home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
kermit.columbia.edu
/
kermit.columbia.edu.tar
/
kermit.columbia.edu
/
newsgroups
/
misc.19990422-19990725
/
000099_news@watsun.cc.columbia.edu _Wed May 26 13:52:52 1999.msg
< prev
next >
Wrap
Internet Message Format
|
1999-07-23
|
3KB
Return-Path: <news@watsun.cc.columbia.edu>
Received: from newsmaster.cc.columbia.edu (newsmaster.cc.columbia.edu [128.59.59.30])
by watsun.cc.columbia.edu (8.8.5/8.8.5) with ESMTP id NAA18655
for <kermit.misc@watsun.cc.columbia.edu>; Wed, 26 May 1999 13:52:52 -0400 (EDT)
Received: (from news@localhost)
by newsmaster.cc.columbia.edu (8.8.5/8.8.5) id NAA18554
for kermit.misc@watsun.cc.columbia.edu; Wed, 26 May 1999 13:50:05 -0400 (EDT)
X-Authentication-Warning: newsmaster.cc.columbia.edu: news set sender to <news> using -f
From: dold@99.usenet.us.com
Subject: Re: kermit process hangs around after terminal disconnect
Date: 26 May 1999 17:49:07 GMT
Organization: a2i network
Message-ID: <7ihc6j$7s2$1@samba.rahul.net>
To: kermit.misc@watsun.cc.columbia.edu
Jeffrey Altman (jaltman@watsun.cc.columbia.edu) wrote:
: If you want the host to detect that the connection is no longer
: valid, the telnetd will have to be modified to attempt to send some
: message on a periodic basis to test the connection. This could be
: a telnet go ahead or a telnet data mark message. Anything that the
: client will process and not respond with real data.
One of my customers offered Convergent computers to college libraries
around the country, and they could check other campus' inventories.
PCs would 'telnet' in and check stuff, and then quit improperly. I think
it was early Macs that were at fault. Eventually there wouldn't be any
more telnet processes/ports available for fresh logins. I recall that the
limit by default was either 16 or 32.
Telnetd was modified for this customer, but the mods were never moved into
SVR2 releases. Later, this same customer changed what telnet did when it
did get a proper disconnect. I don't recall the detail, but it was a
fairly polite kill to child processes, that he wanted changed to a
buffer-dumping merciless kill.
Like lots of other things in unix, the default is very generic. If your
situation has tighter requirements, it is possible to add more layers,
without affecting the underlying stateless connection.
I wrote a very simple 'killidle' that is invoked at login time, and
monitors the "tty" associated with the login. If there's no activity for
five minutes, it does a programmatic shutdown of the process that it knows
was started... In my case, I know what the process is, and that it is okay
to terminate it. That can't be the case generically.
And, if you are using Procomm to AIX, then kermit out to a modem
server, you should probably be doing the connection directly to the
modem server in the first place.
--
---
Clarence A Dold - dold@network.rahul.net
- Pope Valley & Napa CA.