Following the upgrade and on the reboot following the GUI install phase BANG! blue screen of death loop with a 0x0000007B error. It looked to have been related to the driver for the virtual SCSI controller.
We tried the usual BartPE injection of 'LSI Logic' drivers into the installation but to no avail. Having tried a repair install and submitting the drivers using the floppy F6 option again we had the same problem.
Time was key and we had to decide wether we wanted to persevere with trying to resurrect the dead DC or move on with the DC we had left. The boss made the call that this server was dead.
First job was to get the other DC up and running as the master DC in the domain (read: carrying the FSMO roles) so with the aid of the handy bookmarked KB article we seized the roles:
A heads up on this is that once you have selected your server you have to 'go back one level' which is easily overlooked in the fear of losing the domain!
In brief:
START - RUN - ntdsutil
roles
connections
connect to server MyDC
q
seize roletoseize
q
q
The 5 roles in the forest are as follows (in terms of naming for the seizing): rid master, pdc, schema master, infrastructure master, domain naming master. This was the subdomain so we only required the seizing of the rid master, pdc and infrastruture master. Also of note is setting the global catalog, as we only had the one DC left this had to run on the infrastructure master (we moved it after we had a second DC back up).
We brought up a second virtual server image, dcpromo'd it up and got it as the GC using the 'Active Directory Sites and Services' tool:
A single checkbox on the NTDS Settings properties is all there is to set. Needless to say it was a long fretful night and today is still ongoing with changing IP's and DNS updates etc.
As far as I can tell DNS is the magic behind most of this, get that running and the network looks after itself.
No comments:
Post a Comment