Pre-/Post-Commands not working properly

[umatz]

Hi,

I was trying to make use of the pre/post-send commands and run into a problem. My configuration looks like this:

dst_a_plan      = 30days=>1day,4weeks=>1week,6months=>1month
dst_a_precmd    = etherwake 00:30:18:a0:11:f9
dst_a_pstcmd    = ssh root@dicky pm-suspend
enabled         = on
mbuffer         = /usr/bin/mbuffer
mbuffer_size    = 1G
post_znap_cmd   = off
pre_znap_cmd    = off
recursive       = off
src             = tank/sysbackup
src_plan        = 7days=>1day
tsformat        = %Y-%m-%d-%H%M%S
zend_delay      = 0

Apart from the issue that the post command in its current form will never return (which I will have to take care of), I run into the problem that the pre- and post-send commands are each run twice. I believe this is an error. After a short look into znapzend/lib/ZnapZend.pm I noticed that both commands are already run in the function(?) refreshBackupPlans and then again in sendRecvCleanup. I am no perl programmer, but I believe that the handling of Pre-/post-send commands should be removed from refreshBackupPlans.

Here is the output from running znapzend with debug logging

[Wed Oct  3 16:48:01 2018] [info] znapzend (PID=19616) starting up ...
[Wed Oct  3 16:48:01 2018] [info] refreshing backup plans...
[Wed Oct  3 16:48:04 2018] [info] **running pre-send-command for root@dicky:tank/sysbackup**
[Wed Oct  3 16:48:04 2018] [info] found a valid backup plan for tank/sysbackup...
[Wed Oct  3 16:48:04 2018] [info] **running post-send-command for root@dicky:tank/sysbackup**
^C[Wed Oct  3 16:48:16 2018] [info] znapzend (PID=19616) initialized -- resuming normal operations.
[Wed Oct  3 16:48:16 2018] [debug] snapshot worker for tank/sysbackup spawned (24125)
[Wed Oct  3 16:48:16 2018] [info] creating snapshot on tank/sysbackup
# zfs snapshot tank/sysbackup@2018-10-03-164816
[Wed Oct  3 16:48:16 2018] [debug] snapshot worker for tank/sysbackup done (24125)
[Wed Oct  3 16:48:16 2018] [debug] send/receive worker for tank/sysbackup spawned (24127)
[Wed Oct  3 16:48:16 2018] [info] starting work on backupSet tank/sysbackup
[Wed Oct  3 16:48:16 2018] [info] **running pre-send-command for root@dicky:tank/sysbackup**
[Wed Oct  3 16:48:16 2018] [debug] sending snapshots from tank/sysbackup to root@dicky:tank/sysbackup
# zfs list -H -o name -t snapshot -s creation -d 1 tank/sysbackup
# ssh -o batchMode=yes -o ConnectTimeout=30 root@dicky zfs list -H -o name -t snapshot -s creation -d 1 tank/sysbackup
# ssh -o batchMode=yes -o ConnectTimeout=30 root@dicky zfs list -H -o name -t snapshot -s creation -d 1 tank/sysbackup
[Wed Oct  3 16:48:25 2018] [debug] cleaning up snapshots on root@dicky:tank/sysbackup
# ssh -o batchMode=yes -o ConnectTimeout=30 root@dicky zfs destroy tank/sysbackup@2018-10-03-162751
[Wed Oct  3 16:48:25 2018] [info] **running post-send-command for root@dicky:tank/sysbackup**
^C[Wed Oct  3 16:48:31 2018] [debug] SIGINT received.
[Wed Oct  3 16:48:31 2018] [info] terminating znapzend (PID=19616) ...
# zfs list -H -o name -t snapshot -s creation -d 1 tank/sysbackup
[Wed Oct  3 16:48:31 2018] [debug] cleaning up snapshots on tank/sysbackup
# zfs destroy tank/sysbackup@2018-10-03-162751
[Wed Oct  3 16:48:31 2018] [info] done with backupset tank/sysbackup in 15 seconds
[Wed Oct  3 16:48:32 2018] [info] znapzend (PID=19616) terminated.

[Beneter]

After reviewing the sourcecode I can confirm, that refreshBackupPlans is executing the pre and post commands. But as far as I can see this is only being done when starting the daemon (probably for immediate feedback if something fails). When the daemon is running refreshBackupPlanswill not be invoked until restart.

[oetiker]

or when it gets a HUP signal ... this is intentional.

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.