Linux SOBR memory usage

[rjv1971]

Have a backup copy job storing backups locally to a (performance tier) backup repository, then an immediate offload to Azure (capacity tier).
The repository is a virtual (vmware) server), RHEL and has 64GB RAM / 32GB swap. I have the load control on the performance tier backup repository set to 8 concurrent tasks.

What I see happening is that 1 backup copy job's offload process, containing 176VMs, is taking up all the 64GB memory plus about 25GB of swap space.

Is this by design?

What is the purpose of load control? I have it set to 8, which means (I think) that even though many backup jobs can run at the same time... within all those jobs only 8 backups at a time will be processed to be copied to the performance tier of the SOBR. If there are more backups they will wait. But what about the offload job that gets submitted after the backup copy job finishes? From my observation, more than 8 VMs were uploaded at the same time, with each offload creating a process on the linux performance tier server, which exhausted memory. A lot of jobs seemed 'stuck' at 99%. Is it supposed to work like that?

[HannesK]

Hello,
it sounds a bit too much RAM consumption and I suggest to check with support. Please post the case number for reference.

For capacity tier offloads, there is a concurrency limit in the object storage settings that might be useful https://helpcenter.veeam.com/docs/backu ... ml?ver=120

Best regards,
Hannes

[rjv1971]

Thanks for your reply! I do have a ticket: Case #06408916. I did check the setting you mentioned and it was unchecked; I have set it to 16, see how it goes tomorrow morning.