Docker容器服务端口探测

1.业务场景

业务需要，所有服务使用docker部署，在服务器重启后，systemctl管理的docker启动容器是并发的，业务服务启动有时候比中间件启动快，由于业务服务第一次启动时只连接中间一次，如果连接失败，就不会再重试了

2.修复方案

（1）业务服务增加重试连接次数（研发修改java代码）

（2）业务容器启动脚本增加对中间件服务的探测，探测后再启动，给中间件启动时间

• 下载wait-for-it.sh脚本

https://github.com/vishnubob/wait-for-it/archive/refs/heads/master.zip
unzip wait-for-it-master.zip
chmod +x wait-for-it.sh

• dockerfile文件

FROM jre-centos7:8u251
MAINTAINER xuxiaoming@zhengjue-ai.com

COPY entrypoint.sh *.jar /device-message/
WORKDIR /device-message/
HEALTHCHECK --interval=10s --timeout=30s --retries=3 CMD \
    curl --fail -sL -w "http_code:%{http_code} \n" \
    -o /dev/null http://localhost:8070/info || exit 1
RUN chmod +x /device-message/entrypoint.sh
ENTRYPOINT ["sh","/device-message/entrypoint.sh"]

•  entrypoint.sh脚本

#!/bin/bash

WAIT_FOR_IT="/device-message/wait-for-it.sh"
TIMEOUT=120

SERVICES=(
  "127.0.0.1:3306"
  "127.0.0.1:6379"
  "127.0.0.1:8848"
  "127.0.0.1:9001"
  "127.0.0.1:9200"
)

log_with_timestamp() {
  echo "$(date +'%Y-%m-%d %H:%M:%S,%3N') $1"
}

if [ -x "$WAIT_FOR_IT" ]; then
  log_with_timestamp "Starting dependency checks"
  
  for target in "${SERVICES[@]}"; do
    "$WAIT_FOR_IT" "$target" -t $TIMEOUT
    if [ $? -ne 0 ]; then
      log_with_timestamp "ERROR: Service $target not available within $TIMEOUT seconds" >&2
      exit 1
    fi
  done
  
  log_with_timestamp "All dependencies are ready - starting application"
fi

log_with_timestamp "Launching Java application"

exec java -Xmx2048m -Xms2048m -jar -XX:+PrintGCDateStamps -XX:+PrintGCDetails -Xloggc:/device-message/logs/gc-%t.log -Dfile.encoding=utf-8 -Duser.timezone=GMT+8 ./device-message-*.jar --spring.config.location=/device-message/config/

•  业务服务在启动时，会探测中间件服务的端口，超时为120s，可自定义
• docker logs -f device-message --tail 200

2025-07-28 12:28:23,052 Starting dependency checks
wait-for-it.sh: waiting 120 seconds for 127.0.0.1:23306
wait-for-it.sh: 127.0.0.1:23306 is available after 0 seconds
wait-for-it.sh: waiting 120 seconds for 127.0.0.1:26379
wait-for-it.sh: 127.0.0.1:26379 is available after 0 seconds
wait-for-it.sh: waiting 120 seconds for 127.0.0.1:28848
wait-for-it.sh: 127.0.0.1:28848 is available after 69 seconds
wait-for-it.sh: waiting 120 seconds for 127.0.0.1:29001
wait-for-it.sh: 127.0.0.1:29001 is available after 0 seconds
wait-for-it.sh: waiting 120 seconds for 127.0.0.1:29200
wait-for-it.sh: 127.0.0.1:29200 is available after 0 seconds
2025-07-28 12:29:32,354 All dependencies are ready - starting application
2025-07-28 12:29:32,356 Launching Java application