--
Op ma 04-08-2003, om 16:27 schreef Colm MacCarthaigh:
> On Mon, Aug 04, 2003 at 03:47:11PM +0200, Wouter Verhelst wrote:
> > My machine has two global IPv6 addresses; the oldest one being
> > 3ffe:100:e:c::1; the newest one being 2001:ab8:2007:20::1. (please don't
> > start wasting my bandwidth :-)
>
> You should really plan to stop using the 3ffe one sooner rather than
> later, already a large part of the IPv6 internet is using 2000::/3
> rather than ::/0 for default routing.
Yeah, I know, but then I'd like the 2001 one to work first :-)
> > On the network layer, they both work perfectly; they're routable, can be
> > pinged, and so on. However, when I try to connect to port 25, with Exim
> > 3.35 (the one in Debian's stable release) running behind it, the 2001
> > address doesn't work; a connection is made, but a 220 banner is not
> > received; the connection will time out eventually.
>
> Are you using exim in standalone mode or from inetd ?
Standalone; the inetd I've got installed doesn't support IPv6.
> I'm using exim
> 3.35 on Debian with multiple IPv6 addresses without any problems, though
> we use exim 4 on our production boxes (again, on Debian).
>
> > There's a difference when I connect to the 3ffe address; in this case,
> > everything works as it should, i.e. exim replies immediately. Oddly,
> > when I do so, suddenly the 2001 address works too; if a connection was
> > open at that time, it receives a 220 banner, while new connections will
> > work perfectly for a while.
>
> *bizarre*, can you run either strace the daemon, or run "strace -fp <pid>"
> where pid is the exim daemon ?
Sure. Since the output is quite long, I've put it up on
http://www.grep.be/exim-strace
At first glance, I don't see anything out of the ordinary, but that
could be me. It could be interesting to note that the machine from where
I initiated the connections has 2001:ab8:2007:0:2e0:ff:fe9a:702f as its
IPv6 address.
> > I'm not entirely sure whether the fault is exim's or the operating
> > system; for completeness, I'm running an up-to-date Debian Stable with a
> > self-compiled 2.4.20 Linux kernel; the system is an i386-based one.
>
> Same here :)
>
> > Any insighst as to what could cause this are very welcome.
>
> Could be DNS-related, though not very likely.
I don't think that's the case either. If it were, it would take a long
while to connect, but the machine would eventually connect. As it is, it
doesn't; unless I connect to the 3ffe address, nothing happens.
> What happens if you
> connect to ::1 on port 25 ?
That works perfectly, but AFAICS it does not 'release' the 2001 address.
--
Wouter Verhelst
Debian GNU/Linux --
http://www.debian.org
Nederlandstalige Linux-documentatie --
http://nl.linux.org
"An expert can usually spot the difference between a fake charge and a
full one, but there are plenty of dead experts."
-- National Geographic Channel, in a documentary about large African beasts.
--
Content-Description: Dit berichtdeel is digitaal ondertekend
[ signature.asc of type application/pgp-signature deleted ]
--
