What problem does it solve?
Manually reviewing ClickHouse system metrics to identify unusual cluster behavior is time-consuming and prone to missing subtle performance regressions, error spikes, or resource issues that impact production workloads.
Core Features & Use Cases
- Baseline Comparison Anomaly Detection: Compares recent 1-hour activity to a preceding 24-hour baseline to flag statistically significant deviations.
- Multi-Domain Monitoring: Covers error rate spikes, query duration regressions, query volume anomalies, memory usage outliers, part count explosions, and replication lag.
- Actionable Interpretation: Provides clear thresholds to distinguish real anomalies from statistical noise, plus cross-references to specialized skills for root cause resolution.
Use case: A ClickHouse operator can use this skill to quickly confirm if a recent deployment caused a 2x increase in query failures or p95 latency without manually parsing hours of query logs.
Quick Start
Use the anomaly-detection skill to run the error rate spike check on your ClickHouse cluster to determine if recent query failures are abnormally high compared to the previous 24-hour baseline.
