Why preemptively query Sia API?

[forresthopkinsa]

Querying the Sia API and then caching the results for Prometheus to later scrape seems to be at odds with the general Prometheus pull-oriented design.

Why not query Sia in response to a Prometheus request? That way, the metrics collection frequency would be solely managed with the Prometheus job scrape interval.

[mattlqx]

This actually is how Prometheus pull design works in every exporter I've seen. Metrics are collected at an interval and state held in memory so that when a request to /metrics comes in, it's just dumping that state and not having to connect out and wait for a result to then be relayed back to Prometheus or whatever consumer.

[forresthopkinsa]

Weirdly, that's been the opposite of my experience. Additionally, this is called out in the Prometheus documentation:

Metrics should only be pulled from the application when Prometheus scrapes them, exporters should not perform scrapes based on their own timers. That is, all scrapes should be synchronous.