Weird SMTP timeouts

[jabr at blu.org (John Abreau)]

Jerry Feldman wrote:

> I've seen this a number of times. Kevin had one suggestion, but it simply 
> could be that the helmsbriscoe server is currently overloaded, or that the 
> file you are sending is very large. In many cases it takes care of itself 
> over time. 

I tried running tcpdump to watch what happens:

    sudo tcpdump -vv port 25

I then tried telnet to port 25 on helmsbriscoe, and saved the
output. I then ran tcpdump a second time, and ran

    sudo sendmail -v -qRhelmsbriscoe

which tried to deliver four messages in the queue. It appears that
sendmail is never getting a response from helmsbriscoe at all, so it's
failing before it even knows what the message size is going to be.
And yet telnet seems to work fine.

The only difference I see is that sendmail's outbound packet shows
"tos 0x0", whereas telent's outgoing packet shows "tos 0x10" and
the reply packet from helmsbriscoe shows "tos 0x0". I have no idea
what that means, or if it's significant; I don't see a match for it
in the tcpdump manpage.

How can I watch what sendmail is doing when I process the queue? I'd
like to watch the complete protocol as it happens so I can be certain
what it's doing, but I haven't had any luck getting either sendmail
or tcpdump to show me that.



In the telnet session, I issued a HELO command followed by a QUIT
command. tcpdump showed 4 matches: two sent to helmsbriscoe, and
two coming back from helmsbriscoe:

    14:15:01.180114 IP (tos 0x10, ttl  64, id 5751, offset 0,
    flags [DF], proto 6, length: 60)
    brodie.us.zuken.com.57690 > smtp.helmsbriscoe.com.smtp:
    S [tcp sum ok] 2167306626:2167306626(0) win 5840
    <mss 1460,sackOK,timestamp 342192117 0,nop,wscale 2>

    14:15:01.267297 IP (tos 0x0, ttl 104, id 59165, offset 0,
    flags [none], proto 6, length: 64)
    smtp.helmsbriscoe.com.smtp > brodie.us.zuken.com.57690:
    S [tcp sum ok] 1141564426:1141564426(0) ack 2167306627 win 16384
    <mss 1380,nop,wscale 0,nop,nop,timestamp 0 0,nop,nop,sackOK>

    14:15:01.267385 IP (tos 0x10, ttl  64, id 5753, offset 0,
    flags [DF], proto 6, length: 52)
    brodie.us.zuken.com.57690 > smtp.helmsbriscoe.com.smtp:
    . [tcp sum ok] 1:1(0) ack 1 win 1460 <nop,nop,timestamp 342192204 0>

    14:15:01.355630 IP (tos 0x0, ttl 104, id 41424, offset 0,
    flags [DF], proto 6, length: 111)
    smtp.helmsbriscoe.com.smtp > brodie.us.zuken.com.57690:
    P 1:60(59) ack 1 win 65535 <nop,nop,timestamp 651755 342192117>

When I ran the sendmail queue, the four messages each showed just a
single match, sending to helmsbriscoe and never receiving a response:

    14:05:44.790198 IP (tos 0x0, ttl  64, id 58975, offset 0,
    flags [DF], proto 6, length: 60)
    brodie.us.zuken.com.57532 > hbmail01.helmsbriscoe.com.smtp:
    S [tcp sum ok] 1572213780:1572213780(0) win 5840
    <mss 1460,sackOK,timestamp 341635659 0,nop,wscale 2>


-- 
John Abreau / Executive Director, Boston Linux & Unix
ICQ 28611923 / AIM abreauj / JABBER jabr at jabber.org / YAHOO abreauj
Email jabr at blu.org / WWW http://www.abreau.net / PGP-Key-ID 0xD5C7B5D9
PGP-Key-Fingerprint 72 FB 39 4F 3C 3B D6 5B E0 C8 5A 6E F1 2C BE 99
-------------- next part --------------
A non-text attachment was scrubbed...
Name: jabr.vcf
Type: text/x-vcard
Size: 294 bytes
Desc: not available
URL: <http://lists.blu.org/pipermail/discuss/attachments/20060613/4dfc1062/attachment.vcf>