[Pgcluster-general] MasterDB is always the first node from pgreplicate.conf - even if not accessible

[a.mitani at sra-europe.com]

Hi,

When restart Cluster DB, please send just transaction query (like
BEGIN;END;).
Then, replication server find a failed down cluster.

Currentry, the latest version support lifecheck.
In this version, it is not necessary to send this trigger.

Regards,
-----------------
At.Mitani

> I have 3 DB nodes in my cluster which are listed in Cluster_Server_Info
> list
> as db1, db2, db3 (in exactly this order) in pgreplicate.conf
>
>
>
> If I start db in recovery mode ( -R option ), replicator always tries to
> use
> db1 as Master DB, it does want to use any other. For example if db1 is
> stopped, db is running and I start db3 in recovery mode, I want it to
> synch
> with db2, but it doesn't do this - it wait for db1 to be available
>
>
>
> Looks like Master DB in recovery process is always the first node in
> Cluster_Server_Info list.
>
>
>
> Is there a way to force recovery from another node when first one is not
> running?
>
>
>
> Thanks
>
>
>
> Best regards
>
> Alexey Kamenchuk
>
>
>
> _______________________________________________
> Pgcluster-general mailing list
> Pgcluster-general at pgfoundry.org
> http://pgfoundry.org/mailman/listinfo/pgcluster-general
>