This week we've had five win8 slaves reimaged, and zero are working: four haven't connected to a master, one which didn't ever get disabled while it had its hard disk replaced claims to be connected, in a sort of approximate way, but it isn't doing anything. Of the ones I pointed him at, nthomas said none would actually connect via either vnc or ssh. It's our second-worst test slave platform for backlog after 10.8, but 10.8 has 100% of its slaves working, while we're now up to 8% of our win8 slaves out of service, between this bug and bug 1004813.