Author: Tom Kistner Date: To: Sherwood Botsford CC: exim-users Old-Topics: [Exim] Timeout at data close: Redux Subject: Re: [Exim] OT (?) Timeout at data close: Redux
Sherwood Botsford wrote:
> All joking aside, can anyone who has experience in decrypting tcpdumps
> have a look and give me a lead on what's wrong?
15:35:47.851758 postie.sjsa.internal.net.1313 > mail.rogers.nf.net.smtp:
S [bad tcp cksum 8b1e!] 73069535:73069535(0) win 65535 <mss
1460,nop,wscale 1,nop,nop,timestamp 389281490 0> (DF) (ttl 64, id 42306,
len 60, bad cksum 0!)
15:35:48.535852 mail.rogers.nf.net.smtp > postie.sjsa.internal.net.1313:
S [tcp sum ok] 478952694:478952694(0) ack 73069536 win 25612
<nop,nop,timestamp 2787663206 389281490,nop,wscale 0,mss 1460> (DF) (ttl
53, id 42493, len 60)
[...]
The whole transaction is plagued by transmission errors. Since you have
a rather exotic uplink, I suspect MTU/MSS problems in conjunction with
firewalls that block ICMP "please fragment" messages. This would explain
why only some sites are affected, while others work fine.
The effect is that some packets do not fit into your MTU and may get
multilated (depends on the router/FW).
I think this is not exim-related, so I put an OT in the subject.
However, it would be good to have this in the archives since that
problem is pretty common nowadays ...
