phpydaemon

Webservice for calling long-running PHP functions as subprocesses.

Installation

Install from PyPI:

$ pip install phpy

Or install manually:

$ git clone https://github.com/stianstr/phpydaemon.git
$ cd phpydaemon
$ python setup.py install

Configuration

Create a config based on the defaults

$ cp /etc/phpydaemon.json-dist /etc/phpydaemon.json
$ vim /etc/phpydaemon.json

You need to point php.callback to the script you want to use to run PHP jobs

{
...
  "php": {
    "callback": "/path/to/your/app/phpydaemon-callback.php"
  }
...
}

You can find an example callback script in /usr/share/php/phpydaemon/callback-example.php

Configure workers (optional)

To get started, you can skip this part. But when you feel the need to control the number of paralell PHP processes to run for different jobs, you can configure the workers section in the config

{
  "workers": {
    // Allow two instances of myMethod() in class MyClass to run at the same time
    { "method": "MyClass.myMethod", "paralell": 2 },
    // Allow just one at a time for method someMethod() in class OtherClass in namespace \my\namespace
    { "method": "my.namespace.OtherClass.someMethod", "paralell": 1 },
    // Allow two paralell jobs for all other methods
    { "fallback": true, "paralell": 2 }
  }
}

The fallback item here is the default, so if you don't configure anything, phpydaemon will run two PHP calls in paralell at a time.

Usage

Start the daemon

$ phpydaemon start

Queueing jobs

Queue a PHP job to run (using CURL)

$ curl -X POST \
  -H 'Content-Type: application/json' \
  -d '{"method": "myapp.FileSystem.delete", "args": ["/"]}' \
  http://localhost:9713/api/queue

Queue a PHP job to run (using PHP)

require 'phpydaemon/Client.php';
use phpydaemon\Client;
$client = new Client();
$jobId = $client->queue('myapp.FileSystem.delete', array('/'));

The phpydaemon web server will then receive this job, queue it, and when a worker that matches the method is available, it will spawn a subprocess that calls you callback, asking it to do this:

$object = new \myapp\FileSystem();
$object->delete('/');

Getting statistics and job status

The daemon exposes a simple web UI that shows the different workers and running jobs.

google-chrome http://localhost:9713

You can also fetch stats via the webservice:

$ curl -X GET \
  -H 'Content-Type: application/json' \
  http://localhost:9713/api/stats

Or in PHP:

require 'phpydaemon/Client.php';
use phpydaemon\Client;
$client = new Client();
print_r($client->getStats());

Fetch details about running jobs:

$ curl -X GET \
  -H 'Content-Type: application/json' \
  http://localhost:9713/api/jobs

Or in PHP:

require 'phpydaemon/Client.php';
use phpydaemon\Client;
$client = new Client();
print_r($client->getJobs());

Fetch the stats and running jobs as an HTML page:

require 'phpydaemon/Client.php';
use phpydaemon\Client;
$client = new Client();
$html = $client->getStatusHtml();

Preserving userId

In many cases it may be useful to let the PHP background job run as the same user that queued it. You can pass user id as a parameter when queueing the job, and act on this in your PHP callback:

Queue the job with reference to a user id:

require 'phpydaemon/Client.php';
use phpydaemon\Client;
$client = new Client();
$jobId = $client->queue('myapp.FileSystem.delete', array('/'), $myApp->getUser()->getId());

Switch to that user before running the job in your callback:

require 'phpydaemon/Callback.php';
use phpydaemon\Callback;

$job = Callback::getJob();
MyApp::switchUser($job->userId);
Callback::runJob($job);

Todo

Process return data and status
Callback.php prepares return data and stdout as JSON but Python daemon does not use it for anything yet
Detect if PHP job is successfull or not, and add option to log/act upon job status
Clean up python code
Split up into one file for each class
Remove unused imports, code, commented out blocks

Credits

This tool was developed as part of my work at Availo AS and eonBIT as.