>oh of course I'm already on call for the hardware stuff.. but I'm not the one who has to go down to the colo and find another motherboard to swap. eh, my expe

>oh of course I'm already on call for the hardware stuff.. but I'm not the one who has to go down to the colo and find another motherboard to swap. eh, my experience is that actually touching the hardware is a very small part of the hardware janitor job. the hard part, in my experience, is figuring out when it's a software problem vs. when you need to go swap that motherboard. This is why I build all my servers rather than going dell and buying a service contract... at least on the affordable service contracts, sure, they will swap out the motherboard if you can prove that's the problem, but if it's an intermittent problem, you are going to have to work damn hard to get them to do the swap. (to be clear, dell is fine when the problem is obvious.) To me, the actual hands-on-the-hardware part is trivial when compared to the effort required to track down an intermittent problem. I mean, on ec2 they won't troubleshoot intermittent hardware problems for you, but at least you can shoot the node and bring it up on fresh hardware without arguing with anyone. (I have some... anger issues left over from dealing with dell support. )