Thank you again Daniel, I am now testing the recovery process on my test databases now.
Thanks Deepak On Tue, Mar 17, 2009 at 2:13 PM, <[email protected]> wrote: > You are welcome. > > > > > So this is done by using pcp_recovery_node command? > > > > Yes. That command will trigger everything I explained. Also, be aware that > your database needs to have the pgpool stored procedures installed in order > for it to work. And for them to be installed, your database needs to have > the proper pl languages installed, I don’t remember which one, but got to be > either plpgsql or plperlu. I think it’s the former. Read the pgpool tutorial > for info on all this. > > > > Daniel > > > ------------------------------ > > *From:* DM [mailto:[email protected]] > *Sent:* Tuesday, March 17, 2009 4:57 PM > *To:* Crespo, Daniel @ SDS > *Cc:* [email protected] > > *Subject:* Re: [Pgpool-general] Is there a option > > > > Thanks Daniel, Great Answer. > > So this is done by using pcp_recovery_node command? > > Thanks > Deepak > > On Tue, Mar 17, 2009 at 1:30 PM, <[email protected]> wrote: > > Let’s say you have two backends: backend0 and backend1. Supposing backend1 > has failed, recoverying it (by re-synchronizing data from backend0) is > expected to have a pgpool client reconnection. From the point of view of > pgpool, the clients must disconnect and reconnect in order to complete a > data synchronization. However, this disconnection can be very quick. The > interesting part is that data synchronization (a kind of copy of the > postgresql data directory) will happen while backend0 is available por reads > and writes. At the end of the data transfer (1st stage), a postgresql > feature (WAL archiving) helps with the synchronization of newer data > (2ndstage) that has been transferred at the 1 > st stage. At 2nd stage is when clients must disconnect and reconnect. > Pgpool has the ability to normally accept connections during 1st stage. At > 2nd stage, pgpool waits until clients disconnect or until > client_idle_limit_in_recovery seconds happen. Then clients get disconnected. > If clients try to connect while 2nd stage happen, pgpool will block them > until it finishes. Clients won’t get disconnected, just will sit waiting > until connection happens. > > > > I hope this helps. > > > > Daniel > > > ------------------------------ > > *From:* Crespo, Daniel @ SDS > *Sent:* Tuesday, March 17, 2009 4:05 PM > *To:* 'DM'; [email protected] > *Subject:* RE: [Pgpool-general] Is there a option > > > > Yes. Look for the following keywords: > > > > pcp_recovery_node (pgpool client program) > > client_idle_limit_in_recovery (parameter in pgpool.conf) > > pgpool_copy_backup (script) > > pgpool_switch_xlog (script) > > pgpool_remote_start (script) > > > > Daniel > > > ------------------------------ > > *From:* [email protected] [mailto: > [email protected]] *On Behalf Of *DM > *Sent:* Tuesday, March 17, 2009 2:57 PM > *To:* [email protected] > *Subject:* Re: [Pgpool-general] Is there a option > > > > Hi All, > > > > Sorry I wanted to ask the question the other way round. > > In Pgpool is it possible to recover / sync the 2nd DB without quiescing the > first. > > Thanks > Deepak > > On Tue, Mar 17, 2009 at 9:48 AM, DM <[email protected]> wrote: > > Hi All, > > In pgpool Is it possible to do quiescing of a db while (recovering)/syncing > the 2nd DB? > > Thanks > Deepak > > > > >
_______________________________________________ Pgpool-general mailing list [email protected] http://pgfoundry.org/mailman/listinfo/pgpool-general
