[pgpool-hackers: 156] Re: persisting pgpool status

[Meister, Klaus]

Hi Alex,

I think one problem was  a stonith device that did a reboot only and as you said,  there was no alarm.
Now there is a Nagios check that look for the nodes states via pcp_node_info ;)

Thank you.
Klaus



________________________________
Von: pgpool-hackers-bounces at pgpool.net [pgpool-hackers-bounces at pgpool.net]" im Auftrag von "alex [alex at smalldemons.com]
Gesendet: Donnerstag, 8. November 2012 18:19
An: pgpool-hackers at pgpool.net
Betreff: [pgpool-hackers: 154] Re: persisting pgpool status

I think the problem you should be trying to solve is that you didn't know about the failover.

You have several options for monitoring that.  I have the failover_command in pgpool.conf send an alert as well as perform the failover.  In case that somehow doesn't make it through, I have another script tailing the pgpool logs for "health_check: (\d+) th DB node status: (\d)$" and if the second number isn't 2, it sends an alert.  It also makes sure the health check is in the last two minutes, which assures me the health checks are running.  You could get the same information from pcp_pool_status if you'd prefer to monitor remotely through Nagios or something.


  alex


On 11/8/12 12:57 AM, Meister, Klaus wrote:
Hi pgpool-hackers,

we are using pgpool2-3.1.3 in a productive environment with streaming replication.
For the single point of failure pgpool we have a corosync/pacemaker configuration on each database machine.

Once  we had an automatic  failover (because of a network failure) wich worked fine but which we did not realize!(?)
2 weeks later there was another failover (wich worked fine too :)  and we were back on the obsolete database which was still up and open.

To prevent this we are planning to use the pgpool_status file on a shared storage and use it on startup/failover/switchover …
This file seems to be written only in case of cleanly stopping the ppool.
I think it would be useful to do this every time the status of a node is being changed.

Maybe is there another recommended way to solve our problem.

With kind regards
Klaus Meister




--
Dipl.-Phys. Klaus Meister

Datenbankadministrator
mecom Medien-Communikations-Gesellschaft mbH
Mittelweg 143
D 20148 Hamburg

Tel:   +49 40 4113-32824
Fax:  +49 40 451962
Web: <http://www.mecom.de/><http://www.mecom.de/>
Registergericht Hamburg, HRB 43177
Geschäftsführer: Ulrich Wiehsalla, Barbara Bliefert
--------------------------------------------------------------
Sollten Sie nicht der beabsichtigte Empfänger sein,
informieren Sie bitte den Absender dieser E-Mail entsprechend.




_______________________________________________
pgpool-hackers mailing list
pgpool-hackers at pgpool.net<mailto:pgpool-hackers at pgpool.net>
http://www.pgpool.net/mailman/listinfo/pgpool-hackers


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.sraoss.jp/pipermail/pgpool-hackers/attachments/20121108/053e1401/attachment.html>