Clowns to the Left of Me, Jokers to the Right

Forgetting my building access badge this morning cost me an hour to go back home and return at the height of rush hour, a commute made even worse by all the tourists riding the same line on their way to the stadium. Packed into the subway some weasel looked at me and asked me to move left.

"I can't move any further left."
"But more people are trying to get in the train."
"Look! I'm already shoulder-to-shoulder with this woman." She didn't look up from her book.
"But, but I have no room," he whined, trying to push me a bit.
"There is no room to the left. If I move any further left I'll be fucking this lady."
The not-at-all attractive woman smiled; the weasel shut up. Fuckwit.

That was just the prelude. I got my morning coffay. We have these big filter coffee machines which dump into 2-litre, pump-action thermos containers. As I filled my Supermug (which holds about 5 "normal" cups of coffay") I noticed foam. The fuckwit cleaners left soap in the thermos and I had to wait another 15 minutes as I brewed another round.

On to the tickets and I got a doozie:

We're in the process of designing a highly redundant environment and after some research we have decided to test a combination of Cluster Services and load balancing.

Our question is directed at how the $YourBigApp name server and app servers are installed on a cluster server. Our IT department prefers to set an Active-Passive cluster without using a disk array.

"After some research" meaning, "We stuck our thumbs up our asses and pulled out this idea."

The whole point of a cluster is that multiple machines access the same applications and files from the shared disk array. If one machine goes down, the other is still running and handles the load. The name server data has to be stored on the shared disk array. We wrote a great big FAQ on how to set up a cluster with every possible architecture which works.

In some instances this sort of configuration might be possible, but not with $OurBigApp. The name server includes all the system configurations. Without a shared disk array, you'd have to shut down the system in order to replicate the information across all the cluster nodes.

So they don't really understand how a cluster works, they didn't pay attention in class when the entire $OurBigApp system architecture was described, and didn't even look in the Knowledge Base.

While it may appear that I get a sweetie out of all this, I'm certain they're going to come back to me a dozen times whining like children about how they don't want to use a shared disk array and couldn't they maybe just do . This will be followed by them questioning my knowledge of clusters, then their business needs, and finally a request for management involvement and re-assignment to another monkey before finally accepting that their cockamamie idea won't work.

I'm sitting at a brand new desk but already one can make out a slight, my-head-shaped dent in front of the keyboard.

Because my Root Cause: 17-Fuckwit has yet to be implemented, they'll get off lightly with either a Root Cause: 6-No Research or maybe a Root Cause: 6.5-No Understanding. It's hard to choose.

x-posted to HuSi.