In our Exchange 2000 Server environment, whenever a Global Catalog (GC) server fails, we must reboot all Exchange servers that were talking to the GC server to force them to find another GC server. Is this typical behavior?
Under typical circumstances, Exchange 2000 Service Pack 2 (SP2) and later should automatically recover from the loss of a GC server. However, several things, most of which relate to network problems between the Exchange server and the recovered GC, can interfere with this recovery. In Exchange 2000 SP2 and later, you can increase the level of diagnostic logging on the MSExchangeDSAccess service and look for event ID 2080 in the event log. Every time Exchange's DSAccess component discovers a domain controller (DC) or GC server, it logs this event. The Microsoft article "XADM: Event ID 2080 from MSExchangeDSAccess" (http://support.microsoft.com/?kbid=316300) explains the fields in the event message and how you can use them to troubleshoot the problem.