Possible misbehavior of BMCA with redundant links
Reported by a user (https://forums.ohwr.org/t/wrs-and-multiple-slave-ports/848457/6):
I read part of your thesis and tried the network topology proposed in fig 5.12.b, except that I don’t have a Grand Master switch, only a free-running master (no external time ref). I set all the other switches in boundary clock, using BMCA and ports are set to auto. It happens that after disconnection and re connection of one of the redundant links of switches 5 and 6, none of the links comes back up. One is marked as PASSIVE, IDLE, EXT_OFF in wr_mon, and one is SLAVE, IDLE, EXT_OFF but… there is no synchronisation, LEDs are orange for both ports. If I disconnect the passive port, the slave one stays OFF. It also happens that slave port is on and keeps trying to syncronize, but doesn’t reach teh “track phase state” (it also happens that it does). I’ve also seen the port that is supposed to be passive, trying to be master instead. Have you experienced this kind of problems with BMCA algorithm ?
Fig 12:
Check behavior of BMCA....