A good hard reboot seems to have fixed things ... this is mysterious and I don't like it.
At a guess patching made some part of the web stack unhappy and for some reason the first reboot wasn't good, but sending a hard reboot via the management UI seems to have actually done it.
That disturbing thing decommissioning servers, its like leaving the house and wondering if you left the oven on - "Did I leave any files on that thing? Is there any scripts or services still pointed to it?"
There's going to be some downtime and migrations of stuff coming up like in a month or two, this is basically warning to stop people from moaning when it happens. It'll be minimal, trust me.