NetNews Usenet Archive 1992 #18/NN_1992_18.iso/spool/comp/unix/aix/8828

Internet Message Format | 1992-08-14 | 3.4 KB

Path: sparky!uunet!stanford.edu!unixhub!unixhub.slac.stanford.edu!alhy
From: [e-mail] (J. Scott Berg)
Newsgroups: comp.unix.aix
Subject: Re: xterm: no available ptys
Message-ID: <[e-mail]>
Date: 15 Aug 92 00:10:07 GMT
References: <[e-mail]> <[e-mail]>
Sender: [e-mail]
Reply-To: [e-mail]
Organization: Stanford University, dept of Physics / SLAC
Lines: 77
Nntp-Posting-Host: courant.slac.stanford.edu

In article <[e-mail]>, [e-mail] (Mark Rahner) writes:
|> [e-mail] (Bjorn Ivar Stark) writes:
|> : When I try to start more than a certain limit of xterms I get
|> : this error-message:
|> :     fosli@asterix: xterm
|> :     xterm: no available ptys
|>
|> This has given us massive headaches.  Our system administrator has
|> increased one of our RS/6000s to what he says is the limit of 64.
|> Sorry, but I don't know how.
|>
|> When we first reported this to IBM, IBM said "Xterm is provided on the
|> RISC/6000 as a binary sample only and is unsupported."  (More on this in
|> a minute.)  After explaining to the #!%&*s that the problem wasn't with
|> xterm, but with their system chewing up ptys, they reexamined the problem
|> and determined that the software that controls our many X Stations was
|> not properly releasing the ptys when the X Stations were being turned
|> off.  As yet, this has not been fixed.
|>

If your problem is what I think it is, the following fix MAY (no
guarantees) work:  before exiting mwm (I'm assuming that's what you're
running), execute the following perl script:

#!/usr/local/bin/perl
#
# Exit gracefully from an x windows session on an RS6000--i.e., don't leave
# behind any csh's.  At the same time, don't kill anything unnecessarily.
#
$indx = 0;
# Get a list of processes
open(psout,"ps lgxw |");
while (<psout>) {
  $psout[$indx++] = $_;
}
close(psout);
$cmd =  "kill -9";
for ($i = 0 ; $i < $indx ; $i++) {
  @psline = split(/ +/,$psout[$i]);
  if (("$psline[3]" eq "$<") && ("$psline[5]" ne "$$") &&
      ("$psline[11]" eq "EVENT") && !($psline[12] =~ /hft\/[0-9]/) &&
      (($psline[14] =~ /^-?[bck]?sh/) || ($psline[14] =~ /^-?tcsh/))) {
    for ($j = 0 ; $j < $indx ; $j++) {
      @xtst = split(/ +/,$psout[$j]);
      last if (("$psline[5]" eq "$xtst[4]") && ($xtst[13] =~ /^xterm/));
    }
    if ($j != $indx) {
      $cmd = "$cmd $psline[4]";
    }
  }
}
exec $cmd;


Here's what it does:  it does a ps lgxw to find all the shells that
are children of xterms and have this "EVENT" state set, and kills
them.  These seem to be the things that are hanging around that are
eating pty's.  As best as I can figure, the problem is in mwm, not the
server.  That's a guess (running R3 or R5 server with twm doesn't seem
to exhibit the problem at the moment).

I hope this is what you are looking for.  Good luck, and God bless

                -Scott Berg

--
-------------------------------------------------------------------------------
The opinions expressed here are, of course, my own and nobody else's.
-------------------------------------------------------------------------------
J. Scott Berg
email: [e-mail]
real mail: Varian Physics
           Stanford  CA  94305-4060
phone:    [phone] (w)
     [phone] (h)
-------------------------------------------------------------------------------