Thanks everyone for your patience with this rolling out!
I found a bug in the monitoring client that I’d introduced in an unrelated change a late Sunday evening a couple months ago. We missed it in testing because most of the beta monitors never got that particular version.
In each test the monitor sends multiple “probes” to a server. The intention is that it picks the best response it gets as the monitoring result. The bug I introduced made it so if any of the probes had an error the whole result got marked with that error, rather than ignoring it in favor of a successful response.
it’s fixed now and the monitors run by the project have been updated. The other monitors should get updated in the next ~12 hours or so.