allow metric exclusions about druid-exporter

[iamabhishek-dubey]

You can actually do that at Prometheus level.
https://www.robustperception.io/dropping-metrics-at-scrape-time-with-prometheus

[bbecker-te]

Thanks, I will look into it.

[bbecker-te]

The metrics emitted by the druid-exporter are all generically grouped into a single druid_emitted_metrics" metric. Individual metrics are specified with the metric_name label. Here are some sample lines from the prometheus scrape target to illustrate:

druid_emitted_metrics{datasource="http_1m_rollup",host="10.202.7.177:8100",metric_name="jetty-numOpenConnections",pod="10-202-7-177",service="druid-middleManager-tier2"} 0
druid_emitted_metrics{datasource="http_1m_rollup",host="10.202.7.177:8100",metric_name="jvm-bufferpool-capacity",pod="10-202-7-177",service="druid-middleManager-tier2"} 5.25921679e+08
druid_emitted_metrics{datasource="http_1m_rollup",host="10.202.7.177:8100",metric_name="jvm-bufferpool-count",pod="10-202-7-177",service="druid-middleManager-tier2"} 7

While the suggestion of using metric_relabel_configs works to drop or keep metrics by name, it does not seem like it would work in this case, because we want to restrict by label, not by metric name (everything is part of the druid_emitted_metrics metric). Hopefully I am misunderstanding and there really is an easy way to do it.