[comp.dcom.sys.cisco] CISCO ASM-LAT problems

mikey@canisius.UUCP (Mike Szymendera) (10/18/90)

 I recently installed 3 Cisco ASM terminal servers (CSC/2 processor, 1MB
memory, 48 lines) w/ 8.1(19) software. An older ASM was also upgraded to
8.1(19) in order to run LAT. All of these servers seem to run telnet 
and rlogin connections to our 2 VMS hosts (that have Wollengong WIN) with
no problems. Unfortunately, some users are demanding LAT support for better
terminal I/O and reverse connections for printers. While running LAT, our 
users started seeing the following scenario:
  - server starts to get really sluggish
  - connections start dropping- users get a "connection closed by foreign host"
  - if I enable LAT event debugging, a message of the form "connection closed-
         reason 5" seems to coincide w/ the connections being dropped.
  
It seems to me that some resource on the terminal server (memory/buffer pool)
is getting used up, causing the delay of packets sent/received from the cisco.
The VMS hosts find this delay unacceptable and send some sort of LAT control
packet back to the cisco to force it to terminate the connection.  Cisco 
confirms that the 'reason 5' is indeed the foreign host sending a
message to the terminal server to terminate the connection. Just for 
the hell of it, I tried changing the "vc-timer circuit" parameter on 
the cisco terminal  to higher/lower values (it is set to 80 milliseconds
as a default) . When the value is increased (to 100, 120, 140, 200) lat 
connections become less reliable (and at some point won't even be made 
to the requested host). As I lower the value of vc_timer, we have noticed 
that the servers aren't 'crashing' as much (once every 2 days
as opposed to 2-3 times a day). Logically, if I lower the value, it would seem
that packets are getting flushed out of the server faster- requiring less
buffer memory usage (going back to my original premise)- network traffic will
increase, but that doesn't seem to be our problem.  


Another possiblity is that something on the network is garbaging the ciscos
(we do see some bad packets (when using a Cabletron LAN SPECIALIST) w/ CRC,
frame, and address errors)) but this doesn't seem that likely...



I have a bunch of irate users/managers that are ready to trade the cisco's
in for DECSERVERS. 

Any hints/suggestions (besides the trade-in option :-) ?
Has anyone seen this problem? Is anyone besides me running LAT on a cisco
terminal server ?? I have a bunch of irate users/managers that are ready
to trade 'em in for DECSERVERS (yech !)

Any info via voice/FAX/e-mail is greatly appreciated- I'll post a followup
to the net w/ any info I receive.



--
------------------------------------------------------------------------------
USMAIL:	Mike Szymendera / Canisius College / 2001 Main St. / Buffalo,NY 14208
OFFICE: 42 55m 32s N / 78 51m 10s W / basement | (716) 888-2411, 888-2414
UUCP:	...!acsu.buffalo.edu!canisius!mikey  FAX: (716) 888-2525
BITNET: SZYMENDERA@CANISIUS	INET:	@relay.cs.net:mikey@canisius.edu
-------------------------------------------------------------------------------