I have a newly built Exch 2007 mailbox cluster (2 node, single copy) which was missing the virtual directories (/exchange, /exadmin, and /public) on both nodes needed for cas/owa redirection to work. The cluster is part of a mixed 2003/2007 organisation. I have followed the procedure here http://support.microsoft.com/kb/947802/en-gb to create the virtual directories from the active node and all appears to work as expected. However the second node is still missing the virtual directories, as such errors are experienced when running from this node. I have tried re-running the new-owavirtualdirectory commands from the second node but receive errors stating that the virtual directorie(s) already exist for the mailbox server. I have tried manually creating the virtual directories but this results in owa returning a internal server (500) error when running from this node. Any advise on how to overcome this problem? Many thanks, Phil

The Client Access Server role, which is not installed on your cluster, is what is responsible for these directories. I'd recommend getting a book on Exchange before proceeding. This is a fundamental concept. Mike Crowley Check out My Blog!

Thank you for the reply - although not terribly helpful if you don't mind my saying so! I am well aware of the Client Access role which we have running on separate hosts. The virtual directories are required on the 2007 mailbox servers to accommodate CAS redirection for OWA, that is, because we are running a mixed 2003/2007 environment users access webmail via the address CASserver/exchange, which for, 2007 hosted mailbox users redirects to CASserver/owa. As per the referenced support article above, this requires the virtual directories to be present on the 2007 backend. So my question/problem stands, how do I create the virtual directories on the second mailbox cluster node? Thanks.

Phil, my mistake! It was unclear you already had a CAS in your environment, and I thought you were trying to proceed with only a SCC cluster. When you completed the adsi tasks mentioned in your article, did you do this to the CMS or the node itself? I believe the owner of the CMS is the only box that needs to have the directories. If you’re having a problem with failover, have a look here: http://social.technet.microsoft.com/forums/en-US/exchangesvravailabilityandisasterrecovery/thread/70f012b8-757d-4f18-a45d-61c05f1d8ab8/ Mike Crowley Check out My Blog!

I have now resolved this issue by exporting the IIS configuration from the working node and importing it on to the alternate node, thus creating the required virtual directories on both nodes. Mike - thanks for the response, for info, the ADSI task to create the HTTP object was performed against the CMS, indeed the physical nodes do not appear/register within the Administrative group. Hence I could understand to an extent why the error was occurring, as the virtual directories were already registered with the CMS, it was just a case of creating the IIS config on the second node. Although I now have this working I am still not 100% convinced that it should have been necessary to configure the servers in this way - surely this is basic configuration requirements that should be accommodated by default. We will be building some additional mailbox clusters soon, it will be interesting to see if the same issue occurs..! Phil

I agree with your concern. I dont know that all that was necessary. Why did you create the IIS directory in the first place - was it because of an error? I would expect the 2nd node to automatically create the vdir when a failover occurs. Mike Crowley Check out My Blog!

The virtual directories were never present (on either node) following the initial install, as such an error 404 was returned when connecting via the CAS URL (interestingly though this only occured when connecting via redirection from the /exchange url, when connecting directly from the /owa url everything worked a treat - I have seen this behaviour documented elsewhere and can only assume that the redirection process connects to the CMS in a different manor). Anyway, this was resolved by running the new-owavitualdirectory commands on the active node, however in performing a cluster failover the vdirs never appeared on the 2nd node and the 404 error returned. I had therefore (possibly incorrectly) assumed that they should have been there by default! It would be good to know for sure if the vdirs should be created as part of the failover? So far I have not been able to find any formal documentation to support if this is the case! Furthermore there is nothing in any of the logs to suggest errors during the failover process... Thanks, Phil