http_request_duration_seconds_bucket{job="n9e-proxy", le="0.1"} 500 http_request_duration_seconds_bucket{job="n9e-proxy", le="1"} 700 http_request_duration_seconds_bucket{job="n9e-proxy", le="10"} 850 htt http_request_duration_seconds_bucket{job="n9e-proxy", le="20"} 1000 http_request_duration_seconds_bucket{job="n9e-proxy", le="+Inf"} 1000 根据这个数据，我们可以计算出落在各个延迟区间的请求数量，如下： 0 ~ 0.1 : 500 0.1 ~ 1 : 200 ，我们可能希望以模块为颗粒度，分别计算每 个模块的90分位延迟，写法是： histogram_quantile( 0.9, sum by (job, le) (rate(http_request_duration_seconds_bucket[10m])) ) 注意，这里通过job标签来区分模块，le是计算histogram_quantile必须的，所以也要放到sum by后面，如果我们要计算全部数据的

temperature_celsius WHERE \ country=" germany" GROUP BY city rate(errors{job=" foo" }[5m]) / rate(total{job=" foo" }[5m]) SELECT errors.job, errors.instance, [...more labels...], \ rate(errors.value, 5m) / rate(total rate(total.value, 5m) \ FROM errors JOIN total ON [...all label equalities...] \ WHERE errors.job=" foo" AND total.job=" foo" Richard Hartmann & Frederic Branczyk @TwitchiH & @fredbrancz Intro to Prometheus

Quick intro OpenMetrics Outro Thanks! Thanks for listening! Questions? Email me if you want a job in Munich. See slide footer for contact info. Richard Hartmann, RichiH@{freenode,OFTC,IRCnet}, richih@{fosdem