No announcement yet.

Strange Occurence with CAS server

  • Filter
  • Time
  • Show
Clear All
new posts

  • Strange Occurence with CAS server

    We have a strange thing happening with our Exchange 2010 CAS/HT server. Our network is connected via a CenturyLink MPLS circuit with our Exchange at a central site and around 40 remote sites with clients. Our CAS/HT server is connected to a Cisco core switch that serves as the Default Gateway. the gateway on the server is setup to xx.xx.0.1. This works normally for all internal clients as long as the network is stable. For internet traffic, the switch reroutes to xx.xx.0.3. Now for the issue. when we have a site go down for any length of time, once the link to the site is reestablished, any clients that were connected to the Exchange server when it went down, either via MAPI or RPC over HTTP, cannot connect back. I looked at the server and saw that traffic to those IP addresses was being sent directly to xx.xx.0.3 instead of the normal DG (used Wireshark). But, any other addresses on the same subnet as the ones that cannot connect, as long as they were not connected via outlook at the time of the outage, are fine. Traffic going to the affected subnet is OK but traffic to the specific IPs is not. My Network people don't believe me so I need to find out just why this is happening. We were on an ATT MPLS (in fact, a few sites still are) and this issue never happened then. My question is, does the CAS server have a hidden ARP cache that tracks each connected destination IP address? If I clear the windows arp cache, the problem is solved but when I look at the ARP cache in windows, it looks normal. I only see addresses on the same subnet as the Server (what I expect). The routing table also looks normal.
    Can anyone offer any explanation to this behavior. I think there is something with the network that is causing this but I need to prove it isn't the box first.