To: Scott Marlowe <scott.marlowe@gmail.com>
cc: Brian Cox <brian.cox@ca.com>,
	"pgsql-performance@postgresql.org" <pgsql-performance@postgresql.org>
Subject: Re: multiple threads inserting into the same table 
In-reply-to: <dcc563d10903231338y38879310j438eb07b64593c42@mail.gmail.com> 
References: <49C7F050.7030505@ca.com>
	<dcc563d10903231338y38879310j438eb07b64593c42@mail.gmail.com>
Comments: In-reply-to Scott Marlowe <scott.marlowe@gmail.com>
	message dated "Mon, 23 Mar 2009 14:38:52 -0600"
Date: Mon, 23 Mar 2009 18:57:34 -0400
Message-ID: <4396.1237849054@sss.pgh.pa.us>
From: Tom Lane <tgl@sss.pgh.pa.us>

Scott Marlowe <scott.marlowe@gmail.com> writes:
> On Mon, Mar 23, 2009 at 2:25 PM, Brian Cox <brian.cox@ca.com> wrote:
>> This generates and executes a single SQL insert. Since, as you point out,
>> postgres seems to think that this transaction isn't doing anything,
>> it's hard to figure out what the read is doing.

> Might you have a firewall that's killing the connections?  What does
> netstat -an on the client side say about these connections?

netstat will probably say the connection is open on both sides ---
otherwise the sockets would have closed.  It looks like both sides
still think the connection is open.  A firewall timeout is still
a possibility, but you'd have had to have a fairly long idle time
for that to happen.  Are any of the threads issuing commands that might
have run for very long intervals (tens of minutes)?

			regards, tom lane