Lightning Talk: How to detect and delete unused metrics?
summary
Unused metrics consume memory and disk; two practical detection methods are proposed: the manual method that pulls all metric names from Prometheus/VictoriaMetrics, parses alerting and recording rules plus Grafana queries and subtracts them, and the simpler method that tracks metric names on ingestion and logs all queried metric names over a day then subtracts to find unused. Use TSDB status to rank metrics by time-series count and identify high-memory candidates, then drop them with metric_relabel_configs or VictoriaMetrics’ drop-by-name/prefix/regex; VictoriaMetrics shows request count and last-query time to confirm zero-use.
AI-written summary · about
related talks
28mThe Sixth Sense: Finding “Dead” Metrics and Silent Alerts in Your Monitoring
28mHow to reduce expenses on monitoring with VictoriaMetrics
19mCutting Through Metrics Cardinality Noise with VictoriaMetrics
24mManage metrics & logging costs with Grafana Cloud + Log Volume Explorer demo
13mLightning Talk: Taming Metric Cardinality: Practical Cost Reduction Wit...
40mScale Your Monitoring Solution With the VictoriaMetrics Ecosystem
06mObservability Diet: Your 5-Step Plan To Trim the Data Fat
11mLightning Talk: “Metrics That Lie”: Understanding OpenTelemetry’s Cardinality Capping...