HP3000-L Archives

April 1999, Week 5

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:
"Stigers, Greg [And]" <[log in to unmask]>
Reply To:
Stigers, Greg [And]
Date:
Thu, 29 Apr 1999 15:31:23 -0400
Content-Type:
text/plain
Parts/Attachments:
text/plain (37 lines)
X-no-Archive:yes
Ron Horner asks a valid question: what changed? Getting the answer may not
be easy, and could actually be impossible, if what changed is not
configuration, or at least not immediately obvious configuration, but usage.


Some things I found helpful while troubleshooting our ftps across congested
and contorted LANs and nationwide WANs thru a firewall and with lots and
lots of frame relay were pinging with increasing packets sizes between the
two machines. This can be done inside ping.net, or from the CI as
ping.net.sys 'ipaddress,number of packets,packetsize'. Some connections
restrict packet size to less than MPE's 2048 limit, so failure above certain
numbers, usually 1024 to 1536, * could * just be hitting that limit. Another
technique is to telnet between the two systems, and LISTFILE /ACCOUNT/,2 or
something similar - start small. This starts a non-trivial stream of data
between the two systems, and herky-jerky, off and on scrolling text with
disturbingly long pauses indicates that something is impeding traffic
between the two systems; hangs and disconnects even more so.

These will at least help quantify the problem, and may get someone to admit
that something isn't working that really ought to be, so they can at least
start looking for a cause.

3000 side issues worth looking at (at least they were for us) are your MPE
version and patch levels and inbound and outbound memory buffers. When we
increased these, certain problems went away. Say you have four times as much
memory as the machine ships with. Increase these buffers by a factor of
four, assuming the machine is not memory bound, in which case this could
hurt other performance, and that these ftps are production. HPRC was quite
helpful with these.

I would also be suspicious of the frame relay connection. Not every provider
provides what they would like you to think they are, and not all
administrators configure correctly. If your provider can have a look at what
they see, and is honest about it, they can at least help provide diagnostic
information if the problem is not theirs.

ATOM RSS1 RSS2