Recently, my team was notified on an issue happening to outlook clients. Outlook was not able to launch for all users in a particular site. The outlook connection status indicated that it is trying to connect to the F5 load balancer/CAS array, but immediately drops the connection.
We suspected an issue at the load balancer end and hence planned to bypass the load balancer and directly connect to the CAS server instead.
We tried to manually configure outlook for a test mailbox with the fqdn of the CAS server, but the profile configuration was not successful. DAG was configured and the mailbox database in which the mailbox resided had a copy on another Mailbox server. I tried to activate the passive copy and the observed the status, but the issue persisted.
It was then noticed that the RpcClientAccessServer attribute for the mailbox database was configured to point to the fqdn of the cas array, which was expected. In order for outlook to bypass the cas array and directly connect to the CAS server this attribute value should be changed and made to point to the CAS server.
Following cmdlets were used for this :
First to view the current configuration, the below command was executed in Exchange shell :-
Get-MailboxDatabase -identity MailboxDatabase | select rpcclientaccessserver
Now to modify th attribute to point to CAS server,
Set-MailboxDatabase -identity MailboxDatabase -RpcClientAccessServer ‘CASServerFQDN’
Once this change was performed, outlook connected directly to CAS and the connection was established. Later it was identified that there were some network/port issues which prevented the connection. The above change was then reverted and the RpcClientAccessServer attribute was pointed to CAS array.