For long-running processes, it might be useful to see the graphs produced in real-time. We should do this through the existing graphing trait, but expand it to include pluggable backends. This might include the following:
The current solution, where output is only generated when the detector is closed. (Low feedback levels, but simple and already implemented)
A realtime output, where graphs on disk are updated as new measurements arrive. (Much higher load on the detector's host, but useful to look at for long processes)
Realtime output to InfluxDB -> Grafana. (Ideal, since Grafana is dynamic and powerful, and would offload the graph creation overhead to an on-demand process that's designed for it)
In GitLab, by Daniel Oosterwijk on 2019-11-29
For long-running processes, it might be useful to see the graphs produced in real-time. We should do this through the existing graphing trait, but expand it to include pluggable backends. This might include the following: