> -----Original Message----- > From: francisco [mailto:[EMAIL PROTECTED] > Sent: Sunday, April 09, 2006 4:07 PM > To: Barry, Christopher > Subject: Re: IO fencing question > > > Thanks everyone for your ideas on this. As it turns out, > the issue is > > indeed the switch's redundant fiber port not releasing. As > soon as power > > hits the server's motherboard, a link is present on the > switch - even > > though all of my fiber NICs are in PCI slots. The only way I can > > reliably failover the switch port is to remove power > completely from the > > router. > > Are these managed switches, and if so, can you login and flush the > switches arp cache? A script to do this upon carp event might be the > better solution. > > -f > http://www.blackant.net/ >
heh, it's a conundrum wrapped in an enigma, tied with a paradox. ...or maybe just a catch:22? Because the routers attach to each of three switches directly into their redundant MDAs, the master router is the only guy that can talk to them. The backup router is 'fenced out' by the MDA itself. If the master craps out, but the switch is still hanging onto him, who will be able to access the switches arp table to flush it? Short answer: nobody. The only way to do this is to have another host out on that network that can detect if the router is down, and then do this, maybe via snmp. The problem with that is I do not control the hosts out on that net typically, and it becomes another point of failure, and spreads the system of redundancy a bit too thin in my view. Another *interesting* problem with this topology choice I've made is what happens when the redundant fiber on the switch that is connected to the master goes down? Until this happens, I guess I just cannot know. Thanks, Chris
