Incident Response

General triage procedure for Tuitbot issues.

Step 1: Check Health

# Liveness probe (no auth):
curl http://localhost:3001/api/health

# Deep health (requires auth):
curl -H "Authorization: Bearer $(cat ~/.tuitbot/api_token)" \
  http://localhost:3001/api/health/detailed

The detailed endpoint returns: - status: healthy, degraded, or unhealthy - checks.database: reachable, latency, WAL mode - checks.runtime: running, task count - checks.circuit_breaker: state, error count, cooldown

Step 2: Check Logs

# If running via systemd:
journalctl -u tuitbot -n 100 --no-pager

# If running in tmux/screen:
# Check the terminal output or log file

# If running as server:
journalctl -u tuitbot-server -n 100 --no-pager

Look for: - error or warn level entries - AuthExpired messages -> see Auth Expiry - Circuit breaker OPENED -> see Rate Limit Storms - Database errors -> see Database Maintenance

Step 3: Identify Subsystem

Symptom	Subsystem	Next step
AuthExpired / 401 errors	Auth	auth-expiry.md
429/403 errors, breaker open	X API rate limits	rate-limit-storms.md
Database unreachable	Storage	database-maintenance.md
No posts going out	Posting queue	Check runtime status, approval queue

Step 4: Validate Fix

tuitbot test

This checks authentication, X API connectivity, and LLM provider.