Richard,
> We have an issue with our mailrelay system.
> Amavisd with spamassassin.
Backend is Postfix with HASH db and mysql database. When the mail starts
coming in, everything looks fine, but after a few minutes we see a lot
of these errors.
----------------------------------
Main.cf :
Works fine :
virtual_alias_maps = hash:/home/postfix/namen
Errors :
virtual_alias_maps = hash:/home/postfix/namen,
mysql:/home/postfix/forwardsqlconnect.cf
----------------------------------
Temporary lookup failure; from=<[email protected]>
to=<[email protected]> proto=ESMTP helo=<localhost>
Jun 26 14:25:27 mx4 amavis[8448]: (08448-02) smtp resp to RCPT (pip)
(<[email protected]>): 451 4.3.0 <[email protected]>: Temporary lookup
failure
The "451 4.3.0 ..." comes from the back-end Postfix instance.
This happens only if the mysql backend is being used. If the mysql is
being turned off in the config everything works as expected. Our setup
is as followes :
--> 25 / Postfix / smtpd
--> 127.0.0.1:10026 / amavisd
After spamcheck
--> 127.0.0.1:10027 / Postfix smtpd
Delivers to final recipient.
We have noticed that recipient translations are being done twice. First
before, and again after the spamcheck. We think the second translation
is going wrong sometimes. Is there a way to disable recipeint check on
second smtpd hop. (More of a question to postfix list)
It is a postfix question. It is possible to have virtual alias
mapping individually enabled/disabled in the front or the backend
instance, although it is a bit of a hackery, as the virtual_alias_maps
setting is property of a cleanup service, so you need to have
two cleanup services, then configure smptd and pickup services
to use one of the other. This is documented in
README_FILES/README.postfix in the amavis distribution.
For simpler cases it may suffice to use:
-o receive_override_options=no_address_mappings
on one smtpd service in master.cf.
Can anyone help with this error ? Mysql database performance issue is
not the case. After extensive monitoring we found that max connections
and memory is fine.
Definitely looks like an issue with MySQL. No idea there.
Mark
