Oops again - Exchange 2003 and DCPROMO :D

[badass]

In my "virtual envirmoment" that I =use for my own training in order to remove all single points of failure from a cluster, I DCpromo'd the second node. Before engaging my brain

So, we have an exchange organisation like this:

server01 = DC, holder of all FSMO roles, primary node in active/passive cluster , DNS server etc
Server02 = member server, DNS server, secondary node in active/passive cluster
Server03 = member server, configured as front end server
I've ran DCpromo on server02
Place bets now as to whether I can get server02 working without giving up and removing all traces of it from AD, reformatting and reinstalling etc.
If this takes more than an hour to get working, I'll give up and rebuild.

P.S. I bet some muppet points out that promoting/demoting a DC with exchange already installed is not an MS supported configuration and is a bad idea

[Moby-Dick]

you goon

I wouldn't cluster a DC for a start

[badass]

you goon

I wouldn't cluster a DC for a start

Ahh the beauty of a training envirmonent - good pratice can go out of the window

[dave87]

Heh, have fun

I think you'll try for an hour, then go sod it & rebuild it

[chuckskull]

8.22pm, I reckon he's still going at it

[Shorty]

So let's take a look at this.

You built an exchange failover cluster & DC promo-ed it..?



What kind of configuration were you aiming for exactly

You do realise that promoting/demoting a DC with exchange already installed is not an MS supported configuration and is a bad idea..?

I reckon he's still going myself and probably has half the night. Shall we make him a coffee..?

[badass]

Heh - the answer is known. I can confrim that one of the above is definately correct. Ill see what people think over today before announcing the result.
I decided as soon as I had done it that I was going to devote a max of an hour attempting a fix before giving up - so a decision was made within an hour
I have to add here that if this was a production environment, none of this would or could have happened as everything would have been planned and neither node would have been considered as a DC anyway.

[Moby-Dick]

I suppose its an interesting "what if" situation

[badass]

Well anyway - yee (including me) of little faith.
It just worked - checked logs etc before doing anything, killed server01 to initiate failover and it failed over successfully and tested fine.
I was suprised.
I'm still not going to try this in a production environment though