jmj@cernvax.UUCP (jmj) (04/27/89)
We have a large bridged Ethernet with about 25 segments connected directly or indirectly with MAC-level bridges to one Ethennet backbone. The bridges are from DEC (DEBET, revision level E8) and from Vitalink (TranslanIV SW version 6.9.4 and 6.9.5). A DEBET is root with priority 32 and another DEBET is the back-up root with priority 64. All other bridges have default priority. The problem is that already on three occasions the root bridge went into a state were it repeatedly initialised itself. The first time we thought that the root bridge had broken down, so we took it out of the network, only to see that the back-up root started to do the same thing. We found that disconnecting a certain Vitalink bridge stopped the problem and everything was fine again after initialising the Vitalink and re-connecting it. We guessed that the Vitalink decided for some reason or the other to become the root, which was of course not liked by the real root. The fact that the incriminated Vitalink had also the highest priority in the Vitalink domain re-enforced our suspicion. We contacted Vitalink, who looked at the whole affair and give us a whole list of reasons why DEBETs start to re-initialise themselves. We thought we could discard all these reasons apart from the bad HELLO in the spanning tree network. They also advised us to make the Translan bridge the root (which we did not do, being good doctors). Hopefully we are able to analyse the HELLO traffic, when the problems shows up again. Has anybody on the net had similar problems? If so what conclusions have been drawn? How about the DEC netwatcher to tell us under what conditions a DEBET starts to initialise itself?