I'll add 2 new empty XFS reflink extents to our current SOBR, the old extent will be reused later as copy target. I could just seal the existing extent and all jobs will start with an active full to the new ones. As we have ~20 jobs and several hundred TB of data this would put a very high load on our production storage. So I'd like to do this somehow staggered. The only way I can think of is to seal the old extent and manually run an active full for a subset of 5 jobs, then unseal the extent again, so that other jobs can use it again. Would this work?
If I understand the placement policy correct backups for jobs with existing chains on old extent will just continue to use this extent. Even if the new ones are empty. Is it sufficient to seal the extent until the active full's started to run or do I have to keep them sealed for the time that the jobs runs? I think the decision about the storge is done at the beginning of a job for all VM's.
- Availability of extents on which backup files reside. If some extent with backup files from the current backup chain is not accessible, Veeam Backup & Replication will trigger a full backup instead of incremental (if this option is enabled). For more information, see Adding Backup Repository Extents.
- Backup placement policy set for the scale-out backup repository.
- Load control settings — maximum number of tasks that the extent can process simultaneously.
- Amount of free space available on the extent — the backup file is placed to the extent with the most amount of free space.
- Availability of files from the current backup chain — extents that host incremental backup files from the current backup chain (or current VM) have a higher priority than extents that do not host such files.
So the decision about the storage extent is done when a VM task in a job is stared. Which means we can unseal the old extent only after the job that should be placed on the new extents are completely done.
Yes, you are right. The active full goes to the new active extend. My chosen words weren’t really good
1) You put the extend into sealed mode
2) You run the five jobs. An active full will be made.
3) You get the extend out of sealed mode
4) You run all jobs with normal schedule
If i remember correctly, you could just manually move the newest file of each job to new extent (and the rescan SOBR),then the job will use repository with the last file. During synthetic processing it'll the create new VBK on the new extent.
In the end we just sealed the old extents and run all jobs as active full on this day. Maybe I try to move a vib of a small job to the other extent as see what happens, would be good to know for next time or to load balance between extents/jobs at some point. Currently the two active extents are used 50/50.
