<<<No Message Collected>>> [Was Re: SYSERR(root): readqf: cannot open ./df*]

[Julian Field]

At 10:42 16/12/2002, you wrote:
>I've started a new thread as this is clearly a different problem (despite
>the similar error messages)
>
>Quick recap...
>Two messages appearing in error log:
>Dec  9 10:53:59 mailhost sendmail[19633]: gB9Arxid019633: SYSERR(root):
>cannot flock (/etc/mail/access.db, fd=7, type=1, omode=100000, euid=0): No
>locks available
>Dec 16 10:04:56 mailhost sendmail[8497]: gBGA4pjZ008478: SYSERR(root):
>readqf: cannot open ./dfgBGA4pjZ008478: No such file or directory
>
>Of these the second is the most common & ties up with messages users
>recieve which contain '<<<No Message Collected>>>' instead of the message body
>
>Tried so far...
>splitting mailscanner init script into two to seperate sandmail and
>mailscanner init process
>adding ulimit -n 4000 (and later ulimit -n 8000) to both mailscanner and
>sendmail init scripts
>Overhaul of sendmail.cf (via sendmail.mc) removed some redundant stuff
>Compiling sendmail with fnctl rather than flock (have now reverted to flock)
>
>More observations (some new!)...
>access.db flock message is not generated by one of the two 'normal' parent
>sendmail processes.
>The messages mentioned in the readqf error are logged at level info as
>having two consequtive (or near) delivery attempts which are both logged
>as 'Sent' (as follows - some personal/site information obfuscated)
>
>[ERROR LOGGED]
>Dec 16 10:30:36 mailhost sendmail[13099]: gBGAUTjZ013054: SYSERR(root):
>readqf: cannot open ./dfgBGAUTjZ013054: No such file or directory
>
>[DELIVERY ATTEMPTS FOR THIS MESSAGE]
>Dec 16 10:30:36 mailhost sendmail[13099]: gBGAUTjZ013054:
>to=<REMOVED at bmrb.co.uk>, delay=00:00:07, xdelay=00:00:00, mailer=esmtp,
>pri=120695, relay=exchange.bmrb.co.uk. [XXX.XXX.XXX.XXX], dsn=2.0.0,
>stat=Sent (
><8CDB922F2BB1D6119F5400508BBE1F0E011E6FE7 at XXXXXXX.XXXXXXXX.net> Queued
>mail for delivery)
>Dec 16 10:30:36 mailhost sendmail[13107]: gBGAUTjZ013054:
>to=<REMOVED at bmrb.co.uk>, delay=00:00:07, xdelay=00:00:00, mailer=esmtp,
>pri=120695, relay=exchange.bmrb.co.uk. [XXX.XXX.XXX.XXX], dsn=2.0.0,
>stat=Sent ( <8CDB922F2BB1D6119F5400508BBE1F0E011E6FE7 at XXXXX.XXXXXXXX.net>
>Queued mail for delivery)
>
>[PS FOR THESE PID'S - did a ps | grep > file; sleep 1 loop to catch these]
>140 S root     13099     1  3  69   0    -  1406 do_sel 10:30
>?        00:00:00 sendmail: ./gBGAUXjZ013084 XXXXXXXXXXXXXXXXX: client
>DATA status
>140 S root     13107     1  1  69   0    -  1408 do_sel 10:30
>?        00:00:00 sendmail: ./gBGAUTjZ013053 XXXXXXXXXXXXXXXX: client DATA
>status
>
>This shows that both delivery attempts were from sendmails that were not
>children of one of the master processes - so why are there two consequtive
>delivery attempts for the same message?
>I've also observerd that this only affects incoming messages to our domains.
>
>Setup...
>Mandrake 9.0
>sendmail 8.12.6
>mailscanner 3.23

What happens if you reduce the maximum batch size to 50 messages?
What %-age of messages going through the gateway do you actually scan? All
of them or only a small %-age?
--
Julian Field
www.MailScanner.info
MailScanner thanks transtec Computers for their support