Downtime was because the server had a SSD that failed. No data from LoadAverage was on that SSD, but the hypervisor OS was, so that made things dicey. Sorry for the fuss. There will be a hopefully short downtime sometime in the future to remove the failed drive. A replacement is already in place.
Why the drive failed is unknown, it seemed rather out of the blue, but as far as we can tell, the controller just upped and went. It won't even SMART test.