node.js - How can I keep my php script running as a daemon?

I wrote a server using swoole. I plan to hang it on the server as a daemon and run it continuously. This will also allow the script to restart as soon as possible after unexpectedly exiting due to an error. How can I do this?

Reply content:

I wrote a server using swoole. I plan to hang it on the server as a daemon and run it continuously. This will also allow the script to restart as soon as possible after unexpectedly exiting due to an error. How can I do this?

The linux nohup command runs this php script and then writes a shell to monitor the running status of this script. If the script dies, restart it

Just write the script as a loop, such as while(1), so that it can run forever.

In this script header, this judgment, if there is already a process of this script, exit directly, and execute if there is no process,

Write another linix scheduled task, such as executing this php script every minute.

After executing the script once, the script will keep running. The scheduled task will execute the script every minute. If the script is alive, it will exit directly. If it dies, it will start execution.

Write the php script. It is recommended to check the memory usage regularly and not write the core logic. This is related to business.

<code>if(memory_get_usage()>100*1024*1024){
    exit(0);//大于100M内存退出程序，防止内存泄漏被系统杀死导致任务终端
}</code>

Assume the path of the php file is /root/run.php
Open the terminal

<code>setsid php /root/run.php > /dev/null &</code>

Edit the process monitoring script to automatically restart the process when it does not exist /root/monitor.sh

<code>#!/bin/bash
alive=`ps aux|grep root\/run|grep -v grep|wc -l`
if [ $alive -eq 0]
then
php /root/run.php > /dev/null &
fi</code>

Add scheduled tasks (detected every minute)

<code>crontab -e
* * * * * /root/monitor.sh > /dev/null &</code>

Handmade

Use a simple and reliable Shell script to protect a PHP service that may exit suddenly

Add the boot command to /etc/rc.local:

<code>nohup /path/to/swoole.sh >>/path/to/swoole.sh.log 2>&1 &</code>

swoole.sh content is as follows:

<code>#!/bin/sh
PREFIX=/home/eechen
INTERVAL=1
nohup php ${PREFIX}/swoole.php >>${PREFIX}/swoole.log 2>&1 & echo $! > ${PREFIX}/swoole.pid
while [ 1 ]; do
    if [ ! -d /proc/`cat ${PREFIX}/swoole.pid` ]; then
        nohup php ${PREFIX}/swoole.php >>${PREFIX}/swoole.log 2>&1 & echo $! > ${PREFIX}/swoole.pid
        echo 'NEW_PID:'`cat ${PREFIX}/swoole.pid && date '+%Y-%m-%d %H:%M:%S'`
    fi
    sleep ${INTERVAL}
done</code>

Among them:

<code>nohup 表示忽略SIGHUP信号(编号1),比如退出终端时发送的SIGHUP信号(kill -SIGHUP PID)会被忽略掉.
>>${PREFIX}/swoole.log 表示把标准输出重定向(>>表示追加,>表示覆盖)到文件swoole.log
2>&1 表示将标准错误(2:stderr)重定向到标准输出(1:stdout).
结尾加上&表示将命令放入后台运行.
$! 表示前面运行在后台的PHP进程PID.</code>

That is, swoole.sh checks whether the directory /proc/PID exists every 1 second. If it does not exist, the service is restarted.
swoole.sh.log records the time when the service restarts.
swoole.log records It is the output of the service itself.

For example, use Shell to protect vmstat:

<code>vmstat.sh
#!/bin/sh
PREFIX=/home/eechen
INTERVAL=1
nohup vmstat 1 >>${PREFIX}/vmstat.log 2>&1 & echo $! > ${PREFIX}/vmstat.pid
while [ 1 ]; do
    if [ ! -d /proc/`cat ${PREFIX}/vmstat.pid` ]; then
        nohup vmstat 1 >>${PREFIX}/vmstat.log 2>&1 & echo $! > ${PREFIX}/vmstat.pid
        echo 'NEW_PID:'`cat ${PREFIX}/vmstat.pid && date '+%Y-%m-%d %H:%M:%S'`
    fi
    sleep ${INTERVAL}
done

#运行
nohup /home/eechen/vmstat.sh >>/home/eechen/vmstat.sh.log 2>&1 &
#杀死,可以看到vmstat被重启
kill `cat /home/eechen/vmstat.pid`</code>

This vmstat.sh script can also be implemented in PHP:

<code>nohup php /home/eechen/vmstat.php >>/home/eechen/vmstat.php.log 2>&1 &
<?php
$prefix = '/home/eechen';
$interval = 1;
shell_exec("nohup vmstat 1 >>$prefix/vmstat.log 2>&1 & echo $! > $prefix/vmstat.pid");
while ( 1 ) {
    if ( !file_exists('/proc/'.trim(file_get_contents("$prefix/vmstat.pid"))) ) {
        shell_exec("nohup vmstat 1 >>$prefix/vmstat.log 2>&1 & echo $! > $prefix/vmstat.pid");
        echo 'NEW_PID:'.trim(file_get_contents("$prefix/vmstat.pid")).' '.date('Y-m-d H:i:s');
    }
    sleep($interval);
}</code>

1.crontab must be there, check if it exists every minute, (ps -ef | grep..), if it exits, if not, create and execute it
2. A process that has been running can loop endlessly...
3.crontab The script can detect the status of your always-running process. If there is a problem, it has been waiting, no response or other problems, you have to kill and restart it yourself

Linux shell execution