we have configured a master slave loadbalancer configuration with the virtual appliance from loadbalancer.org.The configuration was setup as mentioned in there configuration documentation for exchange 2010, but at some point the master server CPU was climbing to the top.
This loadbalancer is placed before new exchange 2010 CAS servers in the environment. This exchange 2010 environment was just in production for routing the traffic for the mobile access. This problem did not happen during the environment test (what else is new 🙂 ), so we gave the master a reboot en will the master rebooted we could connect again to OWA and the PDA where syncing again until the master was up. One way or another the master virtual appliance got disorientated and changed his the virtual Server layer 7 protocol setting from TCP to HTTP. The slave had still the right configuration. This problem happened with version 6.14. Because we where already in production be brought the master loadbalancer down and where limping for a will. After finding out what the problem was (thanks to a engineer from loadbalancer.org) we corrected the problem.
The problem had occurred because we had put 1 CAS server in maintenance from the cluster webpage address and at that moment the HA heartbeat got disorientated. Because of that the heartbeat kept polling and this is why the CPU was climbing to the top.
The best thing to do, according to the support engineer is the give heartbeat HA a reset at the master level from the management page this will usually solve the problem.
This is a know BUG in version 6.14 which we where running this bug is fix in version 6.15 and if you have a support contract you can do a online upgrade. We did this upgrade to 6.15 in a maintenance window and we experience no problem with the master but the slave lost his DNS server address. but this is not hard to fix the upgrade it self takes about 5 minutes.
Thank U loadbalancer.org for the good and quick support!