I'm trying to figure out how to filter range vector correctly and then use it to select other metric values and do some aggregation over it, ie. something like
count_over_time((metric_a and (metric_b > 95)[30m])
The issue I do not understand is if I do not put scrape interval after lookbehind windows into square brackets, I do not get all values of metric_a, just - as it seems - the first and the last one.
As I learnt in other related question, and after some experimentation, I ended up with this
count_over_time((metric_a and (metric_b > 95))[30m])[30m:10s])
and then I (seem to) get all values I want. The scrape interval I specify should match the real, but I may not know it. How can I just get all values from a range vector and filter them without knowing/specifying scrape interval?