From: Gregory Stark <stark@enterprisedb.com>
To: "Tom Lane" <tgl@sss.pgh.pa.us>
Cc: "Josh Berkus" <josh@agliodbs.com>,
	"Alvaro Herrera" <alvherre@commandprompt.com>,
	<pgsql-performance@postgresql.org>
Subject: Re: User concurrency thresholding: where do I look?
In-Reply-To: <12686.1184865938@sss.pgh.pa.us> (Tom Lane's message of "Thu, 19
	Jul 2007 13:25:38 -0400")
Organization: EnterpriseDB
References: <200707190828.59088.josh@agliodbs.com>
	<20070719154912.GG6538@alvh.no-ip.org>
	<200707190922.24497.josh@agliodbs.com>
	<12686.1184865938@sss.pgh.pa.us>
Date: Fri, 20 Jul 2007 12:46:07 +0100
Message-ID: <871wf39u2o.fsf@oxford.xeocode.com>
User-Agent: Gnus/5.110006 (No Gnus v0.6) Emacs/21.4 (gnu/linux)
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii


"Tom Lane" <tgl@sss.pgh.pa.us> writes:

> Josh Berkus <josh@agliodbs.com> writes:
>
>> That's an interesting thought.  Let me check lock counts and see if this is 
>> possibly the case.
>
> AFAIK you'd get hard failures, not slowdowns, if you ran out of lock
> space entirely

I assume you've checked the server logs and are sure that you aren't in fact
getting errors. I could, for example, envision a situation where a fraction of
the transactions are getting some error and those transactions are therefore
not being counted against the txn/s result.

-- 
  Gregory Stark
  EnterpriseDB          http://www.enterprisedb.com