Noel Jones wrote:
Steve Crawford wrote:
While we do not manage a "mailing list" in the traditional sense, we
do send a lot of emails (daily/weekly/monthly reports, instant-alert
messages, etc.) where using VERP to track bounces could prove useful.
We run numerous projects with varying requirements and the messages
could be generated by Perl, Python, Bash/mutt, etc. All
inbound/outbound mail is routed through Postfix.
In my ideal world, I would find a magic setting that said, say, "use
VERP for all messages from f...@pinpointresearch.com" or perhaps from
some specific IP range. Digging through the docs and list archives
indicates this does not exist....
So I'm seeking recommendations/ideas...
Seems to me that if you want to use SMTP and you control the
submitting software it should be easy enough to add XVERP to the MAIL
FROM command:
http://www.postfix.org/VERP_README.html#smtp
Long delay from last message while out of town.....
I had read that and it is easy if I netcat into Postfix and run the
commands manually. But the other utilities do not understand XVERP. I
tried fooling mutt into adding it but have been unable to get them to
add the XVERP to the FROM - the usual result is that they misinterpret
the XVERP as an additional from address.
If there were a _separate_ header, say "X-VERP: true", triggering VERP
delivery would be easy with most of the utilities. The problem is that
they don't understand the modification to the FROM command and try to
munge and/or correct it into something it believes is correct. (If I'm
missing something that is available in the standard mail utilities in
Perl/Python/Ruby or mutt, please let me know.)
It wouldn't be safe to do this by sender address, but you could use a
check_client_access table with a FILTER result pointing to a pipe
transport that does a "sendmail -XV ..." reinjection.
Actually, in our case all system-delivered messages, reports etc. come
from a single standard address and VERPifying all of them would not hurt
anything.
I haven't had much time to experiment with re-injection but my
experiments thus far have failed. Any pointers on how this would be
done? One possibility that would work for me is to have a VERPifying
"pre-server". All messages going to that server would be VERPified and
forwarded to the main mailserver. Any process needing VERP would simply
send to that server instead of the main server.
Cheers,
Steve
