Backup / Backup copy conflicts

[perjonsson1960]

Folks,

I have just started the first backup copy of our 17.5 TB file server cluster. This will take a while. What happens if the backup copy job is still running when the corresponding backup job starts? Will the backup job wait until the backup copy is finished? Or will the backup job cancel the backup copy job? In order to be sure that the backup copy job finishes, should I disable the backup job?

PJ

[foggy]

Hi PJ, it depends on the source job backup method - it will start normally (in parallel with the backup copy) in case the method is forward incremental and will terminate the backup copy in case it is reverse incremental (backup job has higher priority than backup copy). It all comes down to whether it needs access to the same files.

[perjonsson1960]

It is a forward incremental agent backup job, backing up our file server cluster as a failover cluster.

[foggy]

Then there shouldn't be any conflicts.

[perjonsson1960]

Okay. And the backup copy job will disregard the newly created restore point until the next backup cycle? It is a "Periodic copy" job.

[foggy]

During the next interval, it will sync the latest VM state available on the source. Btw, make sure you've temporarily extended the backup copy interval to allow for the initial sync to complete normally.

[perjonsson1960]

And how do I extend the backup copy interval of a running job? The new copy interval was started at 12:00 noon and is 24 hours, so hopefully that will be enough, but who knows...

[foggy]

It will continue from where it left off during the next interval anyway. The only downside is that you will have a warning and an incomplete backup file but the chain will finally heal itself.

[perjonsson1960]

Thanks!

[perjonsson1960]

Well, saying that it will continue where it left off is stretching the truth a bit, because it re-reads ALL the data from the disks that it was copying when the copy interval expired... In this case it means that it will have to re-read approx. 16 TB of data, 8 from one node and 8 from the other... And it had read approx. 6 TB from each when the interval expired...

I never imagined that the backup copy would take this long (way over 24 hours even if it hadn't been interrupted) when the initial full backup of the cluster took only nine hours...

[perjonsson1960]

Sorry, I spoke too soon... It read only about 2.5 TB from each disk, so no worries! It really continued where it left off, as you said, foggy! Good work, B&R!

[foggy]

Thanks for the timely update - I almost started thinking I was missing something.

[perjonsson1960]