couchbase-observabilitylisted

# Couchbase Observability A skill for *monitoring and alerting on* Couchbase clusters in production — metrics, thresholds, Prometheus integration, Grafana dashboards, log aggregation, and health check patterns. Distinct from: - `couchbase-mcp` — calling `admin_stats_*` tools to read current state - `couchbase-security-hardening` — audit log configuration and shipping ## When this skill applies - "How do I monitor Couchbase?" - "What metrics should I alert on?" - "How do I integrate Couchbase with Prometheus / Grafana?" - "What's a good alert threshold for memory / disk / replication lag?" - "How do I know if a node is healthy?" - "How do I aggregate Couchbase logs?" - "How do I set up dashboards for Couchbase?" - "What does the cache miss ratio tell me?" - "How do I detect query performance degradation?" ## Pick the right reference | Question | Read | |---|---| | "What metrics matter and what do they mean?" | `references/key-metrics.md` | | "Prometheus / Grafana setup — scraping, dashboards, recording rules" | `references/prometheus-grafana.md` | | "Alert thresholds — what values should trigger pages vs warnings?" | `references/alert-thresholds.md` | | "Log aggregation — shipping cluster logs to ELK / Splunk / Datadog" | `references/log-aggregation.md` | ## Three core principles **Principle 1 — Alert on symptoms, not causes.** "Disk usage > 80%" is a symptom. "Compaction not keeping up" is a cause. Alert on the symptom (disk), investigate causes during the incident. S