Rsh dies with EADDRINUSE

jgiller has asked for the wisdom of the Perl Monks concerning the following question:

I have a script that loops through a file of ip addresses and pings them. If the ping is successful, it rsh's to the machine and runs a command. I am using Net-Rsh-0.02. Here is some debug output: Errno::EADDRINUSE(/usr/local/perl/5.6.1/lib/5.6.1/sun4-solaris/Errno.pm:153): 153: sub EADDRINUSE () { 125 } DB<1> Errno::FETCH(/usr/local/perl/5.6.1/lib/5.6.1/sun4-solaris/Errno.pm:182): 182: $errno = 0 unless $! == $errno; DB<1> Errno::FETCH(/usr/local/perl/5.6.1/lib/5.6.1/sun4-solaris/Errno.pm:184): 184: return $errno; DB<1> Net::Rsh::rsh(/usr/local/perl-v5.6.1/lib/site_perl/5.6.1/Net/Rsh.pm:42): 42: if($!{EADDRINUSE}) { DB<1> Carp::croak(/usr/local/perl/5.6.1/lib/5.6.1/Carp.pm:119): 119: sub croak { die shortmess @_ } DB<1> Carp::shortmess(/usr/local/perl/5.6.1/lib/5.6.1/Carp.pm:109): 109: { local $@; require Carp::Heavy; } # XXX fix require to not clear $@? DB<1> Carp::shortmess(/usr/local/perl/5.6.1/lib/5.6.1/Carp.pm:109): 109: { local $@; require Carp::Heavy; } # XXX fix require to not clear $@? DB<1> Carp::shortmess(/usr/local/perl/5.6.1/lib/5.6.1/Carp.pm:109): 109: { local $@; require Carp::Heavy; } # XXX fix require to not clear $@? DB<1> Carp::shortmess(/usr/local/perl/5.6.1/lib/5.6.1/Carp/Heavy.pm:247): 247: 1; Seems like rsh is running out of available ports to use? Here is some code

sub runCommands {
  $a=Net::Rsh->new();
    $host="$ip";
      $local_user="root";
        $remote_user="root";
          $cmd="/home/admin/master/pushout/scripts/$program";

  if ( defined($a) )
    {
      $status = $a->rsh($host,$local_user,$remote_user,$cmd);
        if ( ! defined($status) )
          {
            $msg = $rsh->errmsg;
              print "\t$msg\n";
               print LOG "\t$ip\t$msg\n";
                 $rsh->close;
               }
      else
       {
         print PING "$ip  $hostname\n";
            # Run the program
           $a->rsh($host,$local_user,$remote_user,$cmd);
             sleep 1;
           }
    }
  else {
    print "\t$ip\tConnection refused.\n";
      print LOG "\t$ip\tConnection refused.\n";
    }
}
[download]

Any ideas? It barfs after successfully rsh'ing to 5 hosts.

Comment on Rsh dies with EADDRINUSE Download Code

Replies are listed 'Best First'.
Re: Rsh dies with EADDRINUSE by Fletch (Bishop) on Mar 01, 2007 at 19:02 UTC
It's a problem with the r-services protocol itself in that it requires a source port < 1024 (IMSMR). The best solution is to use SSH instead, which doesn't have this limitation and is light years more secure.	[reply]
Re: Rsh dies with EADDRINUSE by merlyn (Sage) on Mar 01, 2007 at 19:12 UTC
/me looks at calendar, sees 2007. "Rsh"? Really, "rsh"? -- Randal L. Schwartz, Perl hacker	[reply]
Re^2: Rsh dies with EADDRINUSE by shandor (Monk) on Mar 01, 2007 at 20:22 UTC
rsh seems to be very popular in the large corporate environment. After my first aneurysm, I decided to ignore that word (and all the glorious security concerns that follow) and pretend that the world is a happy, loving place.	[reply]
Re^3: Rsh dies with EADDRINUSE by Fletch (Bishop) on Mar 01, 2007 at 22:03 UTC
It's popular until you have a wave (wave? gaggle? plague? scourge?) of SOX auditors come through. Then it's persona non grata (much like the auditors . . . :).	[reply]
Re^2: Rsh dies with EADDRINUSE by bluto (Curate) on Mar 01, 2007 at 20:15 UTC
Get out your platform shoes. I know people that use rsh to fire up processes on trusted remote machines this way (e.g. think 1000s of machines connected on a private network). It gets around some authentication overhead, encryption, and having to set things up in something like inetd on each machine (esp if you aren't the admin and politics rule). Sadly, since scaling configuration and performance does become an issue to them, using rsh becomes a very tempting, easy way...	[reply]
Re: Rsh dies with EADDRINUSE by jettero (Monsignor) on Mar 01, 2007 at 18:46 UTC
`$a` is a poor choice for a variable name there... particularly if you don't lexicalize it with my. `$a` is special you see. It has to do with sort and globalness... In any case, you'd probably benifit from closing the rsh when you're done with it. Oh, dear. I just noticed Net::Rsh uses `$a` in it's example and doesn't mention a close method. That module may not be any good — I've never tried it, so I couldn't say, but the documentation is giving it a bad name. -Paul	[reply] [d/l] [select]
Re: Rsh dies with EADDRINUSE by almut (Canon) on Mar 01, 2007 at 19:51 UTC
As background info, you might want to read this discussion related to issues with TIME_WAIT and SO_REUSEADDR in scenarios like you seem to be having... And, if you really, really, REALLY must use something rsh-like for this, you might want to try this guy's implementation (in Perl) of a "multi-rsh" specifically for purposes like these (use at your own risk, though - can't say much about the code quality).	[reply]
Re: Rsh dies with EADDRINUSE by perrin (Chancellor) on Mar 02, 2007 at 17:33 UTC
EADDRINUSE Someone nearby is playing DDR, and the jumping is scrambling the bits. Please stop dancing until the rsh script is finished.	[reply]