著者: Giuliano Gavazzi 日付: To: Florian Weimer, Marc Haber CC: exim-users 題目: Re: [exim] Receiving mail from outblaze and others times out
after initial 220
At 11:36 am +0200 2005/04/19, Florian Weimer wrote: >* Marc Haber:
>
>> The host is behind a NAT router on a DSL line, so a MTU problem seems
>> to be the kneejerk response, but neither the initial 220 greeting nor
>> the HELO/EHLO from the remote site is likely to trip any MTU and PMTUD
>> issues.
>>
>> Do you guys have an idea what could be the problem here?
>
>The TCP implementation on the NAT router is probably broken. Maybe it
>strips TCP options or something like that.
>
>Unfortunately, without packet captures before and after the NAT
>router, this is close to impossible to diagnose. You get what you pay
>for. 8-(
well, I had the most disparate experiences with ADSL routers. I wish
I only had modems and let all the real work done by proper software.
With a USR SureConnect 9003 I have a specific domain from which I
cannot receive or send anything larger than a few k. A tcpdump (this
side) shows a lot of retries, I suppose owing to packet loss, but
with no other explanation, and since at the other end there are deaf
admins I cannot do more than that (except that changing to another
ADSL router fixes the problem). And yesterday, I realised that a
Draytek (whose manuals are close to garbage) Vigor 2600i was
rejecting all externally initiated udp traffic even with a set DMZ
host and no firewall rules set. You can imagine how the DNS server
(that is the DMZ host) was answering to queries... A firmware update
(version 2.5.6) fixed the problem.
I wonder if you really get what you pay for.. is overspending for a
Cisco really the answer? What about the pix firewall... My philosophy
is to stick to something you know, only that models seem to come and
go like cellular phones..
[or just go back to my old couple of Motorola Premier 33.6K modems,
back-to-back into a 2M line, that was reliable!]
