Quote:
Originally Posted by chrapojr
Hoping someone can help.
We have an Exchange 2010 server with BES 5.x. Everything was working fine.
We added a 2nd Exchange Mailbox Server and CAS Server for redundancy. We created a CAS Array in Exchange. We then set the RPCClientAccessServer parameter on the Mailbox Database to point to the same URL as the URL for the CAS Array.
The BES server detected the mailbox database change and restarted some of the sevices. After that, the BES could not open the mailboxes of any users in the Mailbox Database. The error in event viewer was 20400 and mentioned unable to open mailbox of user xxx and also referenced the new FQDN that was used in the RPCClientAccessServer parameter. No BES does not work for those users in the mailbox database.
If we set the Mailbox Database RPCClientAccessServer FQDN back to the actual server name, everything works good again.
Any help would be appreciated. RIM hasn't been able to help so far. Thanks
|
Hi,
You didn’t state what method you used to cluster those CAS servers. If you used WNLB then your problem could be that in the NLB cluster properties, the port rules you have specified for MAPI (TCP 135 [RPC Endpoint Mapper] and TCP 1024~65535 [Dynamic RPC]) are using an affinity setting of "none". The BES appears to use multiple MAPI connections to the mail system and it does not like being balanced across different CAS servers. Change the MAPI port rules to an affinity setting of "Single" and that will cause each host to get pinned to a single, available CAS server in the cluster.
If you are using some other load balancer, check its dispersion settings for something similar. You do not want it to distribute multiple TCP connections from a single host to more than 1 destination CAS server array member.
The other port rules for 80, 443, 25, and 587 do not require the affinity to be set, or "None". This will effectively load balance inbound SMTP from an external MTA at the network edge across both servers.
-Good luck -J