS server and I am writing directly to NFS.

I recommend to write directly to the Linux machine as a Linux repository (with XFS for block cloning recommended): https://helpcenter.veeam.com/docs/backu ... ml?ver=100. As far as I see, you already did that.