HP3000-L Archives

March 1998, Week 3

HP3000-L@RAVEN.UTC.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
"Calandra, Jack" <[log in to unmask]>
Reply To:
Calandra, Jack
Date:
Thu, 19 Mar 1998 10:09:52 -0500
Content-Type:
text/plain
Parts/Attachments:
text/plain (116 lines)
-----Original Message-----
From:   Calandra, Jack
Sent:   Thursday, March 19, 1998 10:06 AM
To:     [log in to unmask]@-INETGW'
Subject:        RE: Telnet sessions Hanging

Greetings,

I find your topic of much interest to me.  My name is Jack Calandra and I am
the Systems Manager at AIG for the Human Resources Payroll system.  We are
running a 3000k 987/150sx and we are about to upgrade to 5.5 pp4.  Our
payroll application is heavily dependent on Telnet sessions.  This because
our third party software provider, Cyborg Systems uses a Telnet emulator
front end to connect sessions to the payroll package running on the hp via
tcpip.  We also use reflections for Ns/VT.  I am interested in keeping track
of this problem, so any information that you can give me will be much
appreciated.  I am currently using hp open view manager to control the
Telnet sessions on a Telnet card on a dtc72.  We are starting to go over the
80 session limit, so to avoid the additional cost of more hardware (another
dtc72 with a TAC.) we have decided to go with the Os upgrade to handle any
overflow.

I am assuming the problem you are speaking about is with the inbound Telnet
server on 5.5.  Is this also a problem with the TAC on the DTC?

Please keep this thread going.

Thanks
Jack Calandra
H.R. Systems Manager
American International Group
New York N.Y.

                -----Original Message-----
                From:   [log in to unmask]@-INETGW
<mailto:[log in to unmask]@-INETGW>
                Sent:   Thursday, March 19, 1998 6:36 AM
                To:     Alicandri, Matthew; Calandra, Jack;
[log in to unmask]@-INETG <mailto:[log in to unmask]@-INETG>
                Subject:        Re: Telnet sessions Hanging

                In article <[log in to unmask]
<mailto:[log in to unmask]> >,
                  "John W. Nesbitt" <[log in to unmask]
<mailto:[log in to unmask]> > wrote:
                >
                > We are running host-based telnet on our 987/150SX, running
5.5, patch level
                > 2.  I have inetd running in a job in the CS queue (ie.
same priority as
                > sessions).  Most telnet users have Reflection.  I am
getting reports
                > (complaints) of sessions 'hanging': the keyboard freezes
and nothing can
                > reset the session short of aborting it.
                >
                > Are other folks running into this class of problem?
                >

                I have been investigating this problem for the last 4
months.

                I have seen this problem even on the most current patches
applied to the HP.
                I wrote a telnet client to help diagnose the problem, and
when these "hanging
                sessions" occur, the HP's telnet implementation won't even
respond to an "Are
                You There" sequence embedded in a Telnet Synch packet.
Under normal
                circumstances, the HP would respond with the string "YES".

                These hanging sessions usually occur after a TCP retransmit
within the
                session.  Sniffer traces show that TCP recovers properly,
but the session is
                in its frozen state.

                Wanna test it for yourself?  :-)

                An easy way to generate this behavior on your systems is to
tie up the
                network, to force the TCP retransmits.  I use a Ping-Flood
against the
                closest router on route to the HP.  Then, do some
interactive command
                interpreter stuff.  The more bi-directional, the better.  We
used a
                Reflections script to automate this.

                All of the above leads me to conclude that there is a
problem with the HP
                software.  I suspect it's in the socket implementation or in
the telnet
                server implementation -- since a Telnet AYT packet doesn't
even get processed
                by the HP.

                I would be interested in talking with an HP representative
about this, but to
                date neither I (nor my affected clients) have not been able
to get an HP
                resource to discuss this.

                If anyone wants to discuss this further, feel free to drop
me an email (lalee
                at pobox dot com).

                Laurence A. Lee
                Software Engineer

                -----== Posted via Deja News, The Leader in Internet
Discussion ==-----
                http://www.dejanews.com/ <http://www.dejanews.com/>    Now
offering spam-free web-based newsreading

ATOM RSS1 RSS2