Dear community,
a strange phenomenon appears here in my multi-master-replication
set-up with bucardo in respect to 'onetimecopy' that drives me insane.
I start off with writing the dump of the primary master to the other
master data bases, add the sync as peer-to-peer
(primary-master-to-other-master, another sync
other-master-to-primary-master, all kicks on demand). Since the
primary master is 24/7 online, I want to initially update the data
contents of the other-masters with a 'onetimecopy=1'. Conflict
management is 'bucardo-source', no delta is set, each dbgroup is
between primary master and the other-masters, stayalive and kidsalive
is false since my application controls the kicks, status for all syncs
is default inactive.
Sometimes a onetimecopy is carried out, sometimes not and the
onetimecopy-flag in the sync table is still 1 after sync.
My application controls the kicks by sending the command sequence via
ssh 'bucardo update sync SYNCNAME status=active onetimecopy=1; sleep
10; bucardo activate SYNCNAME; sleep 10; bucardo validate SYNCNAME;
sleep 10; bucardo kick SYNCNAME 1200; bucardo deactivate SYNCNAME;
bucardo update sync SYNCNAME status=inactive; exit
During the app-initiated sync the primary-master and the
'master-partner' are permantently online, after the sync the
other-master is offline.
In the verbosed bucardo log I could not find a hint why onetimecopy is
not carried out (instead a kind of delta sync is made).
Any ideas about that?
Cheers.
_______________________________________________
Bucardo-general mailing list
[email protected]
https://bucardo.org/mailman/listinfo/bucardo-general
