Charlie Watts wrote:

> You said you're using LDAP? Perhaps you're getting too many simultaneous
> connections to the LDAP server and it's rejecting new ones? That's a total
> guess. I haven't used LDAP for this yet. And so maildrop is just
> delivering into its "best guess" in /var/spool/mail...
>
> If you set "maildrop_destination_concurrency_limit = 1" does that fix it?
> It will slow things down ... maybe enough to fix it. If so you have some
> capacity problem.
>

Thinking about this, I've increased the amount of seconds before timing out in LDAP
lookups, to check what happens. So I've added to my main.cf:

    ldapsource_timeout = 100 (default is 5)

Sure it's too much, but it will help to check it out. By the moment, I've got no more
mails in /var/spool. Well, I'll better wait some more time... If this works, the
explanation is obvious: the cpacity of the system is not enough and it gets overloaded
when having to deliver a mail to a lot of local users, so responses from LDAP are too
slow and time out.

I hope not to lose more mails... :-)
Thanks everyone for your help!
Regards,

    Albert

> This is definitely not a postfix problem ... go to the (oops, there isn't
> a maildrop list.) Go to the courier list.
>
> On Wed, 23 Jan 2002, Albert Martorell wrote:
>
> > Here are the logs:
> >
> > Jan 23 13:29:22 hostname postfix/qmgr[603]: 6A0013BD04:
> > from=<origin_userdomain.dom>, size=5635, nrcpt=50 (queue active)
> >
> > Jan 23 13:29:23 hostname postfix/pipe[30354]: 6A0013BD04: to=<user1domain.dom>,
> > relay=maildrop, delay=22, status=sent (hostname.domain.com)
> >
> > Jan 23 13:29:23 hostname postfix/pipe[30328]: 6A0013BD04: to=<user2domain.dom>,
> > relay=maildrop, delay=22, status=sent (hostname.domain.com)
> >
> > Jan 23 13:29:23 hostname postfix/pipe[30353]: 6A0013BD04: to=<user3domain.dom>,
> > relay=maildrop, delay=22, status=sent (hostname.domain.com)
> >
> > Jan 23 13:29:23 hostname postfix/pipe[30355]: 6A0013BD04: to=<user4domain.dom>,
> > relay=maildrop, delay=22, status=sent (hostname.domain.com)
> >
> > Jan 23 13:29:24 hostname postfix/pipe[30380]: 6A0013BD04: to=<user5domain.dom>,
> > relay=maildrop, delay=23, status=sent (hostname.domain.com)
> >
> > Jan 23 13:29:31 hostname postfix/pipe[30393]: 6A0013BD04: to=<user6domain.dom>,
> > relay=maildrop, delay=30, status=sent (hostname.domain.com)
> >
> > Jan 23 13:29:31 hostname postfix/pipe[30394]: 6A0013BD04: to=<user7domain.dom>,
> > relay=maildrop, delay=30, status=sent (hostname.domain.com)
> >
> > Jan 23 13:29:31 hostname postfix/pipe[30384]: 6A0013BD04: to=<user8domain.dom>,
> > relay=maildrop, delay=30, status=sent (hostname.domain.com)
> >
> > Jan 23 13:29:31 hostname postfix/pipe[30391]: 6A0013BD04: to=<user9domain.dom>,
> > relay=maildrop, delay=30, status=sent (hostname.domain.com)
> >
> >
> > As a result, user1...5 got the mail ok at their homes, but mail for user6...9 went
> > to /var/spool.
> >
> > Albert
> >
> >
> > jcarminatipluspetrol.com.ar wrote:
> >
> > > Albert:
> > >
> > > As it's always said here, prove it, show the logs...
> > >
> > > Jorge Carminati.
> > >
> > > >
> > > >I'll try to explain it with an example: when sending a mail to a list of local
> > > >users, the first of them receive it ok, but the last ones don't. Those lost
> > > >messages stay at /var/spool/mail. These messages are exactly the same for all
> > > >users, so the only difference that I can see is the increasing delay shown in
> > > the
> > > >logs.
> > > >
> > > >This behaviour is independent of the origin of mail (local or from another
> > > >machine).
> > > >
> > > >Messages sent to a single user never get lost.
> > > >
> > > > Albert
> > > >
> > > Wietse Venema wrote:
> > >
> > > > How does a "lost email in /var/spool/mail" differ from mail that
> > > > is not "lost"?
> > > >
> > > > Wietse
> > > >
> > > > Albert Martorell:
> > > > > Adding "maildrop_destination_recipient_limit = 1" in main.cf solved part of
> > > > > the problem, so I thought that I might be missing something else in postfix
> > > > > rate controls.
> > > > >
> > > > > Mails that appear at /var /spool/mail are multiple-local-users destined, so
> > > > > their delays increase with every delivery. Loss of mails seems to be related
> > > > > to this effect.
> > > > >
> > > > > Albert
> > > > >
> > > > >
> > > > > Wietse Venema wrote:
> > > > >
> > > > > > Postfix hands the mail over to the maildrop program. Postfix logs
> > > > > > that maildrop completed successfully. Yet you believe that it is
> > > > > > Postfix that loses the mail?
> > > > > >
> > > > > > Wietse
> > > > > >
> > > > > > Albert Martorell:
> > > > > > > Hi all,
> > > > > > >
> > > > > > > I've been running Postfix + Maildrop + Courier IMAP/POP + LDAP on a
> > > > > > > RH6.2 for two weeks, with maiboxes at /home/directories in maildir
> > > > > > > format. All seems to work fine, except that every day some "lost" mails
> > > > > > > appear in the /var/spool/mail. Users affected vary randomly and senders
> > > > > > > of these mails are indistinctlly local an not local. Permissions in home
> > > > > > > directories are ok.
> > > > > > >
> > > > > > > In the maillog, nothing strange seems to happen, but some delay:
> > > > > > >
> > > > > > > Jan 22 11:12:04 hostname postfix/pipe[1495]: 4E49B3BD10:
> > > > > > > to=<userdomain.dom>, relay=maildrop, delay=425, status=sent
> > > > > > > (hostname.domain.dom)
> >
> > -
> > To unsubscribe, send mail to majordomopostfix.org with content
> > (not subject): unsubscribe postfix-users
> >
>
> --
> Charlie Watts
> cewattsfrontier.net
> Frontier Internet, Inc.
> http://www.frontier.net/

-
To unsubscribe, send mail to majordomopostfix.org with content
(not subject): unsubscribe postfix-users

• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]