[pgpool-general: 478] load balancing seems to be bottlenecked by performance of master

Lonni J Friedman netllama at gmail.com
Tue May 22 05:18:53 JST 2012

I'm running pgpool-II-3.1.2 purely for load balancing in front of a 4
node postgresql-9.1.3 cluster (all running on Linux-x86_64).  I'm
using streaming replication with 3 hot standby servers.  I have the
weighting distributed such that the master weight is set to '1' and
the three standby servers are set to '33' each.  My intent was that
this would nearly eliminate all read queries going to the master,
however what I'm observing is that the overall performance seems to be
bottlenecked by the performance of the master, even when queries
aren't ultimately getting sent to the master.

For example, there are times when the load on the master is much
higher than on the standby servers, and the master's overall
performance is noticeably degraded.  When this occurs, I've found that
establishing a database connection through pgpool with a SQL query
that should normally go to one of the standby servers has very poor
performance.  It seems like pgpool is silently doing something with
the master (which is under load and slower to respond), before passing
the query itself off to a standby server.

Is this expected behavior?


More information about the pgpool-general mailing list