monitoring - Track Events with Prometheus Counters -
using prometheus things per second works great , i've had great success rate , irate. @ loss how graph that's happening , big deal.
so have counter incrementing that's called job_failed. whenever happens shows in instant-vector. if graph directly goes , see bump in graph, isn't giving me clear enough indication job has failed. i'd have spike in zeroed graph.
if rate(job_failed[15s]) spike - it's per second spike it's value 0.1 although change want 1. tried increase(job_failed[1m]) not adding correctly, leaving me values 2.18 etc.
is there way see single spike? seems rather trivial thing can't figure out.
prometheus suited more high volume low volume events, @ low volumes artifacts how keep things accurate on average show up.
so example rate(job_failed[15s]) increase of 1 on 15 seconds 1/15 = 0.066/s. rounding make show 0.1.
https://www.youtube.com/watch?v=67ulrq6dxwa goes more detail how works.
the short version you're doing way it.
Comments
Post a Comment