Forum OpenACS Q&A: can't grep nsd with ps aux...

I worked around this a while ago by altering restart-aolserverto use the servername.nspid files. This way even swapped out nsd's can be restarted. I'll post it if anyone's interested.

You might try using the daemontools package to control AOLserver. I have a short HOWTO at http://zmbh.com/discussion/svc/aolserver+daemontools.html that might help. It also has some specific help for replacing restart-aolserver.

#!/usr/bin/perl
#
# $Id: restart-aolserver,v 1.2 2001/07/20 08:57:30 andy Exp $
#
## Restarts an AOLserver. Takes as its only argument the name of the server to kill.
## bquinn 6/16/2000 with help from {ryanlee, doug}@arsdigita
## This is a perl script because it needs to run setuid root,
## and perl has fewer security gotchas than most shells.
##
## Make sure that $PIDFILE points to the right location.


use strict;
undef %ENV;
$ENV{'PATH'} = '/sbin:/bin';

if (scalar(@ARGV) == 0) {
     die "Don't run this without any arguments!";
}

my $server = shift;
$server =~ /^([w-]*)$/;
my $service_name = $1;
my $PIDFILE = "/web/aol3/log/nspid.$service_name";
my $pid;

$< = $>; # set realuid to effective uid (root)

# Get the PID of the process to kill.

open(IN,"$PIDFILE") || die "No such server
";
while(<IN>) {
    chomp($_);
    $pid=$_;
}
close(IN) || die "Problem closing PID file
";

# Remove the PID file.  We have to delete the file to make sure that a subsequent call 
# to this script will kill some other process.  We delete the file before the process dies
# because if the service is set to respawn then we may delete the new pid file.

# This removing of the PID file is NOT necessary, and in leads to bugs
# as it will remove the PID file even if the kill of the process then
# fails.  Therefore, comment it out entirely.  --atp@piskorski.com,
# 2001/06/27

#my $cmd ="rm -f $PIDFILE";
#$cmd =~ /^(.*)$/;
#my $untaint_cmd = $1;
#`$untaint_cmd`;

# Issue the kill
$pid =~ /^(.*)$/;
my $untaint_pid = $1;
print "Killing $untaint_pid
";
kill 9, $untaint_pid;

# TODO: This is not so good.  We should be trying a normal kill first,
# and then ONLY using kill -9 if the normal kill fails.
# --atp@piskorski.com, 2001/06/27