Upgraded to 4.67.6, MailScanner scans a batch then hangs at 100 percent CPU

[Rick Cooper]

 > -----Original Message-----
 > From: mailscanner-bounces at lists.mailscanner.info 
 > [mailto:mailscanner-bounces at lists.mailscanner.info] On 
 > Behalf Of Julian Field
 > Sent: Wednesday, March 12, 2008 5:51 PM
 > To: MailScanner discussion
 > Subject: Re: Upgraded to 4.67.6, MailScanner scans a batch 
 > then hangs at 100 percent CPU
 > 
 > -----BEGIN PGP SIGNED MESSAGE-----
 > Hash: SHA1
[...]
 > >> In which case go into "sub Explode" in 
 > >> /usr/lib/MailScanner/MailScanner/Message.pm, and add some 
 > >> "print STDERR" 
 > >> lines to generate tracing output so you can see how far 
 > it gets. When 
 > >> you do a "MailScanner --debug" it will show you the STDERR 
 > >> debug output 
 > >> in the terminal session.
 > >>     
 > >
 > >
 > > OK, Here is whats happening.  Its using Explode in 
 > MessageBatch.pm and
 > > not Message.pm.
 > > Here is where it dies in MessageBatch.pm:
 > >
 > > sub Explode {
 > >   my $this = shift;
 > >   print STDERR "messagebatch\n";  #crumley
 > >
 > >   my($key, $message);
 > >
 > >   # jjh 2004-03-12 reap as many as we can.
 > >   # JKF Test 2004-11-23 1 until waitpid(-1, &POSIX::WNOHANG) == -1;
 > >   print STDERR "about to hang\n";  
 > >   1 until waitpid(-1, WNOHANG) == -1;
 > >   print STDERR "we never get here\n";  
 > >   
 > But as the comments in the code show, this code hasn't been touched 
 > since 2004. So I don't understand why you are just seeing a 
 > change in 
 > behaviour. I would suspect you have upgraded something else 
 > in your system.


I missed a bunch of this and I could go back and read but I will ask
instead... Have you had a look at what the hanging process is doing yet with
lsof? Particularly lsof +r -p ?




 > 
 > Are other people seeing the same problem?
 > What OS, distro, version, kernel, etc are you running?
 > Is anyone else running an identical system?
 > If so, are they seeing the same symptoms?
 > 
 >  From the "perl-func" man page:
 >        waitpid PID,FLAGS
 >                Waits for a particular child process to 
 > terminate and returns
 >                the pid of the deceased process, or "-1" if 
 > there is no such
 >                child process.
 > so it should reap processes until there aren't any left to 
 > be reaped. 
 > What does the documentation for waitpid say on your system? 
 > This is a 
 > POSIX function, so should be the same across most systems.
 > 
 > If you take out the waitpid() call, you will collect <defunct> 
 > processes, as they are terminating but never being reaped. 
 > So this call 
 > is very necessary.
 > 
 > I'm not going to touch this code with a 10-foot barge pole 
 > unless I have 
 > *very* good reason to.
 > 
 > Jules


--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.