Hello everyone,

I'm setting up an Exchange 2010 environment as part of a migration from our single physical Exchange 2007 server.  The 2010 environment looks like this:

2x Barracuda virtual load balancing appliances
2x Ex 2010 Standard CAS servers, in a CAS Array, virtualized
2x Ex 2010 Enterprise mailbox servers, in a DAG, virtualized

The virtualization platform is VMware 5.0.  I followed Microsoft's documentation on setting up the 2010 environment and had no issues in any part of that setup.  With the load balancers, I followed Barracuda's documentation on setting things up and even had them on the phone double checking and verifying the settings were correct, which they said they were.

The problem I'm experiencing is that when I move a mailbox into the 2010 environment and start up Outlook, I run into considerable connectivity issues.  I'll explain further --

By default, it sets itself up in Cached Exchange Mode.  From there, when Outlook opens, the pop-up bubble saying I've lost connection shows up within a minute or so.  When I open the Exchange Connection Status screen (CTRL+Right Click on the Outlook icon in the systray) I can watch as the connections to my CAS Array address disconnect and reconnect.  In the Req/Fail column, I see a number of Fail attempts alongside the Requests, and average response time is well within normal limits.

In the event log, the only thing I see on my client machine is Outlook information even 26; stating that Outlook has lost connection, then regained it.  I'm seeing no errors on the Exchange even logs, and no events/warnings are showing up from the load balancers.

From Barracuda, they recommended that I disable one of the two CAS servers in the load balancer configuration.  This would allow my client connection to choose a specific CAS server, as well as rule out persistent connection issues (so they told me).  I had done this and to no avail.  My next step was to open a ticket with Microsoft, but since it's late on a Friday, I think I'll wait until Monday and see what the community may come up with.

Many thanks,

Rick

Hi,

Please try to run Test-OutlookConnectivity to verify the connectivity.

For details:

http://technet.microsoft.com/en-us/library/dd638082(v=exchg.141).aspx

Since all users have the issue, please also run the EXBPA to check the health of servers.

http://www.microsoft.com/downloads/details.aspx?FamilyID=DBAB201F-4BEE-4943-AC22-E2DDBD258DF3&displaylang=en

Hi Kent, thank you for the response.

I ran Test-OutlookConnectivity this morning and did indeed find an error message:

Fatal Autodiscover failure EXPR = <Null>.
    + CategoryInfo          : OperationStopped: (Microsoft.Excha...onnectivityTask:TestOutlookConnectivityTask) [Test-OutlookConnectivity], FatalAutodiscoverException
    + FullyQualifiedErrorId : F79D8CC4,Microsoft.Exchange.Monitoring.TestOutlookConnectivityTask

I have to assume there is something wrong with Autodiscover, so I will pursue that today and see what's going on with it.

Oops, I should clarify after running the test a couple different times.

The error message above was found when running the following command:

Test-OutlookConnectivity -MailboxCredential $cred -protocol:http -Identity ricktest | fl

I used the "Get-Credential" function to fill the $cred variable.  However when I ran the command using TCP as the protocol, it appears everything was successful.  I do not have Outlook Anywhere configured yet, so I have to assume that's why the HTTP protocol test failed.  I'll run the ExBPA now and look for issues there.

Health Check was run against just the four Exchange 2010 servers I have and there were some errors that it reported, though I don't know how severe any of these necessarily are.

• Active Directory Forest -- Unknown Schema Extension Version
  "The Exchange schema extensions are from an unknown version.  Schema version detected is 14734
• Active Directory Sites -- 'MaxPageSize' is set too high
  "The maximum LDAP page size for the default query policy is set too high and may cause performance problems.  The default of 1000 is recommended.  Current value: 20000
• Domain: XYZ -- Unrecognized Exchange Signature
  "Active Directory domain 'XYZ' has an unrecognized Exchange signature.  Current DomainPrep version: 13040"
• Server: CAS1,CAS2,MB1,MB2
  VMware Detected -- Please review the support policy detailed in KB article 897615
  WMI server startup account -- Is not configured to start up as the computer account (LocalSystem).  Current startup account is: localSystem

On the four server errors, I have to assume those can be ignored (from the standpoint of the connectivity issues I'm having).  I'm not sure about the Schema Extension and Unrecognized Exchange Signature errors, as all of the readiness checks passed when I installed our first Exchange 2010 server into the 2007 group.

I should also note that when I run the exact same health check from the Exchange 2010 servers itself, the only error I get is the MaxPageSize error.

This issue has been solved.  It turned out to be a problem not within the Exchange environment but in the load balancers.  They were supposed to be set up in an active/passive cluster, but somehow the load balancers were acting as an active/active cluster, dropping connections constantly.  Worked with Barracuda support and resolved it.