Yep, our host had a 'fault-tolerant' switch fail. We should be properly back now!
A core (supposedly fault-tolerant) switch stack started having trouble passing traffic between stack members today, causing widespread and semi-random issues.
At this time, we believe the issue is resolved, but we are continuing to monitor very careful. We will also, of course, be analyzing the failure of these switches in great detail to determine our best course of action to prevent the problem from reoccurring.
We will leave this issue open for an hour or two while we go over the system with a fine-toothed comb to make sure all is well.
We are very sorry for the trouble!
Cat emoticons to come.