For some reason postgres is spawning >700 postmaster processes for handling database requests and the postgres log file if full of 'unexpected EOF on client connection', 'incomplete startup packet' and 'sorry, too many clients already'. netstat tells me that all the open connections are local and I'm pretty sure are coming from postgres internally. This particular instance has been running just fine for the last 230 days or so and nothing has changed configuration wise. Any thoughts on where I should be looking to try and resolve this issue?

This is my first time diagnosing a problem like this so if there is any steps I can take to help narrow down the cause that would be helpful as well.