[pgpool-general: 5414] Re: handling network failures

[Muhammad Usama]

Hi

On Sat, Apr 1, 2017 at 6:50 AM, Tatsuo Ishii <ishii at sraoss.co.jp> wrote:

> Usama,
>
> Do you have any suggestion on this?
>
> Best regards,
> --
> Tatsuo Ishii
> SRA OSS, Inc. Japan
> English: http://www.sraoss.co.jp/index_en.php
> Japanese:http://www.sraoss.co.jp
>
> > I have pgpool 3.4 running in watchdog / load balancing  mode with two
> backend postgresql nodes and have run into an issue when there is a brief
> network outage (switch reboot).
> >
> > When an outage occurs the result would be:
> >
> > The standby pgpool node will promote itself to active and execute a
> failover command
> > The active pgpool node will remain active and execute a failover command
> > Both postgresql nodes will be running as master
> > Both pgpool nodes will hold the delegate ip
>

Watchdog was completely overhauled and vastly improved in Pgpool-II 3.5 and
3.6, And the most notable improvement of the watchdog in Pgpool-II 3.5 was
its quorum awareness. From Pgpool-II 3.5 onwards the watchdog only acquires
the delegate-IP, when it has the quorum.
The solution for the second problem, The split-brain in backend PostgreSQL
nodes (when both becomes master) is currently n the active development
phase.


>
> > Was this issue addressed in this?
> > http://www.pgpool.net/pipermail/pgpool-committers/
> 2016-January/002926.html


This particular commit is just one of many that deal with the watchdog
overhauling and enhancements.

Thanks
Best regards
Muhammad Usama


> >
> > If not, how is this this handled?
> >
> > Thanks
> >
> > Travis Kirstine
> > Project Supervisor
> > 140 Renfrew Drive, Suite 100
> > Markham, Ontario L3R 6B3 Canada
> > [cid:image002.png at 01D2AA27.51B79D40]<http://www.firstbasesolutions.com/>
> > T: 905.477.3600 Ext 267 | C: 647
> >
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.sraoss.jp/pipermail/pgpool-general/attachments/20170401/17468404/attachment.html>