#!/usr/bin/env bash
# Restart the blog-app Flask server on 0.0.0.0:8090.
#
# Guarantees:
#   - PID file (blog-app/app.pid) tracks the live server.
#   - Old server is killed before new one starts (no "already running" ambiguity).
#   - Health check loops on /healthz up to 30s (1s interval). 启动失败立即报错.
#   - On failure: kill the failed new PID, restore the OLD server if we recorded one,
#     print the last 50 lines of /tmp/blog-app.log, exit 1.
#   - Does NOT touch daily_watchdog.py / cron (independent lifecycle).

set -u
set -o pipefail

SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
APP_DIR="$(cd "$SCRIPT_DIR/.." && pwd)"
PID_FILE="$APP_DIR/app.pid"
LOG_FILE="/tmp/blog-app.log"
HEALTHZ_URL="http://127.0.0.1:8090/healthz"
HEALTHZ_TIMEOUT_S=30
PYTHON_BIN="/usr/bin/python3"

cd "$APP_DIR"
# flask/requests live in the real user's site-packages; ensure python sees them
export HOME=/home/yi

log() { printf '[deploy] %s\n' "$*"; }
fail() {
    log "ERROR: $*"
    log "----- last 50 lines of $LOG_FILE -----"
    tail -n 50 "$LOG_FILE" 2>/dev/null || log "(no log file)"
    exit 1
}

# --- 1. Snapshot current PID so we can roll back if the new boot fails ---
OLD_PID=""
if [[ -f "$PID_FILE" ]]; then
    OLD_PID="$(cat "$PID_FILE" 2>/dev/null || true)"
fi

# --- 2. Kill old server if PID is alive ---
if [[ -n "$OLD_PID" ]] && kill -0 "$OLD_PID" 2>/dev/null; then
    log "stopping old server pid=$OLD_PID"
    kill "$OLD_PID" 2>/dev/null || true
    # wait up to 5s for graceful exit, then SIGKILL
    for _ in 1 2 3 4 5; do
        if ! kill -0 "$OLD_PID" 2>/dev/null; then break; fi
        sleep 1
    done
    if kill -0 "$OLD_PID" 2>/dev/null; then
        log "old pid $OLD_PID did not exit, sending SIGKILL"
        kill -9 "$OLD_PID" 2>/dev/null || true
    fi
fi

# --- 3. Start new server, record PID ---
log "starting new server: $PYTHON_BIN app.py"
nohup "$PYTHON_BIN" app.py > "$LOG_FILE" 2>&1 &
NEW_PID=$!
echo "$NEW_PID" > "$PID_FILE"
log "new pid=$NEW_PID"

# --- 4. Health check loop ---
healthz_ok=0
for ((i=1; i<=HEALTHZ_TIMEOUT_S; i++)); do
    # Is the process still alive?
    if ! kill -0 "$NEW_PID" 2>/dev/null; then
        fail "server process $NEW_PID died during startup"
    fi
    code="$(curl -sS -o /dev/null -w '%{http_code}' --max-time 2 "$HEALTHZ_URL" 2>/dev/null || echo '000')"
    if [[ "$code" == "200" ]]; then
        healthz_ok=1
        log "healthz OK after ${i}s"
        break
    fi
    sleep 1
done

# --- 5. Rollback on failure ---
if [[ "$healthz_ok" -ne 1 ]]; then
    log "health check failed after ${HEALTHZ_TIMEOUT_S}s, rolling back"
    kill "$NEW_PID" 2>/dev/null || true
    sleep 1
    kill -9 "$NEW_PID" 2>/dev/null || true
    rm -f "$PID_FILE"

    if [[ -n "$OLD_PID" ]]; then
        log "attempting to restart previous server (old pid=$OLD_PID)"
        nohup "$PYTHON_BIN" app.py > "$LOG_FILE" 2>&1 &
        ROLLBACK_PID=$!
        echo "$ROLLBACK_PID" > "$PID_FILE"
        log "rollback pid=$ROLLBACK_PID"
    fi
    fail "deploy failed; rolled back (see log above)"
fi

log "deploy succeeded pid=$NEW_PID"
curl -sS -o /dev/null -w 'local:%{http_code}\n' "$HEALTHZ_URL"