Conference 7.286::fddi

482.0. "NFS Timeouts, DECbridge 620 & AT&T Starlan" by CTOAVX::BEAULIEU () Thu Feb 20 1992 16:54

    
    	We have been testing for some time now the DEC FDDI products at our
    customers site. The customer in their great wisdom originally choose the 
    AT&T Starlan products for their FDDI ring.  When AT&T and NCR merged,
    according to the customer, the key engineers left the projects because
    of NCR management. And the Starlan FDDI products, became feature bound.
    
    So they came to us, and the testing began.... 
    
    We have been doing compatibility testing, timing testing and load testing.
    As the last test we took a DECbridge 620 and put it in the production
    FDDI network with the AT&T Starlan gear.  This bridge has been running 
    fine for about a week. Today the network manager said for the past week
    he has noticed a large rise of NFS time outs. He thinks it is our
    bridge, but isn't positive. I would like to proactive about this.
    IP frag is on,  and we are looking at the timers of both our bridge and
    AT&T's.   Is there anything we should be specifically looking for that
    could cause NFS timeouts?
    
    By the way AT&T's FDDI bridge does not do Spanning Tree and has no
    in band or SNMP management. These are some of the reasons they came to
    Digital for help.
    
    Thanks, Mike Beaulieu
    

    You can get occasional high latency through the DECbridge 620 if 1 or
    more of the ethernet ports are not being used, and there is no loopback
    connector installed.  This is because the unused ports will fail self
    test, and the 620 will occasionally retry the test to see if the port
    has been reconnected.  The port tests take several milliseconds, and
    packet forwarding is suspended during the period.  Putting a loopback
    connector on will make the port pass self test and stop the periodic
    testing.
    
    Steve