Hey guys, We've setup a pgpool instance on each one of our databases, verified that it's working and can be connected to. We verified that we have enough postgres connections (200) for the pgpool connections (32*4=128). Our servers are usually only servicing a few queries at a time (~5 average) with occasional spikes into the 30's or 40's when things get bogged down, but for the most part 128 should be way more connections than we need.
In our client apps I disable in-process connection pooling and change the port to 5433. If we leave connection pooling enabled on the clients, we use up all the pgpool connections and are right back to where we started. Next, we start up our servers and everything appears to be working. Except... We occasionally get "Failed to establish a connection" errors for some unknown reason. Our establish connection timeout is 10 seconds which seems very generous to me. Our databases do not appear to be overloaded. In fact, most things work just fine. Across the cluster, we average about 1 error every 15 seconds or so, and we're getting 100's of hits/minute. Without pgpool using a very large # of idle postgres connections, we rarely see these errors. With pgpool, they happen frequently enough that it's a problem. I'm unsure what to do next. Could anybody help? Thanks, Bryan
_______________________________________________ Pgpool-general mailing list [email protected] http://pgfoundry.org/mailman/listinfo/pgpool-general
