HP3000-L Archives

July 1998, Week 1

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:
John Korb <[log in to unmask]>
Reply To:
John Korb <[log in to unmask]>
Date:
Mon, 6 Jul 1998 09:14:49 -0400
Content-Type:
text/plain
Parts/Attachments:
text/plain (106 lines)
From time to time the Navy has had brief problems similar to (but not
exactly the same as) the problem you describe.  It only occurs on PCs which
are one or more routers away from the HP 3000s.

Most of the problems seen here are caused by one or more of the routers
swamping/being swamped by LAN traffic or a LAN segment being swamped by
heavy traffic or bad packets.

In one instance, a router had been broken into and the routing protocol
changed.  Updates were very frequent and the router was basically
continuously sending updates.

In another instance, someone "playing" with one of the routers turned on
IPX pass through and all hell broke loose (and eventually one of the
routers crashed due to an out-of-memory condition).

By far the most common problem has been a NIC going bad.  Some of the
inexpensive NICs seem to go deaf, can't hear other traffic on the segment,
and then send out large volumes of bad packets.  They send and they send
and they send. It seems that all you need is one bad NIC and all the PCs on
the LAN segment (and the routers directly connected to it) get very slow.
One thing we have found - when this problem crops up, telnet becomes VERY
unreliable, drops whole lines at a time, and frequently drops connections,
while NS-VT just gets very slow.

*  What kind of PING times do you get (PC to HP 3000)?  (on a typical
   day the Navy people see ping times of 27 to 64 ms going over a
   4 hop, 56Kb connection).
*  Is there a NOVELL server somewhere on the LAN?  It may be configured
   a TCP server and cause problems (sorry, I'm not a Novell person, but
   some early problems here were pinned on some Novell TCP server feature).
*  Is there an NT server somewhere on the LAN?
*  Is REFLECTION running off the "C:" drive or off the network server? (oh,
   has the Navy had problems running ANY network services off of network
   servers!  they now load networking software on the "C:" drive to boost
   performance)

Quessing that there may be a bad NIC, try the following.  Try turning off
ALL the PCs, then turning one on, logging it onto the HP 3000, running
something to measure the response time, then turning on another PC, logging
it onto the HP 3000, etc.  See if there isn't a point where all of a sudden
everything goes to hell.  If the response time is terrible with the very
first PC, turn it off and try starting with another.  If the response time
is terrible no matter which PC you start with, have your network engineer
check out your routers and your LAN cabling.


You didn't say whether you are using host based Telnet or DTC based Telnet.
 If you are using host based Telnet and running more than 4 or 5 Telnet
sessions, there really isn't a problem, you have just hit the limit of what
host based telnet can support on such a small system.

Also, if you are using Telnet across a WAN, you may find it either very
slow or dropping volumes of data.  A few times I've had to use Telnet to
reach some of the Navy's HP 3000s from home and I've grown to loath Telnet
across the internet (14 hops from home, typically).  The symptoms are that
the character echo takes from a fraction of a second to a minute or more,
all within a 10 minute interval, and that occasionally lines of output (or
characters of input) are lost.  On the other hand, working from home over
NS-VT (with REFLECTION) is far more reliable, far faster, and doesn't loose
data under the same conditions where Telnet is unusable.

Good luck!

John

At 7/5/98 08:26 PM , Lee Courtney wrote:
>Hi folks,
>
>I posted an item recently asking if any other newsgroup readers had
>experienced sluggish performance when using Reflection to access the
>POSIX shell under MPE iX. I haven't seen or received any replies, but
>wanted to post a couple of more data points. We boosted memory in the
>machine (917LX running 5.5) from 64M to 192M with no change in
>performance (at least in regards to this problem). Despite memory
>utilization running at 90%+ we didn't think more memory would make a
>difference, and this was confirmed in practice. Since then I have done
>some tests and notice that direct connections (serial) don't seem to
>have the same problem dropping characters as network (both NS-VT and
>Telnet) connections.
>
>Has anyone else seen the shell dropping characters, type-ahead not
>working, and/or sluggish performance using the POSIX shell thru
>Reflection network connections? If so what was the solution? BTW, I have
>talked to WRQ about this but they were not able to shed much light on
>the problem.
>
>Thanks for any and all feedback!
>
>Lee C.
>--
>Monterey Software Group Inc.              Email: [log in to unmask]
>Suite 300                                 Voice: 408-735-0437
>520 Lawrence Expressway                   Fax:   408-735-8346
>Sunnyvale, California 94086-4025          Pager: 408-237-1705
>
>www.editcorp.com/Businesses/MontereySoftware
>

--------------------------------------------------------------
John Korb                            email: [log in to unmask]
Innovative Software Solutions, Inc.

The thoughts, comments, and opinions expressed herein are mine
and do not reflect those of my employer(s), or anyone else.

ATOM RSS1 RSS2