Conference netcad::hub_mgnt

 Yves,

 can you verify the firmware bridge revision level ,correct code is V3.1

 if you have already this code, can you verify/post the bridge counter (bridge
and line )

 thanks

 Jean-Yves

 Ps: a little picture of your config will be appreciated
 

   Jean-Yves,


    The firmware in the bridge is v3.1. I can tell you it's DEChub 90
    In slot one there is a DECrepeater 90 T ,slot 2 and 3 has terminal servers,
    in slot 7 we have a DECbridge 90 FL, in slot 8 a DECrepeater 90 FL.
    Without interconnecting the two segments with the DECbridge 90 FL ,but by
     just putting the bridge in the backplane we have already the problem -->
    The PC's are loozing there tcp/ip session with the AXP machine on the same
LAN. The only thing we change was putting the HUB Management parameter on the
bridge in DISABLED. The connections stay stable up to two days. Should I QAR
this or are there some other checkings to do before ?

   Regards,


   Yves

    
    A couple more questions:
    
    1) What exactly do you mean by 'the PCs are loosing thier TCP/IP
    session?'; what is the symptom you see?
    
    2) What are the connections outside the hub?  What is the bridge
    connected to? What is the fiber repeater connected to?  Where is the
    AXP machine connected?

    In a conference, maybe this one, there was a note about using the
    TCP/IP (3COM) stack but maybe not limited to just that in Pathworks.  
    There is a known problem with this stack and how the DB90 tries to 
    locate which nodes are on which 90T or 90C ports.  The DB90 will send 
    a SYSID ir XID out and hope the various Ethernet interfaces can
    respond and then trap the port on which the response appeared.  
    Unfortunately the 3COM implementation of TCP can't handle this and hangs.
    
    Disabling hub management is supposed to reduce/eliminate this problem.
    
    dave

    PS.  Check note 93 in UPSAR::ETHERNET.

   Hi,


   Reply to note 843.3 

   1) What I am seeing ?
       From there PC's(PC's are connected to the UTP repeater) they are
connected to the AXP (Which is connected on the fiber of the other repeater)
machine with TCP/IP. On the PC'S they are using two types of emulators and when
the PC blocks the appearence of the problem is different:


        Emulator of CENTURY SOFTWARE ---> ACCESS VIOLATION Message but still
windows applications available.
         Smart term 420 for windows (PERsoft) ---> no control anymore(black
screen ,cursur to the right corner of the screen ).
        They are using Digital PC'S with Etherworks III controllers.


  Important : you just plug in the DB90 in the backplane without interconnecting
the two LAN's and you have the problem after a while .(Also we changed  the DB90
with another one ,same result). The only thing which reduces that problem is
putting HUB Management in DISABLE).




   Regards,


    Yves 

    We just suffered a problem like the one you are mentioned here.  I
    don't know if yours is solved already, if not please let me know and
    I'll ask the guy from the SPS group that worked the problem, which took
    almost three days, with Canada and TCP/IP Support.
    From our side the problem was solved, I work as Project
    Manager/Designer for NIS.  So I really can not tell you the solution
    they made.
    
    
    rgds,
    
    william

Here you'll see a vaxmail that the guy from SPS sent me in reference to a patch
that they installed for the problem that we had, and the one here mentioned.

rgds,

william

-----0-----0-----0-----0-----0----0----0-----0-----0-----0-----0-----0----0

Will,

The patch is cscpat_3061012.zip.  If he need a copy of this patch, let me 
known.

Rergards,

Julio