LinuxQuestions.org - shell script for httpd

- Linux - Newbie (https://www.linuxquestions.org/questions/linux-newbie-8/)

- - shell script for httpd (https://www.linuxquestions.org/questions/linux-newbie-8/shell-script-for-httpd-709412/)

shell script for httpd

Hi, I make this script but I'm not sure if this is correct.

Code:

#!/bin/bash



set -e



SERVICE="httpd"



while true; do

if ps -ef | grep -q $SERVICE; then

 lsof | grep IPv | grep 80 | awk '{print $2}' | xargs -l -i kill -9 {}  

 /usr/sbin/$SERVICE -k restart

fi

 sleep 2

done

I MAKE THIS SCRIPT FOR RESTART MY APACHE DAEMON IF THIS IS DOWN BUT ELSE FOR KILL ALL THE PROCESS THAT USE THE 80 PORT.

I HOPE YOUR HELP. Regards

I USE DEBIAN 4.*

try replacing
/usr/sbin/$SERVICE -k restart
with
/etc/init.d/$SERVICE restart

This script runs forever - doesn't it make more sense to create a simpler script that is run as a cron job? Then you will have much better control about the timing intervals. Your script won't eat a lot of system resources, but it's just bad practice to use endless loops with sleep calls.

----------
T74marcell

Arch Linux

I don't user cron jobs but is a good idea. I have cron jobs of directadmin control panel so this can be work?

Quote:

Originally Posted by urgido

Code:

#!/bin/bash



set -e



SERVICE="httpd"



while true; do

if ps -ef | grep -q $SERVICE; then

 lsof | grep IPv | grep 80 | awk '{print $2}' | xargs -l -i kill -9 {}  

 /usr/sbin/$SERVICE -k restart

fi

 sleep 2

done

I MAKE THIS SCRIPT FOR RESTART MY APACHE DAEMON IF THIS IS DOWN BUT ELSE FOR KILL ALL THE PROCESS THAT USE THE 80 PORT.

This is insane. You're restarting httpd every 2 seconds.

On top of that, your non-precise grep 80 statement could match anything that has, e.g. "80" in the PID. So you'd violently kill something you did not intend to kill, and then start only httpd.

All your script really needs to do is this:

Code:

# pgrep httpd >/dev/null || /etc/init.d/httpd start

(or whatever Debian uses to start services)

I'd also add that this is a poor approach to service management, IMO. If httpd is crashing regularly, diagnose and solve the problem. You should also deploy some form of service monitoring - e.g. Nagios.

Quote:

Originally Posted by anomie (Post 3467340)

All your script really needs to do is this:

Code:

# pgrep httpd >/dev/null || /etc/init.d/httpd start

(or whatever Debian uses to start services)

I love this post. An advanced scripting problem resolved with beautiful simplicity. :)

so my script is wrong and the correct shell script is just: # pgrep httpd >/dev/null || /etc/init.d/httpd start ???

Your script is causing a lot of unnecessary activity (and probably service interruptions) for the httpd daemon. A closely related question is: why is this script you're proposing necessary? Is httpd regularly crashing?

The one-liner I suggested could run as a cronjob every 5 minutes or so.

Is httpd regularly crashing?

YES =(

I'm newbie on Linux. Frequently it's necessary to kill the process of httpd and up again.

LOOK THIS:

IN THIS MOMENT MY HTTPD SERVICE IS CRASHING SO I PROCCED TO CHECK THE ISSUE:

box:~# lsof -w -n -i tcp:80
COMMAND PID USER FD TYPE DEVICE SIZE NODE NAME
h 8858 apache 3u IPv6 1684609 TCP *:www (LISTEN)

wtf? h command?

So if I try to run my httpd service the system print an error like this: You can't use 80 port because is in use.

So for solve the problem I use the follow commands:

box:~# kill -9 8858
box:~# /usr/sbin/httpd -k start
box:~# lsof -w -n -i tcp:80
COMMAND PID USER FD TYPE DEVICE SIZE NODE NAME
httpd 21138 root 3u IPv6 1761832 TCP *:www (LISTEN)
httpd 21139 apache 3u IPv6 1761832 TCP *:www (LISTEN)
httpd 21140 apache 3u IPv6 1761832 TCP *:www (LISTEN)
httpd 21141 apache 3u IPv6 1761832 TCP *:www (LISTEN)
httpd 21142 apache 3u IPv6 1761832 TCP *:www (LISTEN)
httpd 21143 apache 3u IPv6 1761832 TCP *:www (LISTEN)
httpd 21144 apache 3u IPv6 1761832 TCP *:www (LISTEN)
httpd 21145 apache 3u IPv6 1761832 TCP *:www (LISTEN)
httpd 21146 apache 3u IPv6 1761832 TCP *:www (LISTEN)
httpd 21147 apache 3u IPv6 1761832 TCP *:www (LISTEN)
httpd 21148 apache 3u IPv6 1761832 TCP *:www (LISTEN)
httpd 21149 apache 3u IPv6 1761832 TCP *:www (LISTEN)
httpd 21150 apache 3u IPv6 1761832 TCP *:www (LISTEN)
box:~#

AND ALL IS OK! =/ ... But this happeds frecuently so I need a shell script to do this. Regards

SORRY IF MY ENGLISH IS BAD.

Like Anomie already said: if your httpd is crashing regularly combatting symptoms won't make the problem go away. You need to diagnose and solve the problem instead. Do your system and http daemon access and error logs show warnings or errors?

error_log

[Sat Mar 07 00:10:15 2009] [notice] Apache/2.2.11 (Unix) mod_ssl/2.2.11 OpenSSL/0.9.8c DAV/2 PHP/5.2.9 configured -- resuming normal operations

[Sat Mar 07 00:11:02 2009] [notice] caught SIGTERM, shutting down

Quote:

Originally Posted by urgido

[Sat Mar 07 00:11:02 2009] [notice] caught SIGTERM, shutting down

It looks like a program (maybe another one of your shell scripts?) may be sending a kill signal to httpd.

What I'd suggest is editing httpd.conf so that you have the directive:

Code:

LogLevel info

Then reload or restart httpd. Finally, wait until httpd crashes again, and then post the last 30 or so lines of error_log here.

---

Also, are you the only sysadmin on this server? Is anyone else root or a sudoer besides you?

Quote:

Originally Posted by urgido

IN THIS MOMENT MY HTTPD SERVICE IS CRASHING SO I PROCCED TO CHECK THE ISSUE:

box:~# lsof -w -n -i tcp:80
COMMAND PID USER FD TYPE DEVICE SIZE NODE NAME
h 8858 apache 3u IPv6 1684609 TCP *:www (LISTEN)

wtf? h command?

That's definitely odd. Next time you see this h (?) daemon, also post the results of:
# lsof -p <pid_here>

(... in addition to the other items above.)