On Thursday 09 February 2006 16:30, mouss wrote: >Gene Heskett a écrit : >> On Thursday 09 February 2006 03:47, jdow wrote: >>>From: "Gene Heskett" <[EMAIL PROTECTED]> >>> >>>>>Procmail calls SpamAssasin and feeds the return off to the spool >>>>>file. >>>> >>>>Ok, sub getmail for both fetchmail and procmail, since getmail can >>>>handle the SA pipeing you are doing with procmail. Then run >>>> dovecot on that box to serve kmail on this box? I have the kmail >>>> fetching turned off on that box, so I'd assume I can give getmail >>>> a trial run and see if what it fetches it shows up in kmail on >>>> that firewall box as a new mail, if that works, then setup dovecot >>>> as a pop3 server to serve the kmail requests from this box. Have >>>> I got it right? All running as the user gene I'd assume? >>> >>>Only if getmail combines fetchmail and procmail including procmail's >>>ability to write rules for redirecting mail or applying filtering to >>>it. >>> >>>Kmail would simply read from the imap port you create. It's still >>>write to your ISP's mail server. >> >> First, getmail is out as far as putting it on the FW box, the python >> install there is several releases too old. >> >> Further reading on fetchmail tells me that it hands the incoming >> mail off to sendmail via stuffing it into port 110. Datapoint as I >> try to get my head around the mechanics of this. >> >> It is sendmail then that listens on port 110 and writes to >> the /var/spool/mail/username file. Datapoint again. >> >> So there is a potential place to put a |spamc| is there not?, >> between fetchmail and port 110? A hack to fetchmail maybe? >> Datapoint. >> >> Experimental results.. I grabbed a copy of /var/spool/mail/gene to >> another file while it had some content, then did a 'cat >> filename|spamc >> >>>filename2' >> >> This did properly scan & add the headers that it had done so to the >> first of the 3 messages that were merged into filename, but did not >> re-trigger itself on the next 2 messages also in that file, >> therefore they were not scanned and marked up by spamc. >> >> So it appears that wherever spamc is inserted into the path, it must >> be presented with a single message complete with an EOF indicator of >> come kind. Looking at the src file, it doesn't appear there is a >> quick, dirty, and 100% dependable way to filter the output of the >> cat command and break it up into one stream per message. But I >> haven't ran a tcpdump to see how its formatted on the network >> traffic yet. The only thing I can see is linefeed,linefeed,From and >> since theres no way to stop me from doing it in a message I send, it >> doesn't look that reliable to me. >> >> From >> >> Is that treated as a new message? I think not. I'm going to go >> look at the fetchmail code, maybe I can make a patch for it to do >> this. > >once fetchmail has read the message, it can: > >- put it in files. this is what you do.
From re-reading a 'man fetchmail' I don't see the fileing ability. It only presents it to localhost:25 and apparently sendmail takes it from there. The comm thru port 25 is apparently bilateral as it can be told to summarily delete unwanted mail from the server, while sendmail at the some time is deleting its copy. Or at least thats how I'm reading it. >- run an MDA. so you could run procmail or maildrop or a (correct) >script. In short, fetchmail runs a command (it pipes the message). eg sendmail?, which is running here. >- forward to an smtp server. This is the simplest to configure if you >can afford to run an MTA. > >you'll need to choose which method is appropriate for your situation. > if you don't feel yourself installing an MTA (this is not difficult, > but requires some efforts to do it correctly), then go for the MDA > method. reread fetchmail docs in both cases. In further reading tonight, sendmail grew the libmilter freature at 8.12, which is the base version running here, and yum won't update it, says its current. Right now, I'm looking at the <http://www.bmsi.com/python/milter.html> site, trying to see how this is done. But, here is the headache: At no place in the various files sitting in /etc/mail that serve to configure sendmail, is there an example of how to configure sendmail to make use of these feature facilities. Spamassassin 3.10 contains only very scant references to using it with sendmail, apparently sanctioning only the procmail interface, which in turn then is set to call spamc or spamassassin, adding needless time wasting cpu cycles to what should be a pretty simple job. I fail to understand why (although it will take smarter people than me what with sendmails configuration complexity) there is no readily published recipe for incorporating spamc into the sendmail processing chain, either by pipeing, or when the libmilter feature is there? Or am I simply on the wrong mailing list? I've sent 3 subscribe messages to the getmail-user list over the last 3 days with no response which is discouraging. OTOH, now that I know it can't do what I want, who cares. It might be that if there was a manpage for getmail, it might be possible. A pox on software that doesn't come with readable manuals. Back to perusing that web page I guess, sigh. -- Cheers, Gene People having trouble with vz bouncing email to me should add the word 'online' between the 'verizon', and the dot which bypasses vz's stupid bounce rules. I do use spamassassin too. :-) Yahoo.com and AOL/TW attorneys please note, additions to the above message by Gene Heskett are: Copyright 2006 by Maurice Eugene Heskett, all rights reserved.
