[Pgpool-general] health_check pgpool-ii

[Martin Fandel]

Hi again,

an then if I restart pgpool, it doesn't know which database is up to 
date:

        pgpool_test=# select count(*) from test;
         count 
        -------
           115
        (1 Zeile)
        
        pgpool_test=# select count(*) from test;
         count 
        -------
            91
        (1 Zeile)
        
        pgpool_test=# select count(*) from test;
         count 
        -------
           115
        (1 Zeile)
        
        pgpool_test=# select count(*) from test;
        
So I must delete this node in the configuration to avoid differences?

Or...: 
Is it possible to implement a update-status table for each node? So that
only a node is allowed for querys, if the update status is the same as 
the other nodes? (only checked by stopping/starting pgpool; stop->
update-status, start->update-status)

Other topic:
If one node fails, the Web-Interface can't show the Node-Info?!?

Thanks!

Regards,

Martin

Am Dienstag, den 31.10.2006, 16:03 +0100 schrieb Martin Fandel:
> Hi,
> 
> it works. health_check is ok:
> 
>         2006-10-31 15:40:36 LOG:   pid 22677: starting health checking
>         2006-10-31 15:40:36 DEBUG: pid 22677: health_check: 0 the DB
>         node status: 2
>         2006-10-31 15:40:36 DEBUG: pid 22677: health_check: 1 the DB
>         node status: 2
>         2006-10-31 15:40:36 DEBUG: pid 22677: health_check: 2 the DB
>         node status: 2
>         2006-10-31 15:40:36 ERROR: pid 22677:
>         connect_inet_domain_socket: connect() failed: Connection refused
>         2006-10-31 15:40:36 ERROR: pid 22677: health check failed. 2 th
>         host 192.168.6.52 at port 5434 is down
>         2006-10-31 15:40:36 LOG:   pid 22677: set 2 th backend down
>         status
>         2006-10-31 15:40:36 DEBUG: pid 22677: failover_handler called
>         2006-10-31 15:40:36 DEBUG: pid 22677: failover_handler: starting
>         to select new master node
>         2006-10-31 15:40:36 LOG:   pid 22677: failover_handler: do not
>         restart pgpool. same master node 0 was selected
> 
> But the established connection to the socket 9999 hangs / is lost. I 
> must reconnect to the socket for a working process. There is no failback
> function implemented, right? In a fail state, I must redump the failed 
> database?!? Is there any pgpool-posibility for failback including sync
> the latest data?
> 
> Thanks!
> 
> Regards,
> 
> Martin
> 
> Am Freitag, den 27.10.2006, 12:31 +0200 schrieb Martin Fandel:
> > Hi,
> > 
> > I'm really new using pgpool and right now I'm very happy with it. I was
> > using Slony before. Slony was working, but I was not happy with the many
> > ddl changes on our database :).
> > 
> > I've setted up pgool-ii yesterday with the webinterface. All seems to
> > work, but I've a problem with the health check.
> > 
> > My postgresql-Log says:
> > 
> >         could not receive data from client: Es konnte keine Verbindung
> >         hergestellt werden, da der Zielcomputer die Verbindung
> >         verweigert # (in english: access denied from remote host)
> >         unexpected EOF on client connection
> >         disconnection: session time: 0:00:00.00 user=postgres
> >         database=template1 host=192.168.131.12 port)=34758
> > 
> > I've configured pgpool for 4 nodes. For every node my pg_hba.conf is
> > changed to this:
> >         host	all	all	192.168.131.12/32  trust
> > 
> > My pgpool-health-configuration is configured as follows:
> >         health_check_timeout = 20
> >         health_check_period = 10
> >         health_check_user = 'postgres'
> > 
> > I hope someone can help me :)!
> > 
> > Thanks!
> > 
> > Regards,
> > 
> > Martin
> > 
> > _______________________________________________
> > Pgpool-general mailing list
> > Pgpool-general at pgfoundry.org
> > http://pgfoundry.org/mailman/listinfo/pgpool-general
> 
> _______________________________________________
> Pgpool-general mailing list
> Pgpool-general at pgfoundry.org
> http://pgfoundry.org/mailman/listinfo/pgpool-general