[Mailman-Users] maximum hop count exceeded

[Grant Taylor]

On 05/26/10 02:35, Stephen J. Turnbull wrote:
> It fixes it for the Mailman host, yes.  It's still wasting some 
> bandwidth and may indicate more serious problems on an intermediate 
> host.  Typically it's not a problem at either the last hop or the 
> apparent sender, so quite painful to trace in the absence of headers.

I'm not sure that we are talking about exactly the same thing.

If you are trying to diagnose messages after they have been through 
Mailman, and had the headers removed like we have discussed, then yes, 
finding loops before Mailman will be difficult.

This is why I was talking about having a pre-Mailman archive 
specifically for diagnostic purposes.

I feel like I should throw out there that I consider Mailman to be and 
end point / consumer if you will of email.  The email between the 
original sender and Mailman as the recipient is one email transaction. 
What Mailman does with what it receives and send out to other 
subscribers is a series of different email transactions.  I personally 
think trying to squish the two together and treating them as one is 
leading to lots of confusion and / or problems.

> Actually, I'm talking about this kind of thing:
> 
> <snip>
> 
> where the top three Receiveds would really not give me any useful 
> information beyond what I can get from logs in the tsukuba.ac.jp 
> domain.

Hum.  Either you are talking about the first server in the series, or 
you meant to say the bottom three and reference the top (newest) 
Received: header.  (Remember that Received: headers are pre-pended to 
the message.)

The mail logs on uwakimon.sk.tsukuba.ac.jp would record that it received 
the message from shako.sk.tsukuba.ac.jp, but they would not include the 
other Received: headers.  (At least I have never seen an MTA that will 
record all the Received: headers of the messages that pass through it.)

> Stripping three of the four headers might be worthwhile in the OP's 
> case.

I am referring to the MTA receiving the message via SMTP and passing the 
messages to the LDA that 1) archives a copy of the message, 2) strips 
unwanted headers, and 3) passes the message in to Mailman for 
distribution to subscribers.

It is this pre-stripped message in the archive (step 1 above) that would 
have the diagnostic information.

I don't see where this will cause any problems diagnosing / resolving loops.

Please help me better understand your concern.



Grant. . . .