Hello,
Paragraph 3.1 of the exim spec states that "The first form (ie the
local-part@domain form) must be used only with domains."
I understand this, but nonetheless I'd like to use this first form
also for remote domains, in a special case: I'd like to get
quickly rid of all messages with the local part "news".
My problem is, my MTA acts as a relay for the news moderated
newsgroups. Mails come in in the form news-group@???.
I have a long list of all the news-group aliases (today 5158 aliases).
The local part "news-group" is checked against the aliases list,
then the mail is forwarded to the maintainer of the newsgroup,
which is always a remote domain. Most of the times, mail
gets delivered, but from time to time one of these aliases is
no longer valid (fortunately, the alias list gets updated regularly)
or the remote site has implemented an anti-spam mechanism
at SMTP transaction time.
Quite normally, exim then generates an error message, which must
be sent back to the originating address. Unfortunately, this
originating address is some generic address from the sending
NNTP host (that originally sent the moderated newsgroup
message). Most of the time, the sending NNTP host does not
accept SMTP traffic, or if it does, the sending address is not
supported. Which leaves such error messages stuck in my queue
until normal timeout.
By the sheer quantity of this traffic, it is impossible to
educate the responsible NNTP newsmasters.
I could set-up a second instance of exim, that would take care
of just the moderated newsgroup traffic. However, before
setting this up, I'd like to be sure that there is no
less heavyweight solution around. The feature "local part
lookup for remote domain" regarding retry configuration would
do just that. Also, when doing exim -brt news@remote-domain,
the rule gets "correctly" interpreted:
^news@([^@]+)$ refused F,2h,15m; G,6h,2h,1.5
^news@([^@]+)$ * F,2h,15m; G,6h,2h,1.5; F,12h,1h
exim@chx400 /home/exim/cfg> exim -brt news@???
Retry rule: ^news@([^@]+)$ F,2h,15m; G,6h,2h,1.5; F,12h,1h;
How should the spec be interpreted (ie "must not")? Is this
just impossible by the code, or is it a recommendation, and
in this latter case, what am I doing wrong? Thank you for any
hint.
Regards,
Bernard Stern, SWITCH
____________S_W_I_T_CH___Swiss Academic_______________________________________
mail: SWITCH Head Office a Tel: +41 1 268 1520
Limmatquai 138 n Fax: +41 1 268 1568
CH-8001 Zurich d e-mail: stern@???
________________________________________Reseach Network_______________________
