Clustered MSMQ can't send to queue in another forest

Question

I have MSMQ running on a two-node Windows 2003 cluster. When I try to send transactional messages to a queue on a Windows 2008 server in another forest, the messages get stuck in the cluster's outgoing queue. The outgoing queue says "waiting to connect" and has the destination server's IP for the next hop.

I can send messages from the cluster to Windows 2008 servers in the same forest. I can also send messages from either node of the cluster to Windows 2008 servers in the same forest and the other forest. I can send messages from a Windows 2008 server in the other forest back to the cluster.

Edit: Running a network capture on the servers shows that the Windows 2008 server is receiving packets from the cluster, but they're not being passed up the layers to the MSMQ process.

Answer 1

First thing I would do is make sure the Windows 2008 destination is listening on port 3527. http://blogs.msdn.com/b/johnbreakwell/archive/2010/03/05/port-3527.aspx

[March 5th] I would look at network traffic with Network Monitor or Wireshark.

What does the outgoing queue status do? You can restart the MSMQ service to start the outgoing queue connecting. The outgoing queue may be going through a "connecting - connected - disconnect after idle timeout - waiting to connect" loop. Or it may be "connecting - failing - waiting to connect" loop. Depends what the root cause is.

If it never connects then the 3-way TCP handshake never completes. Can you connect to a network share on the remote server? Basically eliminate all the basic networking issues.

The fact the server is in another forest shouldn't matter too much. The sender just needs to resolve the destination to an IP address, assuming you are not already addressing the messages by IP address.

Cheers

John