Date: Thu, 8 Feb 2007 09:29:46 +0100
From: Magnus Hagander <magnus@hagander.net>
To: "Marc G. Fournier" <scrappy@hub.org>
Cc: Kevin Hunter <hunteke@earlham.edu>,
	PostgreSQL www <pgsql-www@postgresql.org>
Subject: Re: email is fast!
Message-ID: <20070208082946.GA27377@svr2.hagander.net>
References: <200702020546.l125kPN11116@momjian.us>
	<20070202082145.GC20955@svr2.hagander.net>
	<45C8FDA1.7050903@earlham.edu>
	<BD2825F7DB1811D988ED6570@ganymede.hub.org>
	<45CA50E1.6090109@hagander.net>
	<4308AA640AC951EDA7E58305@ganymede.hub.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <4308AA640AC951EDA7E58305@ganymede.hub.org>
User-Agent: Mutt/1.5.11

On Wed, Feb 07, 2007 at 07:01:43PM -0400, Marc G. Fournier wrote:
> >> Just to follow up this one ... Kevin sent me an email including the full
> >> headers for one that took 12 hours to be delivered ... using the QUEUE ID, I
> >> checked the logs and it turns out that the mail server couldn't resolve the
> >> MX  the first and second time it tried to deliver, but succeeded on the
> >> third ...
> >
> > This sounds scaringly close to the problem we had on wwwmaster with it
> > not delivering emails. Resolving separately looked fine, but sendmail
> > couldn't resolve.
> > Could it be symptoms of the same problem?
> 
> Does postfix do any 'caching of results'?  For instance, if it got a failed 
> response, would it cache that and re-use that later, pending it timing out in 
> its own cache?  So try 1 failed, try 2 used cache and failed, try 3 was after 
> cache timed out and re-tried, at which point the local DNS server had the right 
> value ...

I don't beleive it does. The resolver library might do it though - some do, others don't.

//Magnus