PDCE = Active Directory PCD Emulator FSMO Role
Windows 2003 Active Directory Environment
It began as random authentication issues around the network, to loss of connections to some server, but not others.
We realized it was authentication issues, isolated to random clients because issues were not evident on neighboring computers with the same logons.
We believe the issue was related to the DC that was servicing authentication to the clients.
We then noticed in a MAC lab (all joined to the domain also) that the clock on the MACS were very inconsistent up to 10 minutes.
Our DC's are single processor Dell PowerEdge 1950 with Xeon 5050 (SL96C) processor (3000/4M/667).
The root cause was a processor in one of the DC; which was the acting PDCE (responsible for external source time sync). Since the system could not keep reliable time with it's external source, it did not service it's windows time clients (the other DC's) which could not then service their clients ( the client workstations ) and all computers began relying solely on thier internal time mechanisms.
Replaced the processor in the PDCE; all back to normal.
What concerns me is that the Dell Diagnostics on the processor did not identify any issues at all.
Finally, after three days of troubleshooting, after another reboot, the Server DIC displayed Processor Failure E1410, and never turned on again. This forced our suspicions to replace the processor, and ultimately resolved our issues.
Is there a better way to test processors for this sort of failure?