背景說明

最近監控到新安裝的es叢集的head服務經常掛掉，導致http://ip:9100 UI頁面打不開。
檢查了es叢集沒有問題，卻發現啟動head的npm grunt服務異常掛掉（es高版本後，預設使用npm的grunt服務啟動web服務），查詢了下日誌暫未發現任何問題，於是打算寫一個grunt服務異常故障重啟shell指令碼臨時過渡一下。

相關程式碼

es_head_self_restart.sh

#!/bin/sh
# Author: angelfish
# Desc: a daemon program for pull and restart the downed ES-HEAD plugin.

curr_time=$(date "+%Y-%m-%d %H:%M:%S")
start_command='/usr/local/elasticsearch-head/node_modules/grunt/bin/grunt server &'

while :

echo '>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> Current time is: '$curr_time

do
curr_time=$(date "+%Y-%m-%d %H:%M:%S")
echo ">> Current path is: " $PWD
runningStatus=$(ps -ef|grep "grunt"|grep -v "grep")
if [ "$runningStatus" ]; then
echo "Npm grunt server for es-head plugin was already started"
echo "here we do nothing"
echo '************************* $runningStatus : '$runningStatus
else
echo "Detected that grunt server was down, will attempt start it again."
#$start_command
/usr/local/elasticsearch-head/node_modules/grunt/bin/grunt server &
fi
sleep 30
done
 

grep + awk相關獲取任務程序號方法

[[email protected] elasticsearch-head]$ ps -ef|grep grunt|grep -v grep|awk -F ‘\s+’ ‘{print $0}’
es_depl+ 12954 12946 0 12:04 pts/0 00:00:01 grunt

[[email protected] elasticsearch-head]$ ps -ef|grep grunt|grep -v grep|awk -F ‘\s+’ ‘{print $1}’
es_depl+

[

[email protected] elasticsearch-head]$ ps -ef|grep grunt|grep -v grep|awk -F ‘\s+’ ‘{print $2}’
12954

一些想法

關於啟動方式：
1、可以以nohup方式啟動該daemon指令碼，如下：nohup es_head_self_restart.sh >> es_head_self_restart.sh.out &
2、crontab方式方式啟動定期檢查狀態，這時可將shell 腳本里的sleep 語句去掉；