[Pgpool-general] pool_read2: read failed error

[Gnanakumar]

Any diagnosing guidance/recommendation on this would be greatly appreciated.

-----Original Message-----
From: pgpool-general-bounces at pgfoundry.org
[mailto:pgpool-general-bounces at pgfoundry.org] On Behalf Of Gnanakumar
Sent: Tuesday, August 24, 2010 1:14 PM
To: pgpool-general at pgfoundry.org
Subject: [Pgpool-general] pool_read2: read failed error

Hi,

Our production server is running PostgreSQL 8.2.3 and pgpool-II-2.0.1 on
RHEL 5.

Recently, we're facing an issue with pgpool.  The issue is suddenly the
application stops working/responding and on checking pgpool log, there is a
pool_read2: read failed error.  We're facing this for the 3rd time in our
production server.  But when I 'ps -ef |grep pgpool', I can see all the
pgpool processes are running.  Also, I'm able to login to psql on port 5432
and able to run SQL commands at that moment.  pgpool log statements snippet
shown below:

2010-08-10 11:18:19 LOG:   pid 29573: connection received:
host=192.168.0.200 port=58475
2010-08-10 11:18:19 LOG:   pid 19569: YYYYY
2010-08-10 11:18:19 LOG:   pid 20151: YYYYY
2010-08-10 11:18:19 LOG:   pid 19569: YYYYY
2010-08-10 11:18:19 LOG:   pid 19569: YYYYY
2010-08-10 11:18:19 ERROR: pid 9911: pool_read2: read failed (Connection
reset by peer)
2010-08-10 11:18:19 LOG:   pid 9911: notice_backend_error: 0 fail over
request from pid 9911
2010-08-10 11:18:19 LOG:   pid 20794: connection received:
host=192.168.0.200 port=46105
2010-08-10 11:18:19 LOG:   pid 4686: starting degeneration. shutdown host
127.0.0.1(5432)
2010-08-10 11:18:19 ERROR: pid 4686: failover_handler: no valid DB node
found
2010-08-10 11:18:19 LOG:   pid 30151: Cannot accept() new connection. all
backends are down
2010-08-10 11:18:19 LOG:   pid 30150: Cannot accept() new connection. all
backends are down
2010-08-10 11:18:19 LOG:   pid 30152: Cannot accept() new connection. all
backends are down
2010-08-10 11:18:20 LOG:   pid 30153: Cannot accept() new connection. all
backends are down
2010-08-10 11:18:20 LOG:   pid 30154: Cannot accept() new connection. all
backends are down

My questions are:
1. What would cause pgpool to report read failed error?
2. Although I'm not using other features of pgpool like replication, load
balance, etc. except for connection management, why it would automatically
switch on its own into failover mode (failover_handler: no valid DB node
found)?

pgpool parameters
================
replication_mode = false
load_balance_mode = false
master_slave_mode = false
parallel_mode = false

Regards,
Gnanam

_______________________________________________
Pgpool-general mailing list
Pgpool-general at pgfoundry.org
http://pgfoundry.org/mailman/listinfo/pgpool-general