Monitoring, alerting, and on-call runbook

[truthixify]

Tier: L (5-7 days) | Type: docs

Context. guides/ops/self-hosted-deployment.mdx (Wave 7) walks getting contracts up. Nothing today tells the operator what to watch, when to page, or how to respond. Teams running Wraith in production need a full runbook: healthy metric baselines, ready-to-import dashboards, alert rules, incident playbooks, and a tabletop exercise. This is more than a page; it is a bundle of ops artifacts, hence the larger tier.

Scope.

• New guides/ops/monitoring-and-on-call.mdx.
• Key metrics: RPC latency, announcement-lag, indexer backlog, watcher event-drop rate, contract error rate.
• Grafana dashboard JSON in guides/ops/dashboards/ (one for RPC, one for indexer, one for contracts).
• Prometheus alerting rules file with rationale per rule.
• Six named incident playbooks: RPC outage, Horizon backpressure, indexer stall, contract mispublish, watcher drop-spike, key-rotation incident.
• Sev matrix mapping incident types to response SLA, cross-linked with the auditor guide's severity ladder.
• Tabletop exercise script for a quarterly drill.

Acceptance.

• Reviewer imports each dashboard into a Grafana instance without error
• Alert rules validated with promtool check rules
• Reviewer walks one playbook end-to-end on a staging deployment
• Renders in mint dev
• Passes Compile docs snippets CI
• Cross-linked with self-hosted-deployment.mdx and reference/auditor-guide.mdx
• Nav entry under Operations

Files. guides/ops/monitoring-and-on-call.mdx (new), guides/ops/dashboards/ (new), guides/ops/alerts/ (new), docs.json.