Conference netcad::hub_mgnt

3758.0. "Lots of Ring Inits received by adding one node" by GIDDAY::STANISLAUS () Wed Jul 31 1996 12:24

    Cross Posted in FDDI notes conference:
    --------------------------------------
    
	Ring Init Received counts keep increasing on all nodes at a very very
high rate and users loose connection to the FDDI nodes, when just one additonal 
node is added to the following configuration. This additional node can be
any FDDI node - a system or a DECbridge (FDDI type) for example. The problem 
appears to be that, that one last straw is breaking the camel's back, but 
doesn't matter what that one straw is - a system or a DECbridge (FDDI type).

Working configuration:  i.e. *No Ring Inits*  *No Problems whatsoever*
----------------------

DC500 = DECconcentrator 500
DB6xx = DECbridge 6xx
DS900EF = DECswitch 900EF
DC900FH = 14 port DECconcentrator 900FH

DS900EF and DC900FH are in the same DEChub 900. For some reason the customer
doesn't have any FDDI back plane connection on this Hub. He has a physical 
cable from the A port of the DC900FH to the B port of the DS900EF.

			
	+----DC500 #2----DC500 #1----DB6xx------+
	|	 				|
	|		FDDI RING		|
	|					|
	+----DC500 #3----DS900EF-----DC900FH----+

All the following systems and FDDI nodes are Digital's H/W - no 3rd party
FDDI H/W.

2 x Alphaserver 2100 systems, each has DAS DEFEAs and are both dual-homed to 
DC500 #1 and DC500 #2.
1 x Alphaserver 1000 system has a DAS DEFEA and is dual-homed to 
DC500 #1 and DC500 #2.
2 x DS900EF switches, both dual-homed to DC500 #1 and DC500 #2.

6 x Systems are connected via SAS cards to DC500 #1 only.
8 x Systems are connected via SAS cards to DC500 #2 only.
    
2 x Alpha 4100 systems are (via DAS cards) dual-homed to DC900FH and DC500 #3.
1 x Alphaserver 1000 is (via SAS card) connected to DC900FH

Problem first noticed as follows:
---------------------------------

A new Alphaserver 1000A with a SAS DEFEA card was installed. Engineer tried to
connect this system to the DC900FH. He brought the whole FDDI network down.
Connections were lost and all the systems logged lots and lots of RING INITS
RECEIVED ERRORS. He couldn't find a node whose RING INIT SENT count increased.
Even the new Alphaserver 1000A logged RING INITS RECEIVED.

So Engineer disconnected this new Alphaserver 1000A and the FDDI network 
stabilised immediately. He then connected this Alphaserver 1000A to DC500 #1.
Same problem again. He then tried to connect to DC500 #2. Same problem again. He
then tried to connect to DC500 #1.Same problem again. He changed the DEFEA card
in the ALphaserver 1000A and tried. Same problem again. He changed the FDDI 
cable between the DEFEA and DC900FH. Same problem again.

Engineer then disconnected one of the Alpha 4100 from the DC900FH and DC500 #3.
Then he connected the new Alphaserver 1000A to the DC900FH. No problems at all.
Then he tried to bring in the Alphaserver 4100A into the network. Again RING
INTS RECEIVED and network unstable as before - doesn't matter where he connected
the Alphaserver 4100 - DC900FH or DC500 #3.

So he left the Alpha 4100 disconnected but the new Alphaserver 1000A connected 
to the network via DC900FH. Everything was working fine.

So the theory is the addition of that one extra FDDI node broke the network.

So engineer brought a spare DB6xxx that this customer had and connected it to
the DC500 #3. Again same problem - RING INITS RECEIVED on every FDDI node and
network unstable again.

This confirms as though that one extra node (doesn't matter whether it is the
Alphaserver 1000A or Alpha 4100 or DB6xxx) breaks the network.

No RING INITS SENT logged by any node.

In the absence of an FDDI analyser how does one troubleshoot this problem.

Alphonse

    I forgot to mention a VAX 4000-90 that is connected to the DC500 #3.
    Don't know what controller it has.
    
    ALphonse