I have CAS array with two nodes and NLB between them in unicast mode. When i shutdown one node - some clients become disconnected and can connect only if this node is up again. Again: Two nodes up. Clients can ping cluster IP. IF One node down - clients still can ping cluster ip, but exchange is OFFLINE untill shut downed node is up again. (if client go to control panel - user preference - mail and try to connect to exchange - it says that cant connect to exchange server). Other part of clients working normally (perhaps those, who connected to second node, that always up in this situation) Be real

just as addition - two nodes working on vmware hosts.Be real

This is because of HTTP session persistance, you can't do nothing about it. Just stop the host in NLB is all you can really do but that itself doesn't help much.James Chong MCITP | EA | EMA; MCSE | M+, S+ Security+, Project+, ITIL msexchangetips.blogspot.com

it this case i suggest microsoft didn't really do this redundancy with CAS array ??? so solution is to stop the node in NLB cluster, then reboot it and everything must work fine ??? or this problem will exist forever, because this is how it supposed to work ???Be real

i am not using outlook over http. I even configured kerberos authentication to my CAS. So the problem must be not in HTTP sessionBe real