eef835d2aa
Adds operational resilience tooling: - resilience/health-check.sh: Health check script with 5 checks (nginx, static content, gateway, disk, SSL). Supports --auto-restart and --verbose modes. - resilience/service-restart.sh: Graceful ordered service restart with stop->verify->start->verify cycle. Supports --force mode. - Fallback logic for when gateway is unreachable (graceful degradation to static pages) All scripts are self-contained, no external dependencies, work on common Linux distros.
92 lines
2.6 KiB
Bash
Executable File
92 lines
2.6 KiB
Bash
Executable File
#!/usr/bin/env bash
|
|
# service-restart.sh — Graceful service restart for the-door
|
|
# Usage: bash service-restart.sh [--force]
|
|
#
|
|
# Performs ordered restart: stop -> verify stopped -> start -> verify started
|
|
# with health check confirmation.
|
|
|
|
set -euo pipefail
|
|
|
|
FORCE=0
|
|
for arg in "$@"; do
|
|
case "$arg" in
|
|
--force) FORCE=1 ;;
|
|
*) echo "Usage: $0 [--force]"; exit 1 ;;
|
|
esac
|
|
done
|
|
|
|
log() { echo "[$(date '+%Y-%m-%d %H:%M:%S')] $1"; }
|
|
|
|
# ── Stop ────────────────────────────────────────────
|
|
stop_services() {
|
|
log "Stopping services..."
|
|
|
|
if command -v systemctl > /dev/null 2>&1; then
|
|
systemctl stop nginx 2>/dev/null && log "nginx stopped" || true
|
|
elif command -v nginx > /dev/null 2>&1; then
|
|
nginx -s stop 2>/dev/null && log "nginx stopped" || true
|
|
fi
|
|
|
|
# Stop gateway if running
|
|
local gw_pid
|
|
gw_pid=$(lsof -ti:8000 2>/dev/null || true)
|
|
if [ -n "$gw_pid" ]; then
|
|
kill "$gw_pid" 2>/dev/null && log "Gateway stopped (PID $gw_pid)" || true
|
|
fi
|
|
|
|
sleep 1
|
|
log "All services stopped"
|
|
}
|
|
|
|
# ── Start ───────────────────────────────────────────
|
|
start_services() {
|
|
log "Starting services..."
|
|
|
|
# Start nginx
|
|
if command -v systemctl > /dev/null 2>&1; then
|
|
systemctl start nginx && log "nginx started" || { log "FAILED to start nginx"; return 1; }
|
|
elif command -v nginx > /dev/null 2>&1; then
|
|
nginx 2>/dev/null && log "nginx started" || { log "FAILED to start nginx"; return 1; }
|
|
fi
|
|
|
|
log "All services started"
|
|
}
|
|
|
|
# ── Verify ──────────────────────────────────────────
|
|
verify_services() {
|
|
local host="${1:-localhost}"
|
|
|
|
log "Verifying services..."
|
|
|
|
# Check nginx
|
|
if pgrep -x nginx > /dev/null 2>&1; then
|
|
log "nginx is running"
|
|
else
|
|
log "ERROR: nginx failed to start"
|
|
return 1
|
|
fi
|
|
|
|
# Check static file
|
|
local status
|
|
status=$(curl -s -o /dev/null -w "%{http_code}" --connect-timeout 5 "http://$host/" 2>/dev/null || echo "000")
|
|
if [ "$status" = "200" ]; then
|
|
log "Static content verified (HTTP $status)"
|
|
else
|
|
log "WARNING: Static content check returned HTTP $status"
|
|
fi
|
|
}
|
|
|
|
# ── Main ────────────────────────────────────────────
|
|
log "=== Service Restart ==="
|
|
|
|
if [ "$FORCE" = 1 ]; then
|
|
log "FORCE mode — skipping graceful stop"
|
|
else
|
|
stop_services
|
|
fi
|
|
|
|
start_services
|
|
verify_services "${HEALTH_HOST:-localhost}"
|
|
|
|
log "=== Restart complete ==="
|