[Slony1-general] unable to remove stale pg_listener entries

[Victoria Parsons]

Hi all,

 

I have a replicating system of just two nodes. The master is on 10.112.0.115, the slave is the one running below. The system was replicating fine on 6th Jan. Then the slave (the node below) was shutdown with the power off button, and so didn't have a chance to tidy up after the slon daemon. You can see from the log that there is no "shutdown requested" message or any clean up done.

 

When it was restarted on 9th Jan and 12:55 (master is already up and running at this point) if fails to delete the old pg_listener entries and so terminates. This repeats every time the slon daemon is started - always trying to remove stale pg_listener entries for pids 2733, 2626 and 2627. I double checked these processes were dead, then I went into postgres and manually deleted all the pg_listener entries for these 3 pids. When I restarted the slon daemon again it worked fine.

 

I can see from the log that slony has tried to delete these old pg_listener entries. What I don't know is why it failed. I am running slony with the superuser "postgres". It was this user I manually deleted the entries with, so I know its not a permissions problem. Is this a bug? I don't understand the bit of the error that says "ERROR:  tuple already updated by self" To avoid it happening again I am taking the precaution of writing a script that removes these entries before starting the slon daemon. I am interested as to what failed though - does any one have any thoughts?

 

By the way I am using 1.1.0 so if this has already been seen and fixed in a new version I apologise.

 

Thanks,

Vicki Parsons

 

2006-01-06 17:32:08 GMT DEBUG1 cleanupThread:    0.068 seconds for cleanupEvent()
2006-01-06 17:32:08 GMT DEBUG1 cleanupThread:    0.056 seconds for delete logs

2006-01-09 12:55:01 GMT CONFIG main: slon version 1.1.0 starting up
2006-01-09 12:55:01 GMT CONFIG main: local node id = 1
2006-01-09 12:55:01 GMT CONFIG main: launching sched_start_mainloop
2006-01-09 12:55:01 GMT CONFIG main: loading current cluster configuration
2006-01-09 12:55:01 GMT CONFIG storeNode: no_id=2 no_comment='CSG 10.112.0.115'
2006-01-09 12:55:01 GMT CONFIG storePath: pa_server=2 pa_client=1 pa_conninfo="dbname=statusdb1 host=10.112.0.115 user=postgres" pa_connretry=10
2006-01-09 12:55:01 GMT CONFIG storeListen: li_origin=2 li_receiver=1 li_provider=2
2006-01-09 12:55:01 GMT CONFIG storeSet: set_id=1 set_origin=2 set_comment='all status tables'
2006-01-09 12:55:01 GMT WARN   remoteWorker_wakeup: node 2 - no worker thread
2006-01-09 12:55:01 GMT CONFIG storeSubscribe: sub_set=1 sub_provider=2 sub_forward='f'
2006-01-09 12:55:01 GMT WARN   remoteWorker_wakeup: node 2 - no worker thread
2006-01-09 12:55:01 GMT CONFIG enableSubscription: sub_set=1
2006-01-09 12:55:01 GMT WARN   remoteWorker_wakeup: node 2 - no worker thread
2006-01-09 12:55:01 GMT CONFIG main: configuration complete - starting threads
2006-01-09 12:55:01 GMT DEBUG1 localListenThread: thread starts
NOTICE:  Slony-I: removing stale pg_listener entry for pid 2726, relname _statusdb00e081641f02_repcluster_Restart
NOTICE:  Slony-I: removing stale pg_listener entry for pid 2733, relname _statusdb00e081641f02_repcluster_Node_1
NOTICE:  Slony-I: removing stale pg_listener entry for pid 2727, relname versions_update
NOTICE:  Slony-I: removing stale pg_listener entry for pid 2727, relname eventlog_insert
NOTICE:  Slony-I: removing stale pg_listener entry for pid 2726, relname _statusdb00e081641f02_repcluster_Event
NOTICE:  Slony-I: removing stale pg_listener entry for pid 2726, relname _statusdb00e081641f02_repcluster_Event
2006-01-09 12:55:02 GMT FATAL  localListenThread: "select "_statusdb00e081641f02_repcluster".cleanupListener(); listen "_statusdb00e081641f02_repcluster_Event"; listen "_statusdb00e081641f02_repcluster_Restart"; " - ERROR:  tuple already updated by self
2006-01-09 12:55:02 GMT DEBUG1 slon: shutdown requested

This message should be regarded as confidential. If you have received this
email in error please notify the sender and destroy it immediately.
Statements of intent shall only become binding when confirmed in hard copy
by an authorized signatory.

_______________________________________________
Slony1-general mailing list
Slony1-general@gborg.postgresql.org
http://gborg.postgresql.org/mailman/listinfo/slony1-general