Comment 7 for bug 1813104

logsink.go needs a bunch more logging added, so that you can set DEBUG and see the level of granularity that we see on /api. It doesn't have to be *every* log message, but you should see a message every time an agent connects, what agent, for what model, and when they disconnect.

We also need to get the metrics around logsink into the engine report and prometheus metrics.
Unfortunately the prometheus metrics we have today are set up during NewServer, but the logsink endpoint doesn't exist until server.Endpoints() is called. And the server doesn't have a reference to the metricsAdapter/PrometheusConfig at that point. Probably we just change the Server to forward metrics from other handlers and change the metricsAdapter to forward from the Server.