Conference netcad::hub_mgnt

    The PEswitches aren't running the most current firmware which is
    v1.1.1. I'm not aware if the problem being seen is a direct result
    of something wrong in v1.1.0, or whether it's an indirect result
    of other known bugs in v1.1.0
    I'd suggest upgrading the PEswitches to v1.1.1, as it may result
    in the problem being seen to go away.
    
    Bob

    Bob,
    
      Thanks for your reply. However, V1.1.0 was a typo, they're both
    running V1.1.1.
    
      I was at the customer site today and got a little bit more info.When
    I wrote "The station where the partition they're trying to mount lies,
    is connected to the GS/FDDI." I should have written is connected to the
    DECconcentrator 900TH. However a BOOTP/TFTP is performed through the
    GS/FDDI, one FDDI to Ethernet switch (don't know if DS900EF or third
    party) and a Synoptics bridge/switch (?) where the host (BOOTP server) is
    connected.
    
    This seems to be NFS-related and not UDP as BOOTP/TFTP and SNMP seem to
    perform normally although I agree that they're doing BOOTP/TFTP to one
    host and NFS mount to another. FDDI to Ethernet "translation" could be
    an issue.
    
    Any hints?
    
    Pedro
    

    >I wrote "The station where the partition they're trying to mount lies,
    >is connected to the GS/FDDI." I should have written is connected to the
    >DECconcentrator 900TH. However a BOOTP/TFTP is performed through the
    >GS/FDDI, one FDDI to Ethernet switch (don't know if DS900EF or third
    >party) and a Synoptics bridge/switch (?) where the host (BOOTP server) is
    >connected.
    
    You now mention a DS900EF is included in the configuration. This
    wasn't mentioned in the base note. You'll have to find out what kind
    of bridge it is (why didn't you verify this when you were there,
    instead of guessing what it is after the fact?).
    The dependency on UDP packet getting through the path between the
    clients and the server are curious when it depends on whether the
    PEswitches are providing IP services or have an IP address. It
    almost sounds like there may be UDP packets addressed *to* the
    PEswitches, and of course without an IP address, they are
    going to be rejected or ignored by the bridge. 
    Are the clients always powered up? 
    
    Get us a more accurate configuration, and perhaps there will be
    then be sufficient information for others to chime in with what
    may be wrong.
    
    Bob

    Bob,
    
    >    You now mention a DS900EF is included in the configuration. This   
    >wasn't mentioned in the base note. 
    
    Because the customet told me they were connected to the GS/FDDI on the
    phone... and I was at one site (it's the University of Geneva who have
    several sites who are a few kilometers apart...)
    
    >You'll have to find out what kind of bridge it is (why didn't you verify
    >this when you were there, instead of guessing what it is after the fact?).
    
    Why do you think so, if this is the part that doesn't cause any
    problem? Maybe would you like to know the model of the Synoptics 8-)
    (I'm sorry but I had to write this after your angry answer...).
    It  was only given as an info when trying to be complete. Regarding the
    guessing, see above. (I don't want to go into details regarding the
    internal problems of the customer, ie. the one having the problem with
    its network management infrastructure...).
                       
    Why do you say NFS packets could be addressed to the PEswitch ?
    
    Cheers,
    
    Pedro

    Pedro, I am not angry, and I apologize if I left you with that
    impression. It was more of an expression of being confused and
    having the imprssion that you had been at the site and was told
    and shown by the customer what the configuration was.
    
    >> Why do you think so, if this is the part that doesn't cause any
    >> problem? Maybe would you like to know the model of the Synoptics 
    
    We are always looking for as complete a picture of what the customer
    configuration is and of course as complete a description of the
    problem as possible. Otherwise we end up having to probe with more
    questions etc.
    
    >> Why do you say NFS packets could be addressed to the PEswitch ?
    
    Because I was told that UDP packets can take the form of management 
    packets being sent to the bridge, so thought that perhaps 
    someone/something was sending management packets addressed to 
    the bridge. 
    
    So all we're looking to find out is all the components in the path
    between the server and the end nodes that are seeing the timeouts.
    
    thanks,
    Bob

    Bob,
    
    >Pedro, I am not angry, and I apologize if I left you with that
    >impression. It was more of an expression of being confused and
    >having the imprssion that you had been at the site and was told
    >and shown by the customer what the configuration was.
    
    It's OK, not to bother... 8-)
    
    Regarding the complete path the NFS client is connected to the PEswitch
    900TX which has its A and B ports in a FDDI ring on the backplane of
    the hub where is also the DECconcentrator 900TH which has an A or B
    port connected to the BP and the another one in front (that connects to
    the GS/FDDI). The NFS server is directly connected to the 900TH (CDDI).
    
    Hope this helps clarify the situation.
    
    Cheers,
    
    Pedro
    

    I've just sent a mail to the customer asking exactly to which 900TH was
    the NFS server is connected and what's its port configuration (A in
    front and B in the BP or the other way around). I'm thinking about
    response times if the packet goes all the way down to the GS/FDDI and
    then comes back (which should be a problem) as the 900TH, the 900TX and
    the GS/FDDI are all on the same ring.
    
    Just an idea, although I don't place too much hope in it...
    
    Pedro
    

    More info:
    
    Slots 1-4 empty.
    On slot 5: conc 900TH A port goes to the GS/FDDI, B to the BP. NFS
    server connected to port 4 (CDDI).
    On slot 6: conc 900TH A port to the BP, B port to GS/FDDI.
    On slots 7 & 8: PEswitches 900TX A and B ports to the BP. NFS client
    connects to any the Ethernet ports.
    
    They don't have an NFS server with an Ethernet adapter so that they can
    test with both the NFS server and the client on the same PEswitch.
    
    The primary ring goes out from port A of slot 5.
    
    Hope this brings some light.
    
    Thanks.
    
    Pedro
     

    When reading all the notes of this entry again, I realized I wrote
    something that I didn't want to in .7
    
    "I'm thinking about response times if the packet goes all the way down to
    the GS/FDDI and then comes back (which should be a problem) as the 900TH,
    the 900TX and the GS/FDDI are all on the same ring."
    
    Of course, between brackets I wanted to write (which SHOULDN'T be a
    problem)...
    
    Cheers,
    
    Pedro