On Fri, Apr 24, 2009 at 4:35 PM, Pedro Silva <[email protected]> wrote:
> 1.I would like to understand if my current configuration is a really > master-master configuration? Sort of. It has a single point of entry, therefore it's not a master-master (it's just master), but commits are done on both nodes then returned to the client, therefore it's a synchronous cluster. > 2. Load Balancing as it is activated will distribute my SELECT queries between > my two backends. Yes. > 3. If I now setup online recovery it doesn't matter if backend1 one or > backend2 > fails because recovery works from the one that is running, right? Yes. > 4. It is my understanding that recovery phase 1 will only finish when clients > terminate their connections. Won't this be a problem due to the fact that web > applications have normally a pool of connections that is kept between the > database and the application? This way the connections will never end and > phase > 1 won't finish, right? It means connections between clients and pgpool-II. pgpool-II will keep those, but cut the ones between pgpool-II and PostgreSQL. Also, pgpool-II will cut the latter and keep the first in a pool that will be attended once the online recovery is done. Also, connections/transactions will be drop after a timeout in case they have not finished/closed by themselves. -- Jaume Sabater http://linuxsilo.net/ "Ubi sapientas ibi libertas" _______________________________________________ Pgpool-general mailing list [email protected] http://pgfoundry.org/mailman/listinfo/pgpool-general
