[pgpool-hackers: 403] Re: Test down node in master/slave streaming replication mode, and reattach it.

[Sergey Logvinov]

Oo, i find copy-past bug, sorry.

variable change i->j

+					else if (pg_standby[i] == 0) // up master node
+					{
+						int lag = 0;
+						int j;
+						// To allow master node need lag=0 for all standby servers
+						for (j=0;j<NUM_BACKENDS;j++)
+							if (VALID_BACKEND(j) && pg_standby[j] == 1)




2013/11/13 Tatsuo Ishii <ishii at postgresql.org>

> Thanks for the proposal. I will look into this.
> --
> Tatsuo Ishii
> SRA OSS, Inc. Japan
> English: http://www.sraoss.co.jp/index_en.php
> Japanese: http://www.sraoss.co.jp
>
> > In streaming master/slave we testing all online nodes in function
> > check_replication_time_lag.
> >
> > I add new feature. I test all nodes include dead nodes.
> >
> > If down slave node now alive, i compare time_lag with master. And try
> this
> > few times (new param in config sr_check_max_retries) If slave node not
> far
> > behind, i mark node as NODE_UP.
> >
> > If mater node now is online, i check time_lag on all slave nodes. And if
> > lag=0, i think this master is working fine, and reattach it.
> >
> > This will solved the problem ticket #17
> > My implementation see in diff file.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.sraoss.jp/pipermail/pgpool-hackers/attachments/20131114/5b52107c/attachment.html>