The demo Grafana dashboards using Prometheus data are configured to interpret timing metrics as provided in milliseconds. This is incorrect as Prometheus provides these values as seconds. This means that these graphs are showing timing metrics with values 1000x too small.
- Enable collection of monitoring via Prometheus and setup a Prometheus server as described here
- Load the demo dashboards (described by the same README linked to above)
- Interact with AM and observe the dashboards
N/A. The Grafana dashboard configurations need to be updated and a new export committed to AM.