Wietse Venema via Postfix-users wrote in <4tqc213rcwzj...@spike.porcupine.org>: |So you're suggesting that as long as an MTA-to Milter connection |is not in an error state, sending | | SMFIC_QUIT_NC | |and later sending | | SMTIC_CONNECT | |are sufficient to make a Milter fully forget a past SMTP session and |to make it ready to handle events from a new SMTP session? | |I'd like to see some documentation for that.
Well. That is how QUIT_NC is documented i would say, i munged it to a_SMFIC_QUIT_NC = 'K', /* QUIT but new connection follows (since Sendmail 8.14) */ So this implies that SMFIC_CONNECT is to be expected next. I have $Id: milter-protocol.txt,v 1.6 2004/08/04 16:27:50 tvierling Exp $ which does not yet have _NC. But the flow is * MTA Milter * * SMFIC_OPTNEG * SMFIC_OPTNEG * SMFIC_MACRO:'C' * SMFIC_CONNECT * Accept/reject action * SMFIC_MACRO:'H' * SMFIC_HELO and * SMFIC_BODYEOB ^ 'E' * Modification action (multiple, may be none) * Accept/reject action * * (Reset state to before SMFIC_MAIL and continue, * unless connection is dropped by MTA) * as well as * 'A' SMFIC_ABORT Abort current filter checks * Expected response: NONE * * (Resets internal state of milter program to before SMFIC_HELO, but keeps * the connection open.) plus in addition * 'Q' SMFIC_QUIT Quit milter communication * Expected response: Close milter connection * So if sendmail 8.14 documents QUIT_NC as a_SMFIC_QUIT_NC = 'K', /* QUIT but new connection follows (since Sendmail 8.14) */ and does not close the socket from its point of view, then the above implies that SMFIC_CONNECT is to be expected next. I mean, if you want to be on the safe side, you can always place an 'A'bort command first, which "you do quite often", ie, i have seen Jan 30 01:25:11 (none) s-postdkim-sign: [5017][debug]: read 1 bytes ok, cmd is E/69 Jan 30 01:25:11 (none) s-postdkim-sign: [5017][debug]: WRITE WROTE LENGTH, PAYLOAD 1 Jan 30 01:25:11 (none) s-postdkim-sign: [5017][debug]: LOOP TICK Jan 30 01:25:11 (none) s-postdkim-sign: [5017][debug]: READ SELECTS Jan 30 01:25:26 (none) s-postdkim-sign: [5017][debug]: READ READ LENGTH, PAYLOAD 1 Jan 30 01:25:26 (none) s-postdkim-sign: [5017][debug]: read 1 bytes ok, cmd is A/65 Jan 30 01:25:26 (none) s-postdkim-sign: [5017][debug]: LOOP TICK Jan 30 01:25:26 (none) s-postdkim-sign: [5017][debug]: READ SELECTS Jan 30 01:25:26 (none) s-postdkim-sign: [5017][debug]: READ READ LENGTH, PAYLOAD 1 Jan 30 01:25:26 (none) s-postdkim-sign: [5017][debug]: read 1 bytes ok, cmd is A/65 Jan 30 01:25:26 (none) s-postdkim-sign: [5017][debug]: LOOP TICK Jan 30 01:25:26 (none) s-postdkim-sign: [5017][debug]: READ SELECTS Jan 30 01:25:26 (none) s-postdkim-sign: [5017][debug]: READ READ LENGTH, PAYLOAD 1 Jan 30 01:25:26 (none) s-postdkim-sign: [5017][debug]: read 1 bytes ok, cmd is Q/81 ie two 'A' in a row. I would therefore assume that if a milter negotiates a version number that implies QUIT_NC, it should be able to deal with. Especially since everybody uses libmilter and libmilter is sendmail which invented QUIT_NC. No? Oh that would be wonderful, Wietse Venema. Only ever QUIT_NC, and i keep relying upon spawn(8) for the milter and let everything up to only main.cf and master.cf, and be out of the business of systemd and all that! --steffen | |Der Kragenbaer, The moon bear, |der holt sich munter he cheerfully and one by one |einen nach dem anderen runter wa.ks himself off |(By Robert Gernhardt) _______________________________________________ Postfix-users mailing list -- postfix-users@postfix.org To unsubscribe send an email to postfix-users-le...@postfix.org
