[e-nsp] Dead MSM in a Black Diamond

[Christian Panigl, ACOnet/VIX/UniVie]

>Date: Tue, 28 Oct 2003 11:40:18 -0300
>From: "Fernando Krahe" <fernando at tche.br>
>
>I have a black diamond running 6.2.2 (build 18) and running layer 2 only
>with a single MSM module. This module sudenly stopped to answer, although
>the switch is still forwarding 160Mpbs of frames correctly and none of the
>ports are in the vlan Default. Both lan and console interfaces of that
>module produce no response at all and the master LED became off (system LED
>is a blinking green, error = amber and environment = green).
>The moment we connected the console, we received a message about an Address
>load Exception and some more register information. After those messages
>there are no more outputs from the console.
>Have any of you seen this?

    Hi Fernando,
    
    that sounds scary, hope you do have a saved copy of your config ;-(
    
    We have been running EW6.2.2b18 on a BD for quite a while at the Vienna
    Internet eXchange. The main problem we had was that it's occasionally
    producing phantom HW warnings which are not always clearly
    distinguishable from real HW warnings.  We had one instance of a failing
    master MSM, however, the failover to the slave MSM worked nicely as
    expected.
    
    On TAC recommendation we recently upgraded one of our BDs to EW6.2.2b134
    which is now producing (at least) different phantom HW warnings :-(
    
    If you have a support contract I'd suggest that you call them asking for
    a replacement MSM with a pre-installed config, which you should send
    them in advance.  I don't see good chances that a failover including a
    synced config will work if your master console is not responding.
    
    During a maintenance window you should then do
    
    # run diagnostics extended
    # run diagnostics packet-memory
    
    on every interface module (disrupting traffic on the resp module)
    and on the MSMs (which reload afterwards)
    
    be aware that every instance takes a couple of minutes !

    Good luck
    CP
    
 --- ---------------------------------------------------------------------- ---
 --- Christian Panigl        : Vienna University Computer Center  -  ACOnet ---
 --- UniVie - ACOnet - VIX   : -------------------------------------------- ---
 --- Universitaetsstrasse 7  : Mail:  Panigl at CC.UniVie.ac.at    (CP8-RIPE)  ---
 --- A-1010 Vienna / Austria :  Tel:  +43 1 4277-14032         (Fax: -9140) ---
 --- ---------------------------------------------------------------------- ---