How to run a shell script on every request?

Question

I want to run a shell script every time my nginx server receives any HTTP request. Any simple ways to do this?

What is the actual problem you are trying to solve with this? As you have explained it this could easily cause more issues. — Burhan Khalid, Apr 06 '14 at 07:26
I am on an RPi and I want one of the LEDs to blink whenever my server gets a request. — Saswat Padhi, Apr 06 '14 at 07:27
What back end language / framework are you using? Most back end languages should allow you to call a shell script. — Chirag Bhatia - chirag64, Apr 06 '14 at 07:41
My question is whether I can modify it within nginx. I don't want to call the shell script manually from PHP or any other language which I may run on every request. — Saswat Padhi, Apr 06 '14 at 07:45

score 83 · Accepted Answer · edited Oct 30 '17 at 08:41

83

You can execute a shell script via Lua code from the nginx.conf file to achieve this. You need to have the HttpLuaModule to be able to do this.

Here's an example to do this.

location /my-website {
  content_by_lua_block {
    os.execute("/bin/myShellScript.sh")
  } 
}

edited Oct 30 '17 at 08:41

Arnaud P

12,022
7
56
67

answered Apr 06 '14 at 07:53

Chirag Bhatia - chirag64

4,430
3
26
35

2

This seems to produce an nginx error... `[alert] 6807#0: waitpid() failed (10: No child processes)` - see http://serverfault.com/q/432609/86531 – Nick Jul 03 '15 at 22:23
1

It seems that the standard error from `os.execute()` goes into the nginx error log, and any standard output is discarded. The `os.execute()` function returns the subprocess exit code, so this example would output an integer. Something based on `io.popen()` might work better for passing through the standard output from the subprocess. – Arto Bendiken Mar 30 '16 at 04:08
There is also https://github.com/juce/lua-resty-shell but that requires an external daemon. – Arto Bendiken Mar 30 '16 at 04:09
1

Usage of [content_by_lua](https://github.com/openresty/lua-nginx-module#content_by_lua) is marked as discouraged, so I updated your answer accordingly – Arnaud P Oct 30 '17 at 08:43
3

Can you get hold of POST or GET parameters from the request to pass the script? – Chris Dec 05 '17 at 21:46
Tried to install `openresty` to do so and nginx failed to start after that. Had to remove `openresty` to restart nginx – dmirg Aug 30 '23 at 16:11

Don Duvall · Answer 2 · 2014-07-11T06:00:51.177

I found the following information online at this address: https://www.ruby-forum.com/topic/2960191

This does expect that you have fcgiwrap installed on the machine. It is really as simple as:

sudo apt-get install fcgiwrap

Example script (Must be executable)

#!/bin/sh
# -*- coding: utf-8 -*-
NAME=`"cpuinfo"`
echo "Content-type:text/html\r\n"
echo "<html><head>"
echo "<title>$NAME</title>"
echo '<meta name="description" content="'$NAME'">'
echo '<meta name="keywords" content="'$NAME'">'
echo '<meta http-equiv="Content-type"
content="text/html;charset=UTF-8">'
echo '<meta name="ROBOTS" content="noindex">'
echo "</head><body><pre>"
date
echo "\nuname -a"
uname -a
echo "\ncpuinfo"
cat /proc/cpuinfo
echo "</pre></body></html>"

Also using this as an include file, not restricted to only shell scripts.

location ~ (\.cgi|\.py|\.sh|\.pl|\.lua)$ {
    gzip off;
    root /var/www/$server_name;
    autoindex on;
    fastcgi_pass unix:/var/run/fcgiwrap.socket;
    include /etc/nginx/fastcgi_params;
    fastcgi_param DOCUMENT_ROOT /var/www/$server_name;
    fastcgi_param SCRIPT_FILENAME /var/www/$server_name$fastcgi_script_name;
}

I found it extremely helpful for what I am working on, I hope it help you out with your RaspberryPI project.

See also https://www.nginx.com/resources/wiki/start/topics/examples/fcgiwrap/ and https://www.nginx.com/resources/wiki/start/topics/examples/fastcgiexample/ — Justin M. Keyes, Nov 11 '15 at 23:13
I needed to set the right permissions on the shell script `sudo chmod 755 /var/www/www.example.com/cpu.sh` (found here https://www.howtoforge.com/serving-cgi-scripts-with-nginx-on-centos-6.0-p2) — adriaan, Dec 23 '18 at 13:15
I found this article also quite helpful for this approach: https://sleeplessbeastie.eu/2017/09/18/how-to-execute-cgi-scripts-using-fcgiwrap/ — fei0x, Feb 19 '21 at 15:35

score 3 · Answer 3 · answered Jul 26 '22 at 14:05

If you prefer full control in Python:

Create /opt/httpbot.py:

#!/usr/bin/env python3
from http.server import HTTPServer, BaseHTTPRequestHandler
import subprocess

class Handler(BaseHTTPRequestHandler):
    def do_GET(self):
        self._handle()

    def do_POST(self):
        self._handle()

    def _handle(self):
        try:
            self.log_message("command: %s", self.path)
            if self.path == '/foo':
                subprocess.run(
                    "cd /opt/bar && GIT_SSH_COMMAND='ssh -i .ssh/id_rsa' git pull",
                    shell=True,
                )
        finally:
            self.send_response(200)
            self.send_header("content-type", "application/json")
            self.end_headers()
            self.wfile.write('{"ok": true}\r\n'.encode())

if __name__ == "__main__":
    HTTPServer(("127.0.0.1", 4242), Handler).serve_forever()

No concurrency/parallelism here, so httpbot runs one command at a time, no conflicts.
Run apt install supervisor
Create /etc/supervisor/conf.d/httpbot.conf:

[program:httpbot]
environment=PYTHONUNBUFFERED="TRUE"
directory=/opt
command=/opt/httpbot.py
autorestart=true
redirect_stderr=true
stdout_logfile=/var/log/httpbot.log
stdout_logfile_maxbytes=1MB
stdout_logfile_backups=10

Add to your nginx server:

    location /foo {
        proxy_pass http://127.0.0.1:4242/foo;
    }

Run:

chmod u+x /opt/httpbot.py

service supervisor status
# If stopped:
service supervisor start

supervisorctl status
# If httpbot is not running:
supervisorctl update

curl https://example.com/foo
# Should return {"ok": true}

tail /var/log/httpbot.log
# Should show `command: /foo` and the output of shell script

score 2 · Answer 4 · edited Aug 14 '18 at 11:53

Install OpenResty (OpenResty is just an enhanced version of Nginx by means of addon modules ) Refer https://openresty.org/en/getting-started.html for this
Configure aws cli on the instance
Write a shell script which download a file from specified S3 bucket
Do the required changes in nginx.conf file
Restart the nginx server

I have tested the http request using curl and file gets download in /tmp directory of respective instance:

curl -I http://localhost:8080/

OutPut:

curl -I http://localhost:8080/
HTTP/1.1 200 OK
Server: openresty/1.13.6.2
Date: Tue, 14 Aug 2018 07:34:49 GMT
Content-Type: text/plain
Connection: keep-alive

Content of nginx.conf file:

worker_processes  1;
error_log logs/error.log;
events {
    worker_connections 1024;
}
http {
    server {
        listen 8080;
        location / {
           default_type text/html;
           content_by_lua '
                ngx.say("<p>hello, world</p>")
           ';
        }

        location / {
            content_by_lua_block{
            os.execute("sh /tmp/s3.sh")
            }
        }

    }
}

I haven't seen any shell script output and also, it says my file /tmp/s3.sh doesn't exist even though i created that file and it runs when i execute it directly. — Prashanth, Apr 08 '19 at 03:43
My os.execute writes to stdout (i see it in nginx logs) but I got empty page in browser. — Dzenly, Mar 03 '21 at 18:10

score 1 · Answer 5 · answered May 11 '22 at 10:17

You can use nginx's perl module which is usually part of a repo and can be easily installed. Sample to call system curl command:

   location /mint {
       perl '
            sub {
               my $r = shift;
               $r->send_http_header("text/html");
               $r->print(`curl -X POST --data \'{"method":"evm_mine"}\' localhost:7545`);
               return OK;
            }
       '; 
}

score 0 · Answer 6 · answered Aug 23 '19 at 14:45

You can also use the nginx mirror module and poxy_pass it to a web script that runs whatever, in my case I just added this to my main site location {...

mirror /mirror;
mirror_request_body off;

and then a new location called mirror that I had run a php script that executed whatever...

location = /mirror {
    internal;
    proxy_pass http://localhost/run_script.php;
    proxy_pass_request_body off;
    proxy_set_header Content-Length "";
    proxy_set_header X-Original-URI $request_uri;
}

https://nginx.org/en/docs/http/ngx_http_mirror_module.html

How to run a shell script on every request?

6 Answers6

Linked