mimir-prometheus-troubleshoot

PromQL queries for error rates use rate() functions on counter metrics, while latency percentiles use histogram_quantile() with appropriate time ranges and aggregations. This Skill guides you through constructing these queries with proper syntax and optimization to avoid performance issues in Mimir and Prometheus environments.

Missing metric data typically stems from cardinality issues, scrape failures, or incorrect label combinations. This Skill helps diagnose these problems by teaching metric debugging techniques, identifying gaps in your data, and applying best-practice query patterns to verify data presence across time ranges.

High-cardinality labels degrade performance by creating too many unique metric combinations. This Skill provides cardinality optimization strategies, including label design best practices and query construction patterns that minimize metric explosion while maintaining observability.

Yes, PromQL aggregation functions work across configurable time ranges. This Skill teaches how to apply sum by for dimensional analysis, use histogram_quantile for percentile calculations, and adjust time windows to balance accuracy with performance in your monitoring setup.

Recording rules pre-compute frequent queries and store results as new metrics, reducing runtime query load. This Skill covers recording rule creation, aggregation strategy selection, and optimization patterns aligned with Prometheus concepts to improve dashboard and alert performance.