Ascend Archive
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

re: (ASCEND) 5.0Ap27 and the HDLC problem

>From: Kevin Smith <kevin@ascend.com>
>Date: Wed, 01 Oct 1997 19:49:06 -0700
>Subject: (ASCEND) 5.0Ap27 and the HDLC problem
>
>At 11:15 PM 9/30/97 -0400, Jason Nealis wrote:
>>
>> OK, I'm confused, I must have missed Kevin's message, Is there a bug??
>>HDLC? ap27? say what? I've heard if you initiate a HDLC call into a ap27
>>the box will explode. :)
>>
>> Actually, I'm concerned, cause I'm about 3 hours away from deploying this
>>into my network. Can anyone confirm if there is an issue on HDLC calls?
>
>This thread started with a question about an HDLC related problem that was
>reported against 5.0Ap16. Since it's still on the open TR list, all patch
>versions since then have this vulnerability. If you have been running any
>of 5.0Ap16, 5.0Ap20, 5.0Ap23 or 5.0Ap24 and have not seen this problem,
>then you are VERY UNLIKELY to see this problem in 5.0Ap27, since nothing has
>changed with respect to it.
>
>How bad is the problem?
>It is listed as HIGH priority in Engineering.
>A few customers have decided to remain on 5.0Ap13 because of it, they tend
>to be >50% ISDN calls. If you are predominantly a modem service provider,
>you are unlikely to see it at all. So far it has been reported by only a
>handful (six) of customers (most MAX users have a vast majority of modem
>customers rather than ISDN or a 50/50 mix).

Kevin

Well I'll tell you I am surprised that only six of your customers have 
seen this problem. We first reported this problem two months ago 
PR 126490 and had diagnostics traces within a day. With Patch level P16 Ascend
released some type of merged 56K code and ISDN HDLC driver code that
makes the box stop accepting ISDN calls.

A few side notes:

* I have seen this problem on Max4048 with one or two PRI's attached and
  Max4004's with 1-4 PRI's attached. With the Max4004 with one PRI there 
  was only about 12 open ISDN channels. 

* We normally see C=2 P=10 in the syslogs

* The problem will clear up sometimes after 30 to 60 minutes, we have
  syslogs monitors now that page us when it happens so we can reboot!

* We have found out the Max Stacking does not work 9 times out of ten! We
  attempted to spread out the load of 5 PRIs hunted across four Max boxes
  and one the ISDN problem still occurred and two it was pretty impossible
  to have ISDN customers get two channels to bond! The Max's are on a
  10baset Ethernet switch. Max stacking fails with Ascend pipeline 
  customers as well as non Ascend ISDN equipement.

* On the Max stacking problems it seem like Ascend is not very interested
  in fixing the problems since I keep getting told buy a TNT for ISDN.
  When I was looking at Livingston verses Ascends I was told by the
  Ascend sales folks that the Max4xxx series would handle at least 
  seven boxes stacked. I want this feature (stacking) to work!

  The only reason why we have bought Ascend equipment is their long
  history of being an ISDN vender using their Max's. Don't break ISDN
  support unless your are willing to take back these Max's I own!

Since we mix both ISDN and modem calls together in a single hunt
group we have been burned really badly. K56Flex is too unstable to run
at below P16 so we have stayed the course. Some good news is that Ascend
is going to loan us a TNT until this is fixed. 

Questions:

* Why is a major change released in a patch release?

* Why has it taken two+ months to fix this?

* What's the story with Max Stacking?


A bleeding Ascend customer!

Regards,

Tom Wye



>
>How does it manifest itself?
>The TR reports that after about 24 hours, the MAX stops accepting ISDN calls.
>There are debug messages that refer to HDLC resources being unavailable.
>Calls are disconnected with c=2, p=2.
>A system reset temporarily resolves the problem.
>
>What to do if you are seeing it?
>Unless you have already done so, contact Ascend tech support and tell them
>you are experiencing the problem with ISDN calls failing - TR#2554.
>If it is feasible to do so, you could down-grade the MAX to 5.0Ap13 until
>we announce the fix.
>
>In any extreme cases, ask to talk to the Escalations Director:
>>Any customers having this problem should call into support and then we will
>>get them the code [next week] and if it extends we will spread the Pri's out
>>across maxes.
>>
>>Tom
>
>*We have a fix for this in the incremental branch, but have not yet made a
>version that we feel to be stable enough to release. That should change within
>the next week-or-so, and we will have the fix in the next major release which
>should also appear within the next few weeks.
>
>
>Kevin
>
>
>++ Ascend Users Mailing List ++
>To unsubscribe: send unsubscribe to ascend-users-request@bungi.com
>To get FAQ'd:   <http://www.nealis.net/ascend/faq>
>


********************************************************************************
* Thomas J. Wye                         Tel: (408) 260-5000 
* Bay Area Internet Solutions           Fax: (408) 260-5003
* 900 Lafayette Street, Suite 710    E-mail: twye@BAYAREA.NET
* Santa Clara, CA 95050          HTTP: http://WWW.BAYAREA.NET
*                                Info Robot: INFO@BAYAREA.NET       
* 33.6Kbps dial-up, 56Kbps-T1 (Leased*Frame Relay), server colocation, 
* Virtual Domain hosting and World Wide Web development. 
********************************************************************************

++ Ascend Users Mailing List ++
To unsubscribe:	send unsubscribe to ascend-users-request@bungi.com
To get FAQ'd:	<http://www.nealis.net/ascend/faq>