i have two different problems. one seems to clearly be an arp cache problem. the other may be also, but i'm not sure.

i have a two-node bordermanager cluster without shared storage. i have a cluster resource that is simply a secondary ip address. i can move it from one server to another, but if i do that quickly several times while my workstation is pinging the ip address. i will eventually start getting "TLL expired in transit" errors. once this happens, i basically have to offline the resource until i start to get "request timed out".

after this happens the other problem come up and it is more difficult. once i get the "request timed out", i can only online the resource on the node where it worked last. the other node (where i tried to migrate it to when the TTL errors started) gives me the error "secondary ip address x.x.x.x is not associated with any local binding". the resource will go comatose every time on that node, but will load just fine on the other node. if i leave it on the node where it will run for a while, maybe 15 minutes or so, then i can move it back and not get the secondary address issue.

are they both arp cache or is there something on the server itself that is causing the second problem? how can i clear this out.