Hello, I have a customer whose Outlook and Blackberry clients are occassionally losing connectivity to Exchange and the only solution the customer has found is to reboot the server. The server is Windows Server 2003 SP2 with Exchange 2003 SP2. The server is also a domain controller and serves DNS. It is also a global catalog server. There are three domain controllers in the environment, two at this particular AD site. However, when I look at the directory access tab in Exchange system manager it only knows about itself as a DC/GC. Shouldn't it know about both the DCs at this site? And what would cause the clients to think the server is down? They use DPM 2007 to backup the environment but the outages are infrequent and do not correspond with the backup jobs.

Hi, Please provide some more information Both DC within site also is GC ? Have you run dcdiag and netdiag ? Please also check the network latency Dinesh

Yes, all the domain controllers are global catalog serves. NetDiag reports errors related to WINS (there is no WINS in the environment.) DCDiag passes all tests except examination of the system log. The messages reported refer to a kerberos error and a browsing error.

When Exchange is installed on a DC it will look only to itself. That's reason #32 not to install Exchange on a DC. Ed Crowley MVP "There are seldom good technological solutions to behavioral problems."

Got that. This was not my design or install. What would cause Exchange to think there were no available GCs and how can I see that that indeed is the problem? Would you consider a Tazer a good technological solution to a behavioural problem? :-)

Does anyone know of a way to restart the global catalog functionality without restarting the domain controller? This would be a Windows Server 2003 domain controller. I really think what is happening is that Exchange is losing the ability to communicate with the GC. Thanks.

That's because that's the way it is. Microsoft would call it "by design". And if you think about it, it makes some sense. Why would you install Exchange on a domain controller if you didn't want it to use it?Ed Crowley MVP "There are seldom good technological solutions to behavioral problems."

Your real solution is to build another Exchange server on a member server and move everything to it, then remove Exchange from the DC. Sorry that I don't have any better solutions.Ed Crowley MVP "There are seldom good technological solutions to behavioral problems."

It turns out that this issue occurs during a Data Protection Manager 2007 backup cycle. I've finally been able to correlate the outage time with events in the logs. There are VSS errors in the logs. They are: Source: ESE Event ID: 206 Description: Information Store (4304) First Storage Group: Database D:\Exchsrvr\MDBData\termTMmbs\Termed TM Mailboxes.edb cannot be incrementally backed-up. You must first perform a full backup before performing an incremental backup. Source: MSExchangeIS Event ID: 9611 Description: Error code 0xfffffdd0 when freezing the Storage groups Source: ESE Event ID: 2002 Description: Information Store (4304) Shadow copy instance 282 encountered error -560 on freeze. Any help or suggestions would be appreciated.Mike H

Get a bigger boat? :) When are these backups running? ( And it sounds like you need to run a full backup first)

I'm not that familiar with DPM but I believe it is configured to backup theExchange server every 60 minutes.Mike H

Ok, you may want to ask about best practices in the DPM for Exchange Forum: http://social.technet.microsoft.com/Forums/en/dpmexchbackup/threads It appears however that the server may not be sized correctly for those hourly backups.

I was not aware of that forum. Many thanks.Mike H