Sophie child processes hanging

Mark Robinson <[email protected]> Fri, 19 Mar 2004 16:47:39 +0000
Newsgroups gmane.comp.security.virus.vtools
Organization Reason Technology Limited
Message-ID <[email protected]>
Hello vtools,

  
I'm having trouble with Sophie hanging, and stoppiong delivery of
mail. I'm running amavisd-new and postfix.

The problem seems to be as follows:

Sophie seems to start child processes fine, but when they've been idle
for a while it kills them as follows:

Mar 19 16:34:07 ns0 sophie[5327]: Sophie child has timed-out (no data received in 30 seconds) - process killed
Mar 19 16:34:08 ns0 sophie[5357]: Sophie child has timed-out (no data received in 30 seconds) - process killed
(I brought the timeout down for testing)

However they still appear in the process list:

ns0:/usr/local/packages/sophie-3.04rc2/sample_apps/sock # ps -e | grep sophie
 4068 ?        00:00:08 sophie
 4775 ?        00:00:00 sophie <defunct>
 5327 ?        00:00:00 sophie
 5357 ?        00:00:00 sophie

and no further mail is then scanned.  The messages sit in the queue
with a * against them saying they've been processed.  If I manually
these kill processes using
kill 5327 5357
then the emails are scanned fine and delivered.  These stuck messages
never seem to time out and fail over to command line sweep, so amavisd
must be connecting to the socket ok, indeed the log shows:

Mar 19 16:35:12 ns0 amavis[5325]: (05325-10) Using Sophie: (built-in interface)
Mar 19 16:35:12 ns0 amavis[5325]: (05325-10) Sophie: Sending /var/amavis/amavis-20040319T163241-05325/parts/\n to UNIX socket /var/run/sophie

But it never returns!

Please help.

In the meantime I've set the timeout to 3600 to try to get round the
problem. Can I set an infinite timeout?

TIA,
Mark