[Pgpool-general] pgpool-ii health checking

Fri Nov 16 02:57:00 UTC 2007

Hi,

From: "Oleg Laktionov" <olaktionov at gmail.com>
Subject: Re: [Pgpool-general] pgpool-ii health checking
Date: Wed, 14 Nov 2007 19:54:46 +0300

> >You need to perform pcp_attach_node command. pgpool do not reattach a
> >node in health checking.
> 
> And why pgpool does not balance connections when one node fails? I cannot
> connect to the database through the pgpool if some of the nodes is down.
> Only restart of the pgpool helps...

Thank you for the report. Could you apply the following patch?

http://cvs.pgfoundry.org/cgi-bin/cvsweb.cgi/pgpool/pgpool-II/pool_auth.c.diff?r1=1.3&r2=1.3.2.1&f=u

--
Yoshiyuki Asaba
y-asaba at sraoss.co.jp

> 
> 2007-11-13 13:48:55 LOG:   pid 25848: connection received: host=
> tmpwc.test.sportbox.ru port=46253
> 2007-11-13 13:48:55 DEBUG: pid 25848: Protocol Major: 3 Minor: 0 database:
> sportdb user: zope
> 2007-11-13 13:48:55 DEBUG: pid 25848: new_connection: connecting 0 backend
> 2007-11-13 13:48:55 DEBUG: pid 25848: new_connection: connecting 1 backend
> 2007-11-13 13:48:55 DEBUG: pid 25848: new_connection: connecting 2 backend
> 2007-11-13 13:48:55 DEBUG: pid 25848: new_connection: skipping slot 2
> because backend_status = 3
> 2007-11-13 13:48:55 DEBUG: pid 25848: pool_read_message_length: slot: 0
> length: 8
> 2007-11-13 13:48:55 DEBUG: pid 25848: pool_read_message_length: slot: 1
> length: 8
> 2007-11-13 13:48:55 DEBUG: pid 25848: trying clear text password
> authentication
> 2007-11-13 13:48:55 DEBUG: pid 25848: trying clear text password
> authentication
> 2007-11-13 13:48:55 DEBUG: pid 25848: trying clear text password
> authentication
> 2007-11-13 13:48:55 DEBUG: pid 24485: reap_handler called
> 2007-11-13 13:48:55 DEBUG: pid 24485: reap_handler: call wait3
> 2007-11-13 13:48:55 DEBUG: pid 24485: child 25848 exits with status 11 by
> signal 11
> 2007-11-13 13:48:55 DEBUG: pid 26148: I am 26148
> 2007-11-13 13:48:55 DEBUG: pid 24485: fork a new child pid 26148
> 2007-11-13 13:48:55 DEBUG: pid 24485: reap_handler: normally exited
> 
> 
> 
> 2007/11/14, Yoshiyuki Asaba <y-asaba at sraoss.co.jp>:
> >
> > Hi,
> >
> > From: "Oleg Laktionov" <olaktionov at gmail.com>
> > Subject: Re: [Pgpool-general] pgpool-ii health checking
> > Date: Tue, 13 Nov 2007 14:40:44 +0300
> >
> > > Pgpool successfully realized the failure...
> > >
> > > Further I start the Postgresql server on the node tmpwc ...
> > >
> > > 2007-11-13 13:44:04 DEBUG: pid 25879: I am 25879
> > > 2007-11-13 13:44:14 LOG:   pid 24485: starting health checking
> > > 2007-11-13 13:44:14 DEBUG: pid 24485: health_check: 0 the DB node
> > status: 2
> > > 2007-11-13 13:44:14 DEBUG: pid 24485: health_check: 1 the DB node
> > status: 2
> > > 2007-11-13 13:44:14 DEBUG: pid 24485: health_check: 2 the DB node
> > status: 3
> > > 2007-11-13 13:44:24 LOG:   pid 24485: starting health checking
> > > 2007-11-13 13:44:24 DEBUG: pid 24485: health_check: 0 the DB node
> > status: 2
> > > 2007-11-13 13:44:24 DEBUG: pid 24485: health_check: 1 the DB node
> > status: 2
> > > 2007-11-13 13:44:24 DEBUG: pid 24485: health_check: 2 the DB node
> > status: 3
> > >
> > > Nevertheless pgpoll does not see this event... Ordinary connection to
> > the
> > > tmpwc works fine...
> >
> > You need to perform pcp_attach_node command. pgpool do not reattach a
> > node in health checking.
> >
> >
> > > And I have some more question. Can I disable failover mode for the
> > pgpool?
> > > For example I wish to use only replica.test.sportbox.ru as the master
> > node.
> > > Is it possible anyhow?
> >
> > No, if a master node is down, pgpool determines a new master node.
> >
> > --
> > Yoshiyuki Asaba
> > y-asaba at sraoss.co.jp
> >
> > >
> > >
> > > 2007/11/13, Yoshiyuki Asaba <y-asaba at sraoss.co.jp>:
> > > >
> > > > Hi,
> > > >
> > > > From: "Oleg Laktionov" < olaktionov at gmail.com>
> > > > Subject: Re: [Pgpool-general] pgpool-ii health checking
> > > > Date: Mon, 12 Nov 2007 10:37:06 +0300
> > > >
> > > > > I use pgpool-ii-1.3.
> > > >
> > > > Hmm... I can't reproduce it.
> > > > Could you show us pgpool's error log?
> > > >
> > > > Regards,
> > > > --
> > > > Yoshiyuki Asaba
> > > > y-asaba at sraoss.co.jp
> > > >
> > > >
> > > > >
> > > > >
> > > > > 2007/11/12, Yoshiyuki Asaba < y-asaba at sraoss.co.jp >:
> > > > > >
> > > > > > Hi,
> > > > > >
> > > > > > From: "Oleg Laktionov" < olaktionov at gmail.com>
> > > > > > Subject: [Pgpool-general] pgpool-ii health checking
> > > > > > Date: Fri, 9 Nov 2007 16:19:43 +0300
> > > > > >
> > > > > > > About pgpool-ii...
> > > > > > > Can anybody explain me the mechanism of performing backend
> > health
> > > > > > checking?
> > > > > > > Pgpool-ii successfully understands that some of the backends is
> > > > down.
> > > > > > > Nevertheless when some of the backends is online again, pgpool
> > does
> > > > not
> > > > > > see
> > > > > > > this event.
> > > > > > > Only restart of the pgpool-ii helps...
> > > > > >
> > > > > > Which version do you use?
> > > > > >
> > > > > > --
> > > > > > Yoshiyuki Asaba
> > > > > > y-asaba at sraoss.co.jp
> > > > > > _______________________________________________
> > > > > > Pgpool-general mailing list
> > > > > > Pgpool-general at pgfoundry.org
> > > > > > http://pgfoundry.org/mailman/listinfo/pgpool-general
> > > > > >
> > > > >
> > > > >
> > > > >
> > > > > --
> > > > > best regards,
> > > > > Oleg Laktionov
> > > >
> > >
> > >
> > >
> > > --
> > > best regards,
> > > Oleg Laktionov
> > _______________________________________________
> > Pgpool-general mailing list
> > Pgpool-general at pgfoundry.org
> > http://pgfoundry.org/mailman/listinfo/pgpool-general
> >
> 
> 
> 
> -- 
> best regards,
> Oleg Laktionov