a very confused mailscanner?

[Julian Field]

At 23:14 10/11/2003, you wrote:
>Hi there,
>
>We've recently set up a testbed to test mailscanner under constant medium
>to heavy usage scenarios.  After a particularly good beating mailscanner
>went out to lunch with:
>Nov 10 17:10:35 d3 nosa_MailScanner[834]: Could not open
>file >/var/spool/mqueue/nosa/mailscanner_work/834/hAALp9dR077394.header:
>Too many open files
>Nov 10 17:10:35 d3 nosa_MailScanner[834]: Cannot create + lock headers
>file /var/spool/mqueue/nosa/mailscanner_work/834/hAALp9dR077394.header,
>
>
>At first I thought it was insufficient file descriptors available so I up'ed
>that limit and rebooted.  No effect.  The in queue has only 1400 emails to
>be processed, system is idle, tons of all resources available.

Sometimes the shell started to run the check_mailscanner script imposes its
own default limits on the number of open file handles. Modify
check_mailscanner and put the necessary ulimit commands in there.
The other option is that you are hitting a system-wide limit.

You can usually prove whether this is the problem by reducing the number of
messages per batch and the number of child processes.

Doing a systune (guessing you are running IRIX :-) produces this:
  group: resource (statically changeable)
         rlimit_pthread_max = 1024 (0x400) ll
         rlimit_pthread_cur = 1024 (0x400) ll
         rlimit_nofile_max = 2500 (0x9c4) ll
         rlimit_nofile_cur = 200 (0xc8) ll

What is the betting that rlimit_nofile_max == maximum number of files open?

Note this from /var/sysgen/mtune/kernel
* Note: rlimit_nofile_max must not be set to an unreasonably large value
* since many daemons/programs use rlimit_nofile_max as an indication
* of how many file descriptors to close when they want to close them all.

--
Julian Field
www.MailScanner.info
MailScanner thanks transtec Computers for their support

PGP footprint: EE81 D763 3DB0 0BFD E1DC  7222 11F6 5947 1415 B654