Possible bug in domain lookups
\ Peter Rabbitson (28 Jul 2007)
. \ mouss (28 Jul 2007)
. \ Victor Duchovni (28 Jul 2007)
. . \ Peter Rabbitson (29 Jul 2007)
. \ Victor Duchovni (29 Jul 2007)

6 msgreject_unknown_client different on different dokus
1 msgLDAP lookup for relay_domains problem (again)
3 msgsend mail limits
2 msgqmail -> postfix
5 msgAnother small policy server question
2 msgLocal Delivery of only a few users
2 msgBouncing instead of rejecting?
1 msgMore of spamassassin+amavisd question but...
3 msgusing different hostname when sending mail
13 msgPIX problems with DKIM header fields
6 msgsend mail thru HTTP proxy
7 msglocal_recipient_maps: prevent the sender that t...
22 msgPostfix 2.3 or 2.4 on Centos 4.4 via yum?
4 msgEmail monitoring
10 msgconnect to 127.0.0.1 refused
4 msgPostfix + 2 amavis instances sends forwarded me...
3 msgA Thank you to the Developers for the Free Soft...
1 msgUser Bulletin implementable with Postfix
2 msgPostfix | Dovecot | LDAP (and perhaps qmail.sch...
Subject:Possible bug in domain lookups
Group:Postfix-users
From:Peter Rabbitson
Date:28 Jul 2007


 
Hello,

I have a Postfix/DBMail server, so that DBmail is the sole
*virtual_transport* in the system. I am using a mysql lookup for
*virtual_mailbox_maps* to validate existing users. It all works
wonderfully.

Since I am not using any catch-all users 'user@' I figured it will be
nice to make *virtual_mailbox_domains* a mysql lookup as well. Here is
my table:

user = dbmail
password = XXXXXXX
hosts = 127.0.0.1
dbname = dbmail
query = SELECT `alias` FROM dbmail_aliases WHERE `alias` like '%%@%s'

And it works just as expected. The problem is that when I stopped the
mysql server to see what impact it would have, every new smtpd process
blocked indefinitely. After about 2 minutes I eventually got this:

Jul 28 08:54:56 Arzamas postfix/master[4191]: warning: service "smtp"
(25) has reached its process limit "100": new clients may experience
noticeable delays
Jul 28 08:54:56 Arzamas postfix/master[4191]: warning: to avoid this
condition, increase the process count in master.cf or reduce the service
time per client

I started back mysql - a rush of queries came in, and the server kept
working as before.

I looked in the docs for any sort of table lookup timeout, but did not
find anything. Did I do something wrong or is this a problem with postfix?

By the way the same does not happen with the *virtual_mailbox_maps*
lookup: query = SELECT `alias` FROM dbmail_aliases WHERE `alias` = '%s'.
If the mysql server is unreachable the client immediately receives:

451 4.3.0 <address>: Temporary lookup failure


Thank you for your help.

Peter


© 2004-2008 readlist.com