核心服務(wù)器上跑了一堆的腳本、程序，難免有時(shí)候會(huì)出現(xiàn)僵尸進(jìn)程，死不死活不活的在那里占用資源，最初只是寫了個(gè)根據(jù)關(guān)鍵字查殺進(jìn)程的linux shell腳本，后來發(fā)現(xiàn)很多時(shí)候進(jìn)程死在那里的時(shí)候其實(shí)是內(nèi)部調(diào)用子進(jìn)程的時(shí)候出現(xiàn)了問題，這時(shí)候光殺父進(jìn)程根本沒解決根本問題。比如說rsync的時(shí)候通過ssh來連接，rsync本身沒問題，但可能ssh死掉了。因此重新寫了腳本，遞歸查找子進(jìn)程。

代碼如下:

#!/bin/sh

# 遞歸找到導(dǎo)致進(jìn)程僵死的最底層子進(jìn)程并殺除.

ParentProcessID=$1;

if [ "x${ParentProcessID}" = "x" ] ; then

echo "Please Supply the top Parent Process ID to be killed!"

echo "Usage:sh $0 PID [-v]"

echo "PID The Parent Process ID as root"

echo "-v is this argument supplied,no real kill operation will be performed,only process tree be show."

exit 1

fi

let IsRealKillDo=1;

if [ "x$2" = "x-v" ] ; then

let IsRealKillDo=0;

fi

echo "Begin Kill the Leaf Process of process ${ParentProcessID}" >&2

killpidList=""

function loopNextSubProcess(){

local nParentProcessID=$1

local tmpPidList=""

tmpPidList=`ps -A --format='%p%PisParent' --width 2048 -w --sort pid|grep "${nParentProcessID}isParent"|grep -v grep|grep -v "$$" | awk '{ printf $1 }'`

ps --format='%p%P%a' --width 2048 -w -p ${nParentProcessID}|grep -v grep|grep -v "$$" >&2

if [ "x${tmpPidList}" = "x" ] ; then

echo "****Got One Leaf = [${nParentProcessID}]****" >&2

killpidList="${killpidList}\n${nParentProcessID}"

return

fi

for theNextPid in ${tmpPidList} ; do

loopNextSubProcess ${theNextPid}

done

}

loopNextSubProcess ${ParentProcessID}

if [ ${IsRealKillDo} -eq 1 -a "x${killpidList}" != "x" ] ; then

for curpid in `echo -e ${killpidList}` ; do

if [ "x${curpid}" != "x" ] ; then

echo "kill -9 ${curpid}"

kill -9 ${curpid}

fi

done

else

echo -e ${killpidList}

fi

更多信息請(qǐng)查看IT技術(shù)專欄