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

Re: (ASCEND) Ascend Frame relay Problem



        James Voyes said:

>Has anyone out there Had a problem with Frame Relay where if you loose the
>host site , You also loose all Remotes. The problem we are seeing is if the
>host side goes down we loose all the remotes and when the host side comes up
>the remotes do not recover on there own. They have to be power cycled in
>order to recover. Anyone!

        Been there, done that, still have the emotional scars
        from a small network of 10 Pipeline 130s "hosted" by
        a Cisco router.

        From what we have seen from local laptops plugged into
        the P130s that are "down", the specific P130s >>DO<< see
        the LMI packets from the frame relay switch, and >>DO
        NOT<< have an alarm condition.  We have seen a lack of
        incrementing in the transmitted packet counts on
        occasion.

        The FNIDs to which the Pipelines are connected also
        show no error lights, but these are very primitive
        (nearly obsolete, in my view) devices.

        Since the majority of these outages have been due to
        power flickers at the customer site (out core sites 
        never go down), we expect that there is some strange
        power-up sequence incompatibility between the P130
        and the FNID.

        We have found that, in some cases, power-cycling the
        FNID will fix the problem, where in others, power
        cycling the FNID is not enough, and the P130 must
        also be power-cycled.  The lack of a consistent
        scenario is problematic.

        There is no ISDN service in the area in question, so
        dial-back-up is impossible in these cases.

        What I would suggest would be:

                a)  Send your own people on site, and
                    see what the pipeline says to a laptop
                    plugged into the console port.  If you have
                    an alarm condition, you have a "properly
                    working P130 (or whatever)", and a beef with 
                    the telco.

                b)  Check the telco interface device.  PairGains
                    also have a serial port suitable for plugging
                    a laptop into to see status.  The lights should
                    tell most of the story.

                c)  Try power-cycling the telco customer-premise
                    equipment (FNID, PairGain, whatever), and wait 
                    3 mins before doing anything else (Frame Relay 
                    DLCIs can require up to 2 mins to "recover" 
                    once LMI packets start flowing again).

                d)  Ask the telco to overtly send some LMI packets
                    downstream from the frame relay switch in the
                    reverse of the "normal" situation.  This is 
                    fairly easy to do, and can be used to verify
                    that the Pipeline "sees" the packets.

        In our case, the majority of the outages "went away",
        after about 6 months of daily problems.  We therefore
        tend to lay "blame" at the feet of the telco, even if
        the solution was to upgrade the Ascend/Cascade frame 
        relay switch code.
        
        My resulting opinions would be one or more of:

                1) FNIDs suck.  PairGains rule.

                2) Ascend has(had) an initialization problem with 
                   the P130, and/or the Cascade switches, a problem 
                   that they have denied to date.

                3) It is possible to train non-technical customer
                   staff to be decent "field techs", and they can
                   be of great help in debugging such a problem
                   once given some clear instruction about status
                   lights, telnet, Pipeline status screens, etc.

                4) It is never possible to get the telco frame
                   relay switch admin and Ascend on the phone
                   at the same time due to the Hisenberg Uncertainty
                   Principle, so make it clear to the telco 
                   that you expect THEM to address the problem 
                   with Ascend/Cascade.  They likely buy more 
                   toys from Ascend/Cascade than you do, so 
                   perhaps they will get more respect/attention.

  	  Who Knows What Evil Lurks In The Heart of MENSA?

    james fischer                     jfischer@supercollider.com

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