The API garbled a few posts (540262, 540259, 540250), and these were creating issues for the Global stream in particular. They were removed from the stream, and will be looked into. It appears to be markdown link parsing.
Confirmed: "We are currently seeing network issues at the St Louis data center, and are currently working with the issue at this time. We will update your team as we have information available."
All services were down from 12:29 to 12:42. As far as we can tell, it was an Internet outage in the building. Nothing on the server was affected.
There is a complicated bug that sometimes causes a leak, and eventually interferes with database connections. This is fixed on v0.9.0, but the fix is caught up in a major rewrite of App Streams.
There's a good chance the v0.6.1 patch will be out by then.