I’m working with a client this week on migrating them from Exchange Server 2003 to Exchange Server 2007.
Their Exchange 2003 environment is a weird one – a A/A/A/P/P cluster. That’s right – Active/Active/Active/Passive/Passive. Whoever sold them that solution made some serious money! 🙂
I’m moving them to a more standard solution – dual front end servers (CAS + HT) and a CCR backend for the mailboxes. Obviously, we are going to have coexistence for some period of time – there are thousands of mailboxes involved.
One of the standard steps in a co-existence scenario is to create a inter-server-version (InterOp) routing group connector (RGC) which allows the Exchange 2003 environment to communicate (i.e., transfer email) with the Exchange 2007 environment.
Today I finished setting up the CAS+HT servers and the CCR servers and began to set up the InterOp RGC. I used the same syntax I’ve used with dozens of other migrations:
New-RoutingGroupConnector -name Interop-RGC -SourceTransportServers server1, server2, server3 -TargetTransportServers ExchCas1, ExchCas2 -BiDirectional $true -PublicFolderReferralsEnabled $true
And… crap. It bombs.
WTF?!?!?
New-RoutingGroupConnector : Active Directory operation failed on dc3.example.com. This error is not retriable. Additional information: The name reference is invalid.
This may be caused by replication latency between Active Directory domain controllers.
Active directory response: 000020B5: AtrErr: DSID-03152392, #1:
0: 000020B5: DSID-03152392, problem 1005 (CONSTRAINT_ATT_TYPE), data 0, Att 315c30e2 (msExchTargetBridgeheadServersDN)
At line:1 char:26
+ new-routinggroupconnector <<<< -domaincontroller dc3.example.com -identity "Exchange Administrative Group (FYDIBOHF23SPDLT)\Exchange Routing Group (DWBGZMFD01QNBJR)\InterOp-RGC" -sourcetransportservers server1, server2, server3 -targettransportservers ExchCas1, ExchCas2 -BiDirectional $true -PublicFolderReferralsEnabled $true
Well, it has nothing whatsoever to do with “replication latency between Active Directory domain controllers”. I spent far too long heading down that path before I started googling around to see what else might be on the “Interwebs” about this issue.
It turns out that Exchange 2003 assigns indexes to each SMTP Virtual Server on a computer. The New-RoutingGroupConnector command will only connect to SMTP Virtual Servers that have a index value of 1. Is this a bug? Absolutely. Has it been reported? Absolutely. Has it been fixed? Absolutely not. Note that the same error can occur with the attribute msExchangeSourceBridgeheadServersDN – depends on how you construct the cmdlet.
As of the Exchange 2010 beta, this issue still appears.
In a A/A/A/P/P cluster, the indexes for the SMTP Virtual Servers go 1, 2, 3. This means that you can only specify a single SMTP Virtual Server in the New-RoutingGroupConnector cmdlet, and you can only specify the SMTP Virtual Server for the Exchange Server that has the index of 1. This ALSO means that, for the InterOp-RGC to work, that the Exchange 2003 server with the index of 1 must be the routing group master!
Finding those things out cost me several hours of my life. I hope this blog post helps you avoid the same timesink I had!
Until next time…
If there are things you would like to see written about, please let me know!
Follow me on twitter: @EssentialExch