Marco Aurelio V. da Silva wrote: > Dears, > > After stop the comunication between master and slave for hours, the > replication stop and log file have many lines with: > > 2007-03-01 08:45:39 BRT DEBUG2 syncThread: new sl_action_seq 1 - SYNC 21205 > 2007-03-01 08:45:39 BRT DEBUG2 localListenThread: Received event 2,21205 > SYNC > 2007-03-01 08:45:49 BRT DEBUG2 syncThread: new sl_action_seq 1 - SYNC 21206 > 2007-03-01 08:45:57 BRT DEBUG2 localListenThread: Received event 2,21206 > SYNC > 2007-03-01 08:45:59 BRT DEBUG2 syncThread: new sl_action_seq 1 - SYNC 21207 > 2007-03-01 08:46:05 BRT DEBUG2 localListenThread: Received event 2,21207 > SYNC > 2007-03-01 08:46:09 BRT DEBUG2 syncThread: new sl_action_seq 1 - SYNC 21208 > 2007-03-01 08:46:13 BRT DEBUG2 localListenThread: Received event 2,21208 > SYNC > 2007-03-01 08:46:19 BRT DEBUG2 syncThread: new sl_action_seq 1 - SYNC 21209 > 2007-03-01 08:46:21 BRT DEBUG2 localListenThread: Received event 2,21209 > SYNC > 2007-03-01 08:46:29 BRT DEBUG2 syncThread: new sl_action_seq 1 - SYNC 21210 > 2007-03-01 08:46:29 BRT DEBUG2 localListenThread: Received event 2,21210 > SYNC > > have solution for prevent this ? > > i don´t use pgpool, or perltools, i use the basic script for replication. > i don´t use failover or switchover, these options are recommended in this > cases ? > >
Well, the fact that you don't have two log files suggests to me that one of the slon processes wasn't running. If *that* is the case, then replication will have a tough time proceeding, because both slon processes need to run in order to get replication to work. - There needs to be a slon running against the node(s) originating replication sets, otherwise nothing marks SYNC events on that node, and subscriber nodes won't know what data to pull from the origin. - There needs to be a slon running against each subscriber, as that is the only way that changes will be pulled and applied. It seems to me that one of your slon processes fell over, and replication will naturally stop at that point. _______________________________________________ Slony1-general mailing list [email protected] http://gborg.postgresql.org/mailman/listinfo/slony1-general
