[Pgpool-general] replication/failover feature in Master/Slavemode

[Daniel.Crespo at l-3com.com]

I think there is a misunderstanding...

To clarify (just in case).

1. Pgpool detects non-healthy postgresql backend.
2. Pgpool executes WHATEVER is in "failover_command" entry in
pgpool.conf
3. The end.

In "failover_command" you can put
"executeFailoverOfMyCurrentSlavePostgreSqlBackend.sh".
Which means, in yout pgpool.conf you should have:
failover_command="/path/to/executeFailoverOfMyCurrentSlavePostgreSqlBack
end.sh"

in which you should do something like:

+++++++++++++++++
#!/bin/bash

touch /path/to/where/trigger/file/must/exist

+++++++++++++++++

Of course, you'll want to touch the trigger file of the backend that is
the standby, so you got to know its path. If you don't know it, let us
know.

You might want to try having failover_command execute something that
writes to /tmp/somefile to prove that pgpool is actually executing
failover_command.

Best regards,
-Daniel

> -----Original Message-----
> From: Tatsuo Ishii [mailto:ishii at sraoss.co.jp]
> Sent: Friday, January 28, 2011 9:21 AM
> To: armin.nesiren at intera.si
> Cc: sandeeptt at yahoo.com; ishii at sraoss.co.jp; Crespo, Daniel @ SSG -
> SDS; pgpool-general at pgfoundry.org
> Subject: Re: [Pgpool-general] replication/failover feature in
> Master/Slavemode
> 
> > Also I have same problem.
> > When I do pcp_recovery of my old primary node (node 0),
> failover_command is
> > not executed,
> 
> If you are using 3.0.x, then it's a known problem.
> Please try CVS HEAD (to be released as 3.1).
> --
> Tatsuo Ishii
> SRA OSS, Inc. Japan
> English: http://www.sraoss.co.jp/index_en.php
> Japanese: http://www.sraoss.co.jp
> 
> > I have set failback_command, and that command creates trigger file
> for me.
> >
> > This configuration works fine right now, but I would like to know if
> this is
> > right way to do it?
> >
> > On Fri, Jan 28, 2011 at 12:53 PM, Sandeep Thakkar
> <sandeeptt at yahoo.com>wrote:
> >
> >> Well, that's what my first question was. Why pgpool is not
executing
> the
> >> failover_command in my case? Does it need any more settings in
> pgpool.conf?
> >> I did not set failback_command.
> >>
> >>
> >>
> >> ------------------------------
> >> *From:* Tatsuo Ishii <ishii at sraoss.co.jp>
> >> *To:* sandeeptt at yahoo.com
> >> *Cc:* Daniel.Crespo at l-3com.com; guillaume at lelarge.info;
> >> pgpool-general at pgfoundry.org
> >> *Sent:* Fri, January 28, 2011 2:01:50 PM
> >>
> >> *Subject:* Re: [Pgpool-general] replication/failover feature in
> >> Master/Slavemode
> >>
> >> No. Failover_command will be automatically executed by pgpool-II if
> >> primary goes down. And you can have touch command to create the
> >> trigger file in your failover_command. This should automatically
> >> promote the standby.
> >> --
> >> Tatsuo Ishii
> >> SRA OSS, Inc. Japan
> >> English: http://www.sraoss.co.jp/index_en.php
> >> Japanese: http://www.sraoss.co.jp
> >>
> >> > Thanks Tatsuo. So, does it mean with existing PG version (9.0),
we
> need
> >> to
> >> > manually execute failover script (create trigger file) even when
> using
> >> pgpool?
> >> >
> >> >
> >> >
> >> >
> >> > ________________________________
> >> > From: Tatsuo Ishii <ishii at sraoss.co.jp>
> >> > To: sandeeptt at yahoo.com
> >> > Cc: Daniel.Crespo at l-3com.com; guillaume at lelarge.info;
> >> > pgpool-general at pgfoundry.org
> >> > Sent: Fri, January 28, 2011 1:08:08 PM
> >> > Subject: Re: [Pgpool-general] replication/failover feature in
> >> Master/Slavemode
> >> >
> >> >> Yes, I'm aware that I need to create a trigger file (same as
what
> is
> >> defined in
> >> >>
> >> >> recovery.conf) to start the failover and promote the Standby to
> Primary.
> >> I have
> >> >>
> >> >> tried this when I did not use pgpool and it worked fine. I
> thought, when
> >> pgpool
> >> >>
> >> >> is in picture, it will execute the failver_command, when it
> detects that
> >>
> >> >>primary
> >> >>
> >> >> is down. and since, the failover_command in my case is to "touch
> trigger
> >> file".
> >> >>
> >> >> Shouldn't pgpool execute the failover_command here and create
> that file
> >> >> automatically?
> >> >
> >> > Besides the problem that it's not easy to read and parse
> >> > recovery.conf, creating trigger file might not be the only way to
> >> > promote standby server forever. Actually currently PostgreSQL
> >> > developers are discussing about "pg_ctl failover" command which
> will
> >> > be a new way to promote standby in the next version of
> >> > PostgreSQL. Pgpool-II is designed to be flexible to work with as
> many
> >> > PostgreSQL version as possible. So we don't want to hard code
> "touch
> >> > trigger file" in pgpool itself.
> >> >
> >> >> I think, either pgpool is not able to detect that primary is
down
> >> >> or it is unable to execute the failover_command. Am I missing to
> define
> >> >>anything
> >> >>
> >> >> in the configuration file?
> >>
> >>
> >>
> >> _______________________________________________
> >> Pgpool-general mailing list
> >> Pgpool-general at pgfoundry.org
> >> http://pgfoundry.org/mailman/listinfo/pgpool-general
> >>
> >>
> >
> >
> > --
> > Armin