转载 kestrel php 消息队列

wtuyss

　　We've been using Twitter's kestrel queue server for a while now at work, but only from our service layer, which is written in python.  Now that we have some queueing needs from our application layer, written in PHP, I spent a few days this week adding queue support to our web application.  I thought I'd share what I learned, and how I implemented it.
　　Goals
　　The kestrel server itself was pretty straightforward to get up and running.  The only thing I would point out is that I recommend sticking to release branches, as master was fairly unstable when I tried to use it.  Regarding implementing the client, there were a few goals I had in mind when I started:
　　Since kestrel is built on the memcache protocol, try and leverage an existing memcache client rather than build one from scratch
    Utilize our existing batch job infrastructure, which I covered previously here, and make sure our multi-tenant needs are met
    Keep the queue interface generic in case we change queue servers later
    Utilize existing kestrel management tools, only build out the the functionality we need
　　With these goals in mind, I ended up with 4 components: a kestrel client, a producer, a consumer, and a very small CLI harness for running the consumer.  But before I even coded anything, I set up kestrel web, a web UI for kestrel written by my co-worker Matt Erkkila.  Kestrel web allows you to view statistics on kestrel, manage queues, as well as sort and filter queues based on manual inputs.  Having this tool up and running from the get go made it easy to watch jobs get added and consumed from my test queue, and also easily flush out the queues as needed.
　　The Kestrel Client
　　I couldn't find any existing kestrel clients for PHP, so I started looking at the two memcache extensions: the older memcache, and Andrei Zmievski's memcached, the latter of which is based on the libmemcached library.  I started with memcache, and while it worked fine initially, I quickly found that I could not modify timeouts.  This interfered with the way  kestrel recommends you poll it for new jobs, and I would see timeout errors from the memcache extension if you tried to set the poll timeout to 1 second or higher (the memcache default).  The memcached extension does not have these issues, so I went with it.
　　The first gotcha I ran into was serialization.  You can use memcached's serializer for writing to kestrel, but when it reads the data back, it doesn't recognize that it is serialized.  So I just serialize the data manually in my client, and things work fine. One other thing to note is that you'll want to disable compression, or do it manually, as the memcached extension will automatically compress anything over 100 bytes by default, and will not decompress it when reading from kestrel.
　　The other issue is that if you want to use any custom kestrel commands, you can't.  Since the application layer doesn't need anything fancy, the memcached extension will work fine for it.  Once we need support for the upcoming monitor (batching) in kestrel 2, we may need to implement a kestrel client from scratch.  Kestrel web supplies everything else we need right now.
　　Once the decision was made to use memcached, I wrote a light decorator for it, EC_KestrelClient.  This handles instantiation of the memcached client, serialization, and helpers for some kestrel specific options to the GET command.  It also has support for passing memcached specific options through it.  The class ended up looking like this:

<?php
/**
* A thin kestrel client that wraps Memcached (libmemcached extension)
*
* @author Bill Shupp <hostmaster@shupp.org>
* @copyright 2010-2011 Empower Campaigns
*/
class EC_KestrelClient
{
/**
* The Memcached instance
*
* @var Memcached
*/
protected $_memcached = null;
/**
* The Kestrel server IP
*
* @var string
*/
protected $_host = '127.0.0.1';
/**
* The Kestrel server port
*
* @var string
*/
protected $_port = 22133;
/**
* Optional options, not currently used
*
* @var array
*/
protected $_options = array();
/**
* Sets the host, port, and options to be used
*
* @param string $host The host to use, defaults to 127.0.0.1
* @param int $port The port to use, defaults to 22133
* @param array $options Memcached options, not currently used
*
* @return void
*/
public function __construct(
$host = '127.0.0.1', $port = 22133, array $options = array()
)
{
$this->_host = $host;
$this->_port = $port;
$this->setOptions($options);
}

/**
* Sets job data on the queue, json_encoding the value to avoid problematic
* serialization.
*
* @param string $queue The queue name
* @param mixed $data The data to store
*
* @return bool
*/
public function set($queue, $data)
{
// Local json serialization, as kestrel doesn't send serialization flags
return $this->getMemcached()->set($queue, json_encode($data));
}
/**
* Reliably read an item off of the queue. Meant to be run in a loop, and
* call closeReliableRead() when done to make sure the final job is not left
* on the queue.
*
* @param mixed $queue The queue name to read from
* @param int $timeout The timeout to wait for a job to appear
*
* @return array|false
* @see closeReliableRead()
*/
public function reliableRead($queue, $timeout = 1000)
{
$queue = $queue . '/close/open/t=' . $timeout;
$result = $this->getMemcached()->get($queue);
if ($result === false) {
return $result;
}
// Local json serialization, as kestrel doesn't send serialization flags
return json_decode($result, true);
}
/**
* Closes any existing open read
*
* @param string $queue The queue name
*
* @return false
*/
public function closeReliableRead($queue)
{
$queue = $queue . '/close';
return $this->getMemcached()->get($queue);
}
/**
* Aborts an existing reliable read
*
* @param string $queue The queue name
*
* @return false
*/
public function abortReliableRead($queue)
{
$queue = $queue . '/abort';
return $this->getMemcached()->get($queue);
}
/**
* Set an option to be used with the Memcached client. Not used.
*
* @param string $name The option name
* @param value $value The option value
*
* @return void
*/
public function setOption($name, $value)
{
$this->_options[$name] = $value;
}
/**
* Sets multiple options
*
* @param array $options Array of key/values to set
*
* @return void
*/
public function setOptions(array $options)
{
foreach ($options as $name => $value) {
$this->setOption($name, $value);
}
}
/**
* Gets a current option's value
*
* @param string $name The option name
*
* @return mixed
*/
public function getOption($name)
{
if (isset($this->_options[$name])) {
return $this->_options[$name];
}
return null;
}
/**
* Gets all current options
*
* @return array
*/
public function getOptions()
{
return $this->_options;
}
/**
* Gets a singleton instance of the Memcached client
*
* @return Memcached
*/
public function getMemcached()
{
if ($this->_memcached === null) {
$this->_initMemcached();
}
return $this->_memcached;
}
/**
* Initialized the Memcached client instance
*
* @return void
*/
protected function _initMemcached()
{
$this->_memcached = $this->_getMemcachedInstance();
foreach ($this->_options as $option => $value) {
$this->_memcached->setOption($option, $value);
}
$this->_memcached->addServer($this->_host, $this->_port);
$this->_memcached->setOption(Memcached::OPT_COMPRESSION, false);
}
// @codeCoverageIgnoreStart
/**
* Returns a new instance of Memcached. Abstracted for testing.
*
* @return Memcached
*/
protected function _getMemcachedInstance()
{
return new Memcached();
}
// @codeCoverageIgnoreEnd
}
　　view raw EC_KestrelClient.php This Gist brought to you by GitHub.
　　The Producer
　　The producer is very simple.  It just formats the data into a standard structure, including current tenant information, namespaces the queue so it doesn't collide with other projects, and adds it to the queue.  The producer looks like this:

<?php
/**
* Interface for adding jobs to a queue server
*
* @author Bill Shupp <hostmaster@shupp.org>
* @copyright 2010-2011 Empower Campaigns
*/
class EC_Producer
{
/**
* Adds a job onto a queue
*
* @param string $queue The queue name to add a job to
* @param string $jobName The job name for the consumer to run
* @param mixed $data Optional additional data to pass to the job
*
* @return bool
*/
public function addJob($queue, $jobName, $data = null)
{
$item = array(
'instance' => EC::getCurrentInstanceName(),
'jobName' => $jobName
);
if ($data !== null) {
$item['data'] = $data;
}
// Namespace queue with project
$queue = 'enterprise_' . $queue;
$client = $this->_getKestrelClient();
return $client->set($queue, $item);
}
// @codeCoverageIgnoreStart
/**
* Gets a single instance of EC_KestrelClient. Abstracted for testing.
*
* @return void
*/
protected function _getKestrelClient()
{
if (APPLICATION_ENV === 'testing') {
throw new Exception(__METHOD__ . ' was not mocked when testing');
}
static $client = null;
if ($client === null) {
$host = EC::getConfigOption('kestrel.host');
$port = EC::getConfigOption('kestrel.port');
$client = new EC_KestrelClient($host, $port);
}
return $client;
}
// @codeCoverageIgnoreEnd
}

　　view raw EC_Producer.php This Gist brought to you by GitHub.
　　The Consumer
　　The consumer has a bit more to it, though still pretty straightforward.  It's intended to be run from a monitoring tool like daemontools or supervisord, so there is a very small CLI harness that just passes the CLI arguments into EC_Consumer and runs it.  After parsing the CLI arguments, EC_Consumer polls kestrel for new jobs, and runs them through our standard batch job infrastructure.  Until we have more confidence in PHP's long running process ability, I added an optional maxium jobs argument, which will stop the consumer from processing more than X jobs and then terminate.  The monitoring service (supervisord) will then just restart it in a matter of seconds.  I also added an optional debug argument for testing, so you can see every action as it happens.  The CLI harness looks like this:

#!/bin/env php
<?php
// External application bootstrapping
require_once __DIR__ . '/cli_init.php';
// Instantiate and run the consumer
$consumer = new EC_Consumer($argv);
$consumer->run();
　　view raw consumer_cli.php This Gist brought to you by GitHub.
　　And the main consumer class, EC_Consumer, looks something like this:

<?php
/**
* Enterprise queue consumer interface, called by bin/consumer_cli.php
*
* @author Bill Shupp <hostmaster@shupp.org>
* @copyright 2010-2011 Empower Campaigns
*/
class EC_Consumer
{
/**
* Instance of {@link Zend_Console_Getopt}
*
* @var Zend_Console_Getopt
*/
protected $_opt = null;
/**
* Which APPLICATION_ENV to run under (see -e)
*
* @var string
*/
protected $_environment = null;
/**
* The kestrel server IP
*
* @var string
*/
protected $_host = null;
/**
* The kestrel server port
*
* @var int
*/
protected $_port = null;
/**
* The kestrel queue name to connect to
*
* @var string
*/
protected $_queue = null;
/**
* Whether we should show debug output
*
* @var bool
*/
protected $_debug = false;
/**
* Maximum # of jobs for this process to perform (for memory fail safe)
*
* @var int
*/
protected $_maxJobs = null;
/**
* Current job count
*
* @var int
*/
protected $_jobCount = 0;
/**
* Parses arguments from the command line and does error handling
*
* @param array $argv The $argv from bin/ecli.php
*
* @throw Zend_Console_Getopt_Exception on failure
* @return void
*/
public function __construct(array $argv)
{
try {
$opt = new Zend_Console_Getopt(
array(
'environment|e=s' => 'environment name (e.g. development)'
. ', required',
'server|s=s' => 'kestrel server, format of host:port'
. ', required',
'queue|q=s' => 'queue name (e.g. crawler_campaign)'
. ', required',
'max-jobs|m=s' => 'max jobs to run before exiting'
. ', optional',
'debug|d' => 'show debug output'
. ', optional',
)
);
$opt->setArguments($argv);
$opt->parse();
// Set environment
if ($opt->e === null) {
throw new Zend_Console_Getopt_Exception(
'Error: missing environment'
);
}
$this->_environment = $opt->e;
// @codeCoverageIgnoreStart
if (!defined('APPLICATION_ENV')) {
define('APPLICATION_ENV', $this->_environment);
}
// @codeCoverageIgnoreEnd

// Set server
if ($opt->s === null) {
throw new Zend_Console_Getopt_Exception(
'Error: missing server'
);
}
$parts = explode(':', $opt->s);
if (count($parts) !== 2) {
throw new Zend_Console_Getopt_Exception(
'Error: invalid server: ' . $opt->s
);
}
$this->_host = $parts[0];
$this->_port = $parts[1];
// Set queue
if ($opt->q === null) {
throw new Zend_Console_Getopt_Exception(
'Error: missing queue'
);
}
$this->_queue = $opt->q;
// Set max-jobs
if ($opt->m !== null) {
$this->_maxJobs = $opt->m;
}
// Set debug
if ($opt->d !== null) {
$this->_debug = true;
}
} catch (Zend_Console_Getopt_Exception $e) {
echo "\n" . $e->getMessage() . "\n\n";
echo $opt->getUsageMessage();
// @codeCoverageIgnoreStart
if (!defined('APPLICATION_ENV') || APPLICATION_ENV !== 'testing') {
exit(1);
}
// @codeCoverageIgnoreEnd
}
$this->_opt = $opt;
}

/**
* Polls the queue server for jobs and runs them as they come in
*
* @return void
*/
public function run()
{
$client = $this->_getKestrelClient();
$queue = 'enterprise_' . $this->_queue;
while ($this->_keepRunning()) {
// Pull job from queue
$job = $client->reliableRead($queue, 500);
if ($job === false) {
$this->_debug('Nothing on queue ' . $queue);
continue;
}
if (!isset($job['instance'])) {
echo 'Instance not set in queue job: ' . print_r($job, true);
continue;
}
$instance = $job['instance'];
if (!isset($job['jobName'])) {
echo 'Job name not set in queue job: ' . print_r($job, true);
continue;
}
$jobName = $job['jobName'];
$data = null;
if (isset($job['data'])) {
$data = $job['data'];
}
// Run the job
$returnCode = $this->runJob($instance, $jobName, $data);
if ($returnCode !== 0) {
$client->abortReliableRead($queue);
continue;
}
}
$client->closeReliableRead($queue);
}

/**
* Runs the job via bin/ecli.php
*
* @param string $instance The instance name to run the job under
* @param string $jobName The job name
* @param string $data Optional extra data
*
* @return int
*/
public function runJob($instance, $jobName, $data)
{
$cmd = BASE_PATH . '/bin/ecli.php '
. '-e ' . $this->_environment
. ' -i ' . $instance
. ' -j ' . $jobName;
if ($data) {
$cmd .= " '" . base64_encode(json_encode($data)) . "'";
}
$returnCode = $this->_passthru($cmd);
$this->_jobCount++;
$this->_debug('Job count: ' . $this->_jobCount);
return $returnCode;
}
/**
* Check to see if the job limit has been reached
*
* @return bool
*/
protected function _keepRunning()
{
return ($this->_maxJobs === null) ? true
: ($this->_jobCount < $this->_maxJobs);
}

/**
* Show debug messages
*
* @param mixed $message
*
* @return void
*/
protected function _debug($message)
{
if (!$this->_debug) {
return;
}
echo $message . "\n";
}
// @codeCoverageIgnoreStart
/**
* Calls the passthru() function and returns the exit code. Abstracted
* for testing.
*
* @param string $cmd The command to execute
*
* @return int
*/
protected function _passthru($cmd)
{
passthru($cmd, $returnCode);
return $returnCode;
}
/**
* Gets a single instance of EC_KestrelClient. Abstracted for testing.
*
* @return void
*/
protected function _getKestrelClient()
{
if (APPLICATION_ENV === 'testing') {
throw new Exception(__METHOD__ . ' was not mocked when testing');
}
return new EC_KestrelClient($this->_host, $this->_port);
}
// @codeCoverageIgnoreEnd
}
　　view raw EC_Consumer.php This Gist brought to you by GitHub.
　　Putting it together
　　Now that all the pieces are put together, let's take a look at in action. Adding example job "HelloWorld" to the queue "hello_world" from within our application looks something like this:

<?php
$producer = new EC_Producer();
$producer->addJob('hello_world', 'HelloWorld', array('foo' => 'bar'));
?>
view raw gistfile1.php This Gist brought to you by GitHub.
　　And finally, here's an example of running the consumer from the CLI harness, along with some example debug output of processing the job:
　　./bin/consumer_cli.php -e development -s 127.0.0.1:22133 -q hello_world -d -m 2
Nothing on queue enterprise_hello_world
Nothing on queue enterprise_hello_world
Nothing on queue enterprise_hello_world
Nothing on queue enterprise_hello_world
Running EC_Job_HelloWorld on instance dev under environment development
Hello, world! Here is my data array:
stdClass Object
(
    [foo] => bar
)
And here are my args: ./bin/ecli.php eyJmb28iOiJiYXIifQ==
Completed job in 0 seconds.
Job count: 1
Nothing on queue enterprise_hello_world
Nothing on queue enterprise_hello_world
Nothing on queue enterprise_hello_world
Nothing on queue enterprise_hello_world
Running EC_Job_HelloWorld on instance dev under environment development
Hello, world! Here is my data array:
stdClass Object
(
    [foo] => bar
)
And here are my args: ./bin/ecli.php eyJmb28iOiJiYXIifQ==
Completed job in 0 seconds.
Job count: 2
　　view raw example.txt This Gist brought to you by GitHub.
　　That's it! I'd be interested to hear how other folks are interfacing with kestrel from PHP.