Get pvesr schedule-now to wait until everything is replicated

[redmop]

Is this possible? If not, can we add a --wait paramater to do this? Perhaps a --retry x to try x number of times?

 

[fiona]

Hi,
does using pvesr run --id <JOB-ID> instead work for you? Otherwise, please explain the use case for this in more detail.

 

[redmop]

I want to run a script to backup to an external drive. Nodes 1 and 2 hose running vms. They both replicate to node 3. Node 3 does zfs send to the external. In the script, I want to force replicate now, then wait until the replication is done before doing my zfs send to the external.

 

[fiona]

Is there a problem with using pvesr run --id <JOB-ID> then? AFAICS that command only returns after finishing the replication.

 

[redmop]

pvesr run --id 117-0
trying to acquire lock...
can't lock file '/var/lock/pvesr.lck' - got timeout

It works as long as no other replication is currently running.

 

[fiona]

You're right. What you could try is using

pvesh get nodes/<NODE>/replication/<JOB-ID>/status

to get the current last_sync timestamp. Then run schedule-now and in a loop with a bit of pause do

pvesh get nodes/<NODE>/replication/<JOB-ID>/status

again to monitor the last_sync value. When it changes, you know the job has completed.

Of course you can also check the fail_count and conditionally abort the loop.

 

[redmop]

I did it kinda like that, though a wait and retry would be much much cleaner.