On Thu, Oct 27, 2011 at 1:48 PM, Scott Marlowe <scott(dot)marlowe(at)gmail(dot)com> wrote:
> OK, a few points. 1: You've got a zombie process. Find out what's
To expand on the zombie thing, it's quite possible that you're
managing to make a pg backend process crashout, which would cause the
db to restart midday, which is bad (TM) since that dumps all of shared
buffers and forces all clients to reconnect. So look through the
system logs for segmentation faults, etc.