I work for the web department of a school that is still running (explicative) token ring.
In our new webserver setup, we use the same loadbalancer design/software that anandtech uses. However, we can't get the secondary load balancer to take over for the primary properly. When heartbeat detects the failure, ifconfig's up (as an alias) the new ip, and begins arp spoofing, all of a sudden the entire token ring interface (even the old IP) become un-pingable.
This is clearly a problem with the token ring interface, as we do it to the internal (its a NAT based setup) ethernet card and it works just fine. We also tried using a spare set of ethenet cards as the external interface's, and they worked too.
my guess is its just plain a bug in olympic.o... any ideas?
In our new webserver setup, we use the same loadbalancer design/software that anandtech uses. However, we can't get the secondary load balancer to take over for the primary properly. When heartbeat detects the failure, ifconfig's up (as an alias) the new ip, and begins arp spoofing, all of a sudden the entire token ring interface (even the old IP) become un-pingable.
This is clearly a problem with the token ring interface, as we do it to the internal (its a NAT based setup) ethernet card and it works just fine. We also tried using a spare set of ethenet cards as the external interface's, and they worked too.
my guess is its just plain a bug in olympic.o... any ideas?
Facebook
Twitter