We use Fly to host our applications for a community event for the game Path of Exile. Our current schedule is that we run 7 day events per quarter, more or less.
Post each event we do a retro of our performance, and usually check metrics for potential errors or slow endpoints. Unfortunately, the metrics explorer on the dashboard doesn’t allow arbitrary datetime ranges or such, which would be very useful for us!
The best (but perhaps not he easiest) way to achieve this would be to have your own Grafana dashboard and have the fly.io prometheus metrics as a datasource. Metrics on Fly
There’s no way to increase metrics retention yet. We have some vague plans on doing that sometime in 2022, but for now I’d recommend running a job that grabs the metrics you care about writes them into another data source. We actually aggregate metrics into Postgres for our own purposes!