自动发布脚本导致的Tomcat意外退出

最近公司项目的Dev环境要自己维护，为了避免麻烦，写了一个发布脚本，但是发现每次关闭和服务器的SSH连接后，Tomcat就意外退出，查看日志发现最后退出有如下几行日志：

Nov 10, 2017 2:38:35 PM org.apache.coyote.AbstractProtocol pause
INFO: Pausing ProtocolHandler ["http-nio-8080"]

Nov 10, 2017 2:38:35 PM org.apache.catalina.core.StandardService stopInternal
INFO: Stopping service Catalina

[INFO][2017-11-10 14:38:35] org.springframework.context.support.AbstractApplicationContext.doClose(AbstractApplicationContext.java:960) Closing WebApplicationContext for namespace 'Dispatcher-servlet': startup date [Fri Nov 10 14:37:53 CST 2017]; parent: Root WebApplicationContext
[INFO][2017-11-10 14:38:35] org.springframework.context.support.AbstractApplicationContext.doClose(AbstractApplicationContext.java:960) Closing Root WebApplicationContext: startup date [Fri Nov 10 14:37:52 CST 2017]; root of context hierarchy
Nov 10, 2017 2:38:35 PM org.apache.catalina.loader.WebappClassLoader clearReferencesThreads
SEVERE: The web application [/] appears to have started a thread named [FileWatchdog] but has failed to stop it. This is very likely to create a memory leak.
Nov 10, 2017 2:38:35 PM org.apache.coyote.AbstractProtocol stop
INFO: Stopping ProtocolHandler ["http-nio-8080"]
Nov 10, 2017 2:38:35 PM org.apache.coyote.AbstractProtocol destroy
INFO: Destroying ProtocolHandler ["http-nio-8080"]

最开始怀疑是 OOM 问题，排查代码没有发现相关问题，JVM配置OOM后转存堆内存快照，也没有发现问题，后来发现规律是每次关闭SSH连接一会儿后Tomcat即关闭，度娘后发现如下文章：

tomcat进程意外退出的问题分析

-XX:+HeapDumpOnOutOfMemoryError  -XX:HeapDumpPath=\data\heap\

解决方案

在脚本开始添加 set -m，使用监视模式。

问题原因

SSH关闭后，操作系统kill掉脚本；

Java进程挂在了脚本进程下面；

修改后的脚本

修改后的发布脚本：

#!/bin/sh
# author qiesai

##调用格式：base-war-deploy-tools.sh proName branchName profile
##命名规则：proName作为back、release、war的名称存在，要保持一致
PRO_NAME=$1
BRANCH_NAME=$2
PROFILE=$3

BASE_DIR=`pwd`
CODE_DIR="/data/code/${PRO_NAME}/"
RELEASE_DIR="/data/release/"
BACKUP_DIR="/data/backup/"
WAR_NAME="${PRO_NAME}.war";

function stopTomcat(){
echo "_______________________________"
echo "stop tomcat"

TOMCAT_PID=`ps -ef|grep tomcat | grep -v 'grep\|tail' | awk '{print $2}'`
if [ -n "$TOMCAT_PID" ]; then
echo "tomcat is running : ${TOMCAT_PID} , will stop it"
kill -9 $TOMCAT_PID
fi

COUNT=0
while [ $COUNT -lt 1 ]; do
echo -e ".\c"
sleep 1
COUNT=1

PID_EXIST=`ps --no-heading -p $TOMCAT_PID`
if [ -n "$PID_EXIST" ]; then
COUNT=0
break
fi
done
echo ""
echo "stop tomcat succes"
}

function buildProject(){
source /etc/profile
echo "_______________________________"
echo "pull git code"
cd $CODE_DIR
git branch -q $BRANCH_NAME
git checkout $BRANCH_NAME
git pull origin $BRANCH_NAME
echo "_______________________________"
echo "buil project"
echo `/data/maven/maven/bin/mvn clean package -P${PROFILE}`
echo "buil finish"
}

function releaseWar(){
echo "_______________________________"
echo "deploy war"
cd $RELEASE_DIR
mv -f ./$WAR_NAME $BACKUP_DIR$WAR_NAME
#注意不要 rm -rf /*也不要 cd到/下
rm -rf *
mv -f $CODE_DIR/target/$WAR_NAME ./$WAR_NAME
#set -m是解决Tomcat关闭的关键点。
set -m
nohup /data/tomcat/bin/startup.sh &
echo "deploy success"
}

function main(){
if [ -z "$PRO_NAME" ]; then
echo "please input project name"
exit;
fi

if [ -z "$BRANCH_NAME" ]; then
echo "please input a git branch name"
exit;
fi

if [ -z "$PROFILE" ]; then
echo "please choose a profile"
exit;
fi

echo "=====================begin deploy====================="
buildProject;
stopTomcat;
releaseWar;
echo "=====================deploy succ====================="
tail -f /data/tomcat/logs/catalina.out
}

main;