What problem does it solve? Operating a production Ava cluster means triaging a constant alert stream and diagnosing recurring failure classes without a clear methodology. This Skill gives the cluster operator a structured playbook for classifying alerts, tracing root causes, and verifying fixes instead of chasing symptoms. ## Core Features & Use Cases - Alert triage workflow: Classify each alert window's events as known/closed, transient, or action-needed, with attribution by machine, service, and incident history. - Diagnosis playbooks: Symptom-to-cause-to-response guides for disk bloat (Postgres checkpoint growth), memory leaks, port conflicts, session-backend mismatches, tripped schedule breakers, and stalled message delivery. - Post-rollout verification: A checklist that catches stale processes, ghost daemons, and duplicate schedule sessions that pid-based status checks miss. - Use Case: After a rollout, an operator sees repeated 'duplicate session' launch failures on a schedule. The playbook identifies the session-backend mismatch as the cause, directs clearing the stale session and restarting via the schedules API, and requires verifying the next fire before closing the incident. ## Quick Start Ask the assistant to triage the last two hours of Ava cluster alerts and diagnose any repeating failure patterns using this playbook.
