Conference mvblab::alphaserver_4100

    The max number is 2 for the Madge cards.
    
    This information is available on the web at:
    
    http://www.digital.com/info/alphaserver/tech_docs/alphasrv4100/docs/
      supported_options.html
    
    Todd

    This pointer works too...
    
    http://www.digital.com/alphaserver/tech_docs/alphasrv4100/docs/
    	supported_options.html
    
        Jeff

    Re .1, just for my curiosity, what are the technical reasons for
    supporting only 2 Madge TR adapters at max?
    
    On other platforms (Alpha and Intel), as far as i could tell while
    working with the Madge folks, there were no limitations (i.e. as long
    as there are available slots).
    
    --svb

    Probably isn't a technical reason.  I don't remember for sure, but it's
    likely that we only had 2 during qualification.
    
    Todd

    Re .4, that should be easy to fix.  But, let's try to address it when a
    specific request for more than two Madge TR adapters arrives from the
    customers.  Please send me mail when that happens.
    
    --svb

    
    Thanks for the pointers. 
    
    They both refer to Version 3.51 WNT.  Are they supported for 4.0? 
    
    Thanks. 
    

    
    Ok, 
    
    THe reason I am asking about this is I have a customer who is TRYING
    to run 4.0 WNT on 4100 with 2 madge pcis. 
    
    If they enabled just the first system runs fine.  If they enable the 
    second with IPX AND TCP/IP, the system crashes within hours. 
    
    If they disable TCP on the second and leave IPX, runs fine. 
    
    I have trying to figure out if this is a madge issue (they have latest
    drivers from madge) or a configuration issue (are the cards
    positionally correct on the 4100) or if they have a bad card. 
    
    I haven't been able to get them to try 2 other types of cards (Like 
    2 de500s) to make sure the config is correct. 
    
    Any pointers? 
    
    Mary Jo 
    

    Re .7, if you could, would you get the customer to provide us with more
    details on the configuration:
    
    Which PCI bus(es) are the two Madge TR adapters on?
    
    Does it matter which one being "first" or "second"?
    
    What was the stack dump when the system crashed?
    
    There are two flavors of the Madge TR PCI adapters.  Could you get
    them to provide the etched ID on the cards?
    
    --svb

    
    
    THanks for the reply.  I'll check. 
    
    We have swapped the cards around with no change. 
    
    Using bus 0. 
    
    I'll check about the others. 
    
    Mary Jo 
    

    Here is an update. 

    1.  Cards are installed in Slots 4 and 5 on PCI bus 0 

    2.  He has NOT swapped cards.  He doesn't want to.  I am trying 
        to get him to. 

    3.  Both cards are labeled (BM)2.  (I don't if this is correct or 
        not because he has not been able to open system to physically 
        inspect cards.  

        I had requested yesterday for him to get field service out there 
        to make sure everything was 'hardware' correct'. 

    4.  Since asking for the memory.dmp, it turns out the system is 
        HANGING not crashing.  Just a little difference. 

        The CTRL-ALT-DEL box just freezes. 

    5.  Description of last hang: 

         Card 1  protocols bound to card:  IPX/SPX and TCP/IP 
         CARD 2  Protocols bound to card:   TCP/IP 

        Server was started at 7:30 a.m. (because of failure last night). 
        Lotus Notes server running and replication to other servers in 
        the network.  (This is being done over card 1).  

        At 10:30 a.m. Card 2 failed, would no longer accept a ping.  THey 
        did not discover the failure until 1:00 p.m. when physical 
        inspection of monitor shows normal NT GUI, but moun and keyboard
        would not respond.   Lotus Notes server continued to replicate 
        during this time frame. 

        Reset server and checked Event log -- no errors or information
    regarding the failure. 


    Cards were swapped as of 5:00 pm central time.  Here is the information
    on the cards (the card 1 information is from the original setup, prior 
    to the swap) and is entered in upper and lower case as indicated on the 
    cards:

    Card 1:   Metal tab: Madge 

         Smart
         16/4
         PCI
         Ringnode
         (BM) 2
         3274 
         GLOBE N.J.
         07092
            Top of board:  KCE WV - 0   94V - 0  F1   4296

     Back of board: Board No.  151-101-05 1996 Madge Networks, Ltd.

     Back of board bar code: 151-100-045    740C7B   C44K    Made in
    Ireland

    Card 2:   Metal tab: Madge 
         Smart
         16/4
         PCI
         Ringnode
         (BM) 2
         2636 
         GLOBE N.J.
         07092
            Top of board:  KCE WV - 0   94V - 0  F1   4596

          Back of board: Board No.  151-101-05 1996 Madge Networks, Ltd.

     Back of board bar code: 151-101-05    7702A1   C50A    Made in USA

    As of Sunday a.m., after swapping cards, system was hung.  He could 
    ping card 1 (old card 2) and not ping card 2 (old card 1). 
    

    And additional update: 
    
    
With two cards running TCP/IP they do not have routing enabled. 

We asked him to enable routing.  He doesn't want to.  But he did 
me the following information.  


/Also, since the movement of the cards down a slot did not make a 
/difference, we have change the configuration so Card 1 is in slot 4 
/of PCI Bus 0 and Card 2 is in slot 2 of PCI Bus 1.  System has been 
/running with 27 hours without any problems. 

    Re .11, this is the second case i've run into where putting two
    adapters on the same Rawhide PCI bus would not work correctly, but
    would do so when they are on separate buses.
    
    The other one was the Pathlight DataPump SSA adapter.
    
    --svb