mimir-prometheus-troubleshoot

"帮助构建高效的Mimir/Prometheus查询,解决指标问题,避免高基数问题,并为聚合、记录规则和性能推荐最佳实践。"

📖 中文导读:"帮助构建高效的Mimir/Prometheus查询,解决指标问题,避免高基数问题,并为聚合、记录规则和性能推荐最佳实践。" 以下正文为该技能的原始 SKILL.md 内容(英文),上方安装命令可直接使用。

name: mimir-prometheus-troubleshoot description: "Help craft efficient Mimir/Prometheus queries, troubleshoot metric issues, avoid high-cardinality problems, and recommend best practices for aggregation, recording rules, and performance."

Mimir + Prometheus Troubleshooting & Query-Builder Skill

What this Skill does

Use this skill whenever a user needs help with:

  • PromQL queries
  • Metric debugging
  • Missing data / gaps
  • Cardinality optimization
  • Aggregation strategy
  • Recording rules

Best Practices

Low-cardinality label selection

Use labels such as:

  • job, instance, service, cluster, namespace, env

Avoid:

  • user_id, session_id, request_id, raw UUIDs

Always narrow time ranges

Prefer "5m", "15m", "1h".

Use correct aggregations

  • rate() for counters
  • sum by (...) for grouping
  • histogram_quantile() for latency

Suggest recording rules if query is heavy

Example Queries

User Request PromQL
"Error rate for payments in prod" sum by (job) (rate(http_requests_total{job="payments", env="prod", status=~"5.."}[5m]))
"Latency p95 for frontend" histogram_quantile(0.95, sum by (le) (rate(http_request_duration_seconds_bucket{app="frontend"}[5m])))

When to Suggest Loki or Tempo

For:

  • request IDs
  • root-cause event-level debugging
  • full request paths

→ Recommend Tempo + Loki correlations.

Limitations

  • Skill does not run PromQL