Hello,

I am currently running 2 single role Exchange 2010 servers in coexistence with 2 multirole 2013 servers.  The users on the 2013 setup are the only ones with this issue.  I have a Netscaler in place to handle the load balancing, but for the sake of resolving this issue, I've bypassed that and am using a host file to direct clients to either of the two Exchange servers.  Here's the issue: When a users mailbox DB is active on EX01, but Outlook is balanced to EX02, there is a 45-60 second delay in launching Outlook, it hangs on "Loading Profile".  If they are balanced to EX01 it opens immediately.  If I fail the DAG over and the mailbox is active on EX02 this reverses, so EX02 it opens instantly and EX01 it's slow.  I've opened a case with MS, but they have yet to resolve this after multiple days and escalating it to a higher level engineer.  I should have another call back from them today, but I see no light at the end of the tunnel currently.  Thanks for anything and everything!

Are the URLs within Exchange identical on both servers?

Do they all resolve to the correct place?

I would also use the RCPDAG box (outlook.exe /rpcdiag) to see what Outlook is trying to do during that event. That could give you an indication of the cause of the delay.

Simon.

URL's are the same, typically the URL would resolve to the Netscaler, but I'm pointing directly to each of the Exchange servers in my testing to ensure I reproduce the problem.  When I hit reconnect I can see which one is hanging up, I have four connections in there, one being my 2010 public folders (which is where I thought the problem would be, but it doesn't appear to be).  Exchange Mail is the one that's slow to connect, however it doesn't show failures, Req/Fail is 237/0.

Hi,

According to your description, I understand that Outlook need 45-60 second to load and connect to Exchange server if DAG failover.
If I misunderstand your conern, please do not hesitate to let me know.

By default, it will take more time if DAG failover or move mailbox to another server, Outlook will use autodiscover service to request and connect to Exchange server when it's open.

Please check the response time when Check Name with Netscaler, if the time to resolve is greater than 30 to 40 seconds, please refer to below link for further troubleshooting: https://support.microsoft.com/en-us/kb/15504

This is not in DAG failover situation while Outlook is open, this is the initial launch of the Outlook client.  I have a load balancer that will distribute connections between the two multi-role servers.  If a user gets balanced to a server that the passive database is on (active on other server) they have this 40-60 second delay in launching Outlook.  If they are balanced to the server their DB is active on there is no delay, Outlook opens immediately.

Public Folder connection can cause the delays that you are seeing.
Are you public folder mailboxes in the DAG?

Simon.

Public folders are still on the 2010 server, I haven't migrated them over yet.  When looking at outlook.exe /rpcdiag the public folders connect right away.  The one that isn't connecting right away is "Exchange Mail".  It works fine anytime the CAS and Mailbox DB are on the same server, regardless of which one.  

Do you have a public folder database on the mailbox database of the Exchange 2013 server? Is it the correct one - you can see it on the properties of the database.

CAS is independent of the mailbox role, therefore it should make no difference where the mailbox is located. If you are only seeing this problem when going between the servers, that suggests there is a problem with the configuration.

Are the two servers next to each other on the LAN, or separated by a subnet/WAN connection? Identical in build/patch level?

Simon.

The 2013 server has no public folders on it, none of that has been moved from 2010 yet.  

The servers are next to each other on the LAN.  They are identical and have been upgraded to CU9 when this started, however that didn't fix the issue.

That wasn't what I meant.

On the properties of the Mailbox Database on Exchange 2013, do you have the Exchange 2010 public folder database configured?

There has to be something about your configuration that is causing the problems. Something blocking the traffic or causing name resolution or connection to fail.

Simon.

I've solved my problem!  What it turned out to be was a problem with where I am hosting the servers (AWS).  I upgraded the instance type and as part of that AWS "upgraded" the network adapters.  What I didn't realize is that this changed my settings in the adapters, meaning my DAG replication adapter that was set to not register in DNS was now registering in DNS.  Unchecking that simple box on the network interface fixed my problem!

• Marked as answer by Allen_WangJFMicrosoft contingent staff, Moderator 6 hours 15 minutes ago