[Pgpool-general] vacuum fails due to toast table difference

[Lou Fridkis]

I am using postgres 8.4.1 & pgpool 2.3.3 with:
    replication_mode = true
    load_balance_mode = true
    replicate_select = false
    reset_query_list = 'ABORT; DISCARD ALL'
    master_slave_mode = false
    parallel_mode = false

I have a script that runs every 10 minutes, from the cron, which compares
the row counts and sequence numbers of the primary and secondary DBs to
verify that they are synchronized. The problem is that pgpool is hanging.
This happens about once every 2 weeks, but with no apperently reqular
pattern. When pgpool hangs, it simply fails to respond to a connnection
request. There is no error message either on the command line or in the
pgpool log. There is also no error message in the postgres log. I have put
both pgpool and postres into debug mode, but still no error messages or any
messages that indicate what the problem is. A ps command typically show
about 10 connections running the cron script. Also, there will be the
normal, expected connections from the application, but these are not a large
number of connections. The top command does not show any unusual activity.
Also, performance history from ganglia does not show any unusual activity. I
also checked dmesg, /var/log/messages and /var/log/cron and did not see
anything unusual.

My question is: Does anyone have any suggestions of what I can check or do
to figure out why this is happening and prevent it from happening in the
future?

Thanks,
Lou ...
P Please consider the environment before printing this e-mail or any
attachments.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://pgfoundry.org/pipermail/pgpool-general/attachments/20110510/ee0ac590/attachment.html>