signal-batch/.gitea/workflows/deploy.yml
htlee 3b0c09575e fix(ci): 배포 health check 대기 90초→180초 확장 — 기동 타임아웃 실패 대응
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-19 05:17:23 +09:00

130 lines
4.4 KiB
YAML

name: Build & Deploy
on:
push:
branches: [main]
env:
DEPLOY_HOST: 192.168.1.18
DEPLOY_PORT: 32023
APP_DIR: /home/apps/signal-batch
APP_PORT: 18090
CONTEXT_PATH: /signal-batch
jobs:
build-and-deploy:
runs-on: ubuntu-latest
steps:
- name: Checkout
uses: actions/checkout@v4
- name: Install JDK 17 + Maven
run: |
apt-get update -qq
apt-get install -y -qq openjdk-17-jdk-headless maven openssh-client jq > /dev/null 2>&1
java -version
mvn --version
- name: Build with Maven
run: |
mvn -B clean package -DskipTests \
-Dmaven.compiler.release=17
ls -lh target/vessel-batch-aggregation.jar
- name: Deploy to production server
env:
DEPLOY_KEY: ${{ secrets.DEPLOY_SSH_KEY }}
run: |
# SSH 키 설정
mkdir -p ~/.ssh
echo "$DEPLOY_KEY" > ~/.ssh/id_deploy
chmod 600 ~/.ssh/id_deploy
ssh-keyscan -p $DEPLOY_PORT $DEPLOY_HOST >> ~/.ssh/known_hosts 2>/dev/null || true
SSH_CMD="ssh -p $DEPLOY_PORT -i ~/.ssh/id_deploy -o StrictHostKeyChecking=no root@$DEPLOY_HOST"
SCP_CMD="scp -P $DEPLOY_PORT -i ~/.ssh/id_deploy -o StrictHostKeyChecking=no"
# JAR 전송
echo "=== Uploading JAR ==="
$SCP_CMD target/vessel-batch-aggregation.jar root@$DEPLOY_HOST:$APP_DIR/vessel-batch-aggregation.jar.new
# 안전 배포: 활성 Job 확인 → 종료 → JAR 교체 → 기동
echo "=== Safe Deploy ==="
$SSH_CMD bash -s << 'DEPLOY'
set -e
APP_DIR=/home/apps/signal-batch
BASE_URL="http://localhost:18090/signal-batch"
SAFE_COUNT=0
MAX_WAIT=300
# 1단계: 활성 Job 없음 3초 연속 확인
echo "--- Waiting for running jobs to complete ---"
for i in $(seq 1 $MAX_WAIT); do
RESPONSE=$(curl -sf "$BASE_URL/admin/batch/job/running" 2>/dev/null || echo "UNAVAILABLE")
if [ "$RESPONSE" = "UNAVAILABLE" ]; then
echo "App not responding — proceeding with deploy"
SAFE_COUNT=3
break
fi
RUNNING=$(echo "$RESPONSE" | jq 'length' 2>/dev/null || echo "0")
if [ "$RUNNING" = "0" ]; then
SAFE_COUNT=$((SAFE_COUNT + 1))
echo "No running jobs ($SAFE_COUNT/3)"
if [ $SAFE_COUNT -ge 3 ]; then
echo "Safe to deploy!"
break
fi
else
[ $SAFE_COUNT -gt 0 ] && echo "Active job detected — resetting count"
SAFE_COUNT=0
fi
sleep 1
done
if [ $SAFE_COUNT -lt 3 ]; then
echo "WARNING: Timeout waiting for jobs. Proceeding anyway."
fi
# 2단계: 서비스 중지 (graceful shutdown)
echo "--- Stopping service ---"
systemctl stop signal-batch || true
# systemd TimeoutStopSec=180 에서 graceful shutdown 완료 대기
# 3단계: JAR 교체
echo "--- Replacing JAR ---"
mkdir -p $APP_DIR/backup
if [ -f $APP_DIR/vessel-batch-aggregation.jar ]; then
cp $APP_DIR/vessel-batch-aggregation.jar $APP_DIR/backup/vessel-batch-aggregation.jar.$(date +%Y%m%d-%H%M%S)
fi
mv $APP_DIR/vessel-batch-aggregation.jar.new $APP_DIR/vessel-batch-aggregation.jar
restorecon -v $APP_DIR/vessel-batch-aggregation.jar 2>/dev/null || true
# 백업 정리 (최근 5개만 유지)
ls -t $APP_DIR/backup/vessel-batch-aggregation.jar.* 2>/dev/null | tail -n +6 | xargs rm -f 2>/dev/null || true
# 4단계: 서비스 기동
echo "--- Starting service ---"
systemctl start signal-batch
# 5단계: 기동 확인 (최대 180초 — 64GB 힙 AlwaysPreTouch + 캐시 워밍업)
for i in $(seq 1 180); do
if curl -sf "$BASE_URL/actuator/health/liveness" > /dev/null 2>&1; then
echo "Service started successfully (${i}s)"
curl -s "$BASE_URL/actuator/health"
exit 0
fi
sleep 1
done
echo "WARNING: Startup timeout. Recent logs:"
journalctl -u signal-batch --no-pager -n 50
exit 1
DEPLOY
- name: Cleanup
if: always()
run: rm -f ~/.ssh/id_deploy