Sticky issue when failed server become active

[gkadudfo1]

Hi,

We are operating the ace(csm) and using the sticky option.

Assuming that We have two servers with sicky for redundancy and being opreated active - active, 

I realized that if one server go down and go up, Csm dosen't forword packet to restored server.

I think it is due to the sticky option. We need csm to play loadbalancing two servers once one is restored immediately.

Is there anything opinion or solution?

Thanks,

[mwinnett]

If a real goes down (fails probe) then the csm will remove it from service and not send connections to it. The entry will also be removed from the sticky table. When it comes back, it will receive new connections when it is elgible. If all incoming connections match existing sticky entries then it will take time to gather load. If you configure "predictor leastcons" under the serverfarm and limit the sticky life, then it will allow the failed real to recover load quicker. Matthew