MailScanner & new postfix release

Brian Fistler brian at KIMNBRIAN.COM
Wed Mar 24 21:47:58 GMT 2004


I have 2 postfix systems, one was running postfix 2.18 from Mandrake (9.2) rpm
install on an older P4 system 1G ram, single IDE HD(home) (about 3000
messages/day) I got MailScanner set up and working fine (version 4.29.3 if I
remember correctly) (f-prot, spamassassin)

I then installed the same version of MailScanner on a dual xeon p4 2.8 Ghz (with
hyperthreading turned on) 15Krpm RAID 0-1 SCSI drive 2Gb ram system hooked to a
DS3 line handling approximately 6000 email accounts.
This system had postfix 2.18-20040209 (if I remember correctly)
When I would start MailScanner/postfix the cpu load would jump over a 30 second
period from around 1 to 30+ with the system becoming unresponsive around the
12.5 mark.  If I did a killall MailScanner before it reached 10 or so, the
system would recover.

I upgraded to MailScanner 4.29.4 which stopped that behavior but now I have
another problem.
Postfix.in seems to be working fine, it is accepting messages and moving to the
deferred directory.  MailScanner is seeing messages and processing them, but
they never show up in the inboxes.  I get lots of the following errors in the
mail/errors log on the 2nd server.

vmail dccproc[13876]: missing message body; fatal error

postfix/qmgr[7178]: warning: saving corrupt file "51C7515CB65" from queue
"active" to queue "corrupt"

and the following errors in the mail/warnings directory:

postfix/postfix-script: warning: damaged message: corrupt/9C14F15CE35

I upgraded the postfix to 2.19-20040312 & MailScanner to 4.29.5-1 and there is
no change.

I upgraded my home system from 2.18 rpm, to the same 2.19.20040312 version and
now it has the same problem.  (compiles with all default options) (I know, I
know, go back to 2.18 "rpm") (I have not tried installing the 2.18 "official
release" by compiling yet, I may try that tonight if I get a chance.)

It seems to me that postfix.in is processing everything fine, MailScanner is
picking up & scanning everything just fine, and putting it into the incoming
directory of /var/spool/postfix and then postfix is seeing the message as
corrupted.

Everything looks fine when manually starting the postfix.in process with
"postfix -c /etc/postfix.in start" and when starting the MailScanner manually
with check_mailscanner.  It is when I start the out-going postfix with the
"postfix -c /etc/postfix start" command that the errors show up.

Anyone have any clues?

Thanks!

Brian


-------------------------------------------------
This mail sent through IMP: http://horde.org/imp/



More information about the MailScanner mailing list