elastic / elasticsearch-php Goto Github PK

It might be interesting to depend on the interface provided by the psr/log package instead of the monolog package directly.

This is a remider to create a great helper for bulk operations.

Take a look at #52

Hi,

I am new to this official php library of ES. First I was using third party's library.

My question is: Is there no need to include elasticsearch-1.0.0 in this php library?
that download from the home page of http://www.elasticsearch.org

I run composer somehow to download the files and dependencies.

Is there any front end of what indexes I have created?

More question are in pipeline

Thanks,
Sohail Ahmed

Hi,

How use percolator in elasticsearch-php : http://www.elasticsearch.org/guide/en/elasticsearch/reference/0.90/search-percolate.html

We can't use _percolator/index/type
But we can use /index/type/_percolate :

    $paramsEs['index'] = 'ptf';
    $paramsEs['type']  = 'alerts';
    $client->percolate($paramsEs);

How do for register my percolator ?

Thanks

Hi, looking inside of the elasticseach-php, I saw that guzzle is used, but I think that httpful (32k zip) can do the same, and is very smaller in comparation to guzzle (1M~ zip).

https://github.com/nategood/httpful

Hi there, I just tried to use the mlt endpoint, my request looks like
$query['index'] = 'myIndex'
$query['type'] = 'myType'
$query['id'] = $video['_id'];
$query['min_term_freq'] = 1;
$query['min_term_freq'] = 1;
$query['mlt_fields'] = 'title';
I get:
more_like_this requires 'like_text' to be specified
So than I set the additional parameter $query['like_text'] = 'test';
then I get
Uncaught Exception: like_text is not a valid parameter

this parameter is not even on the whitelist in your endpoint mlt.php so I am wondering how this should work :P.
any idea? bug?
thanks in advance

This isn't exactly an issue with the codebase here, but I need help in identifying what it is that makes the php version requirement 5.3.9.

I have been made to try and implement this onto a PHP 5.3.3 system, and it fails completely.

Changing the requirement in composer allows composer to install all of the requirements, however it still fails.

I have no error report to use, as Codeigniter doesn't output errors (because it's so wonderful!)

Any helpful advice would be appreciated.

Hello,

I am just trying out the Elasticsearch PHP client for the first time.
Now I am getting a strange result, when counting objects in an index.

But from the beginning:
I wrote a small PHP script which adds 2 document to an index called "test".
Afterwards I wanted to valid the inserts by counting the objects in the index.

Calling
curl -XGET 'http://localhost:9200/test/_count?pretty=true'
gives me:

Array
(
    [count] => 2
    [_shards] => Array
        (
            [total] => 5
            [successful] => 5
            [failed] => 0
        )
)

which is correct.
However using:

$result = $client->count(array('index' => 'test'));

there are zero document in $result['count'].
I was already double checking the client log and did not find a solution.

Any ideas ?

Hi

i setup logPath to some value but when i run script i see error:
The stream or file "elasticsearch.log" could not be opened: failed to open stream in class monolog/monolog/src/Monolog/Handler/StreamHandler.php.

When i trace it i find the path in url property is without path i see that only elasticsearch.log value remain and whole path set in logPath is gone.

Hi

It seems passing raw json doesn't work as expected.

I pass the below - when I pass the below in console I get no results as the email '[email protected]' doesn't exist, when I pass via the php client it return every row.
$json = '"filter" : { "nested" : { "path" : "users", "query" : { "match" : { "users.email" : "[email protected]" } } } }';

$params['index'] = 'companies';
$params['type']  = 'company';
$params['body']  = $json;

I then passed:
$json = 'blah blah';

$params['index'] = 'companies';
$params['type']  = 'company';
$params['body']  = $json;

the result was not an error but all the rows.

Please help.

i'm looking for the way how to search in french.
Can you help me how to query with accent
Exemple med* to médecin
i try with

$params['body']['query']['query_string']=array(
'analyze_wildcard'=>TRUE,
'analyzer' => "french",
'query' => "" . $word . "");

$results = $this->elasticSearch->search($params);
but it doesn't work
Thanks

Hello

I'm trying to search an index with a raw json object.
I aint getting the desired results so i went searching in the code and found something that isn't clear to me.

in \Elasticsearch\Client

 $endpoint = $endpointBuilder('Search');
        $endpoint->setIndex($index)
                 ->setType($type)
                 ->setBody($body);
        $endpoint->setParams($params);

You first set the body and then the params.

in \Elasticsearch\Endpoints\Search
the setBody method:

if (is_string($query) === true) {
            $this->params['q'] = $query;
            $this->body = null;
        }

here is the raw json string added to the params array.
Then in setParams the 'q' key in the params array will be overwritten with $this->convertArraysToStrings($params); so my json is gone and no search is done.

    public function setParams($params)
    {
        $this->checkUserParams($params);
        $this->params = $this->convertArraysToStrings($params);
        $this->extractIgnore();

        return $this;
    }

This is my code:

 $json = '{
            "query" : {
                "match" : {
                    "name" : "' . $city . '"
                }
            }
        }';
$searchParams = array();
$searchParams['index'] = $indexName;
$searchParams['type']  = $indexType;
$searchParams['body'] = $querystring;
$queryResponse = $this->es->search($searchParams);

Thnx in advance!

Hi Zachary and other guys,

Say I have a number of hosts and some are dead, if I understand correctly, it takes 60 seconds to confirm the host is dead. In some cases, it's just too long and I am looking for a quicker way to test host and skip it if it is dead and go for another one.

Should I change the configuration (sorry I can't find option in http://www.elasticsearch.org/guide/en/elasticsearch/client/php-api/current/_configuration.html) to lower the timeout? or there are other approaches?

Thanks for reading. :)

Hi,

How can I do facets grouping of data, e.g. date_histogram, by using elasticsearch-php client?

I tried to send the JSON

{
"query" : {
...
},
"facets": {
    "histo1": {
      "date_histogram": {
        "key_field": "post_date",
        "interval": "month"
      }
    }
  }
}

The ['hits'] gives what match but there is no ['facets'] or something similar for the facets data.

I tried to read the doc on https://github.com/elasticsearch/elasticsearch-php/tree/master/docs, but it doesn't mention facets at all. Please correct if i am wrong and thanks for this elasticsearch php client.

Vic

Hi,
could you please update the composer.json in your release version 0.4.0 to use the last version of monolog 1.6.0

thank you.

Client returns "no active nodes" error if curl is disabled for PHP (even though there may be active elasticsearch nodes).

It would help troubleshooting to give a different error indicating the absence of curl.

For some request msearch() method return NULL instead array of responses.

Small sample:

$params = array(
    'hosts' => array('127.0.0.1'),
    'connectionClass' => '\Elasticsearch\Connections\CurlMultiConnection',
    'connectionParams' => array(
        'curlOpts' => array(
            CURLOPT_HTTPHEADER => array('Connection: keep-alive', 'Keep-Alive: 300')
        )
    )
);
$elastic = new \Elasticsearch\Client($params);

$params = array(
    'index' => 'Ads',
    'body' => array(
        array('type' => 'DefManagedBids'),
        array(
            'fields' => array('g', 'hta', 'cap', 'cTtl'),
            'query' => array(
                'filtered' => array(
                    'filter' => array(
                        'and' => array(
                            array('term' => array('sc' => 9000081)),
                            array('terms' => array('adtypes' => array('image_26', 'richmedia_26'))),
                            array('term' => array('countries' => 'HK')),
                            array('term' => array('languages' => 'LG')),
                            array('terms' => array('keywords' => array(
                                '348a448a51d1e0f0f5eee42337d12adc',
                                '8f99e430f3167d6bf65e0beb23896276',
                                '9336ebf25087d91c818ee6e9ec29f8c1',
                                'ee2a23af409b352d8f1819405bc770b2',
                                'ee8147f3ad132c3495314bb77fefa274',
                                'f3aa047479dae4b554c381ccbc2f5019',
                                'fe325cf304ee9155d513be1044bf064b',
                                'dbab14a2fd9a474f5aa30b06f130b368'
                            )))
                        )
                    )
                )
            )
        ),
        array('type' => 'DefRonBids'),
        array(
            'fields' => array('g', 'hta', 'cap', 'cTtl'),
            'query' => array(
                'filtered' => array(
                    'filter' => array(
                        'and' => array(
                            array('terms' => array('adtypes' => array('image_26', 'richmedia_26'))),
                            array('term' => array('countries' => 'HK')),
                            array('term' => array('languages' => 'LG')),
                            array('terms' => array('keywords' => array(
                                '348a448a51d1e0f0f5eee42337d12adc',
                                '8f99e430f3167d6bf65e0beb23896276',
                                '9336ebf25087d91c818ee6e9ec29f8c1',
                                'ee2a23af409b352d8f1819405bc770b2',
                                'ee8147f3ad132c3495314bb77fefa274',
                                'f3aa047479dae4b554c381ccbc2f5019',
                                'fe325cf304ee9155d513be1044bf064b',
                                'dbab14a2fd9a474f5aa30b06f130b368'
                            )))
                        )
                    )
                )
            )
        )
    )
);

$result = $elastic->msearch($params);

If I perform the same request with curl it show correct response:

{"responses":[{"took":0,"timed_out":false,"_shards":{"total":2,"successful":2,"failed":0},"hits":{"total":0,"max_score":null,"hits":[]}},{"took":0,"timed_out":false,"_shards":{"total":2,"successful":2,"failed":0},"hits":{"total":0,"max_score":null,"hits":[]}}]}

If I perform the same request with other values for search criteria it work properly and return array of responses for each query.

Please let me know if you need additional details.

I want to update some Documents with the bulk function. In some cases the id i want to update is not in my index.

In this cases i get the following Exception:

Fatal error: Uncaught exception 'Guzzle\Http\Exception\ClientErrorResponseException' with message 'Client error response
[status code] 400
[reason phrase] Bad Request
[url] http://localhost:9200/products/product/_bulk' in /home/thorsten/import/vendor/guzzle/guzzle/src/Guzzle/Http/Exception/BadResponseException.php:43

Here are my Request and Response from the Commandline:
Request:

curl -X POST "http://localhost:9200/products/product/_bulk" -d '
{"update":{"_id":"10b1894ebb4c44c12146193dff4b4d8d","index":"products","type":"product"}}
{"doc":{"doc_as_upsert":true,"stores":[{"id":"1","price":"45.00","shipping_time":"1"}]}}
{"update":{"_id":"b72dec695f188d99cf6c1b9916884f5c","index":"products","type":"product"}}
{"doc":{"doc_as_upsert":true,"stores":[{"id":"1","price":"43.00","shipping_time":"1"}]}}
{"update":{"_id":"c7212731d67e9731e7c2a9b1657f1dbf","index":"products","type":"product"}}
{"doc":{"doc_as_upsert":true,"stores":[{"id":"1","price":"91.00","shipping_time":"1"}]}}
{"update":{"_id":"ac40fa3f15bbcdf079b97c0f4d2c3eca","index":"products","type":"product"}}
{"doc":{"doc_as_upsert":true,"stores":[{"id":"1","price":"56.00","shipping_time":"1"}]}}
';

Response:

{"took":7,"errors":true,"items":[
{"update":{"_index":"products","_type":"product","_id":"10b1894ebb4c44c12146193dff4b4d8d","status":404,"error":"DocumentMissingException[[products][-1] [product][10b1894ebb4c44c12146193dff4b4d8d]: document missing]"}},
{"update":{"_index":"products","_type":"product","_id":"b72dec695f188d99cf6c1b9916884f5c","_version":2,"status":200}},
{"update":{"_index":"products","_type":"product","_id":"c7212731d67e9731e7c2a9b1657f1dbf","_version":2,"status":200}},
{"update":{"_index":"products","_type":"product","_id":"ac40fa3f15bbcdf079b97c0f4d2c3eca","_version":2,"status":200}}
]}

As you can see the first update statement returns an error but the following updates still working. With the PHP elasticsearch library my update crashs if there is 1 update on an id that doesn't exists.

Hi,
is there a way to retrieve the final url that is generated and used to query ES?
I think this can be very useful for unit testing and debugging.

Thanks

How can I create aliases from Client object

Hello,

I'm trying to delete some documents using the deleteByQuery function but I'm receiving ActionRequestValidationException[Validation Failed: 1: source is missing;].
From the function description I cannot understand what is wrong with the parameters I'm sending. I've tried to pass 'source' parameter also but the same exception occur.
What is the correct syntax here? I'm using 1.0 version of the library.

$params['index'] = 'yyyy';
$params['type'] = 'xxxx';
$params['body']['query'] = array(
'term' => array(
'entity.entity_id' => $id
)
);

try {
$response = $client->deleteByQuery($params);
}
catch (Exception $e) {
....
}

Hi all,

I am learning elasticsearch on windows.

I have run a node elasticsearch-1.0.0\bin\elasticsearch.bat

it connects on 9200

Then I have create an index using port 9200

it will now insert data using 9200

Now I run another node ( elasticsearch-1.0.0\bin\elasticsearch.bat ) it ll run on 9201

Question 1: Is it splits the load and data to new node as I have only configure port 9200 while creating index?
Question 2: After closing or lost on any node then data will remain available? If yes then in which node?

Thanks
Sohail Ahmed

Hello,

I'm writing a function and I want to get all installed plugins.
I'm using the following line, where $client is a parameter of type \Elasticsearch\Client

$client->nodes()->info(array('node_id' => '_all', 'plugin' => TRUE));

This code is throwing this error: plugin is not a valid parameter
I've tried also with 'plugins' but the result is the same.
In the function comments it is written that this is the right syntax.

I can not apply a sort using the Elasticsearch\Client::search method as outlined below. The bug appears to be in AbstractEndpoint::convertArraysToStrings and it incorrectly attempts to implode() any nested array into a comma separated list but the 'sort' search criteria can be a simple string or another nested array of parameters.

$es = new Elasticsearch\Client();
print_r($es->search(array(
    'index' => 'my_index',
    'type' => 'my_type',
    'sort' => array(
        // 'aFieldName', // this would work ... the next line doesn't
        'dateCreated' => array( 'order' => 'desc' ),
    ),
    'body' => array(
        'query' => array(
            'query_string' => array(
                'query' => 'search term'
            )
        )
    )
)));

Just wanted to check if there's any way to pass username/password credentials when configuring the connection?

Thanks!

I'm trying the following and receiving an error:

$params['index'] = 'my_index';
$params['type'] = 'my_type';
$params['id'] = 'document_id';
$params['body'] = array('existing_key' => 'new_value');

$results = $client->update($params);

and I'm receiving the following error:

{"error":"ActionRequestValidationException[Validation Failed: 1: script or doc is missing;]","status":500}

Sorry, but are bulk operations already available? Will it be?

I'm using a php based proxy and cant find a way to set up the connection so that the API uses it.

http://servername/proxy.php

the "/" gets turned to "%2F" if i add it to the [hosts]

If I include "/" with the [index] i get a "+"

https://gist.github.com/hogbinj/bc73ef01a045d9909724

Help!

Thx

J/.

Hello there

In the documentation at the paragraph "example usage" there is on line 2 the following use statement "use Elasticsearch;"

this should be "use Elasticsearch as Elasticsearch" because otherwise you get an warning:
" Warning: The use statement with non-compound name 'Elasticsearch' has no effect "

Hello,

in the composer file, there is a required for php >=5.3.9.

It mean that this library can be used only by php >= 5.4 ? Not by php 5.3.3 ?

Running 0.4.3 currently. Had some DNS updates that hadn't taken yet and attempting to create an index threw a php error at following location:

Undefined index: data
vendor/elasticsearch/elasticsearch/src/Elasticsearch/Namespaces/IndicesNamespace.php @ line 777

return $response['data'];

Once DNS had updated the call worked fine, which means client was just talking params, running the index create call and attempting to just return the data param sans checking for issues.

Side note, tried same call passing no host param and exception was correctly thrown

Hi all,
Iam new ElasticSearch.My task is indexing the huge files(txt,pdf etc..).
able to index the files individualy and search the content,but not with bulk.
(followed by ref: http://www.elasticsearch.cn/tutorials/2011/07/18/attachment-type-in-action.html)
can anyone give me the steps to index the huge files with in small time.

If you pass invalid arguments to the Elasticsearch\Client constructor, the error will cause the HTTP return status to be 0 , which will cause the web server to discard the error and return a 502 Bad Gateway

Software stack:

• Nginx 1.4.1
• php 5.3.10-1ubuntu3.9 (fpm-fcgi)
• Xdebug 2.1.0

Production steps:

$options = array();
// do not give a host.
$client = new Elasticsearch\Client($options);

Resulting output in nginx error log:

upstream sent invalid status "0 Hosts parameter must be an array of strings" while reading response header from upstream

There should also be a connection pool that performs no pinging, no matter what.

This is useful when your network setup is unable to perform HEAD requests, or you don't want to waste round-trips on pings (but simply want requests to timeout and retry if the node is down).

Related to #18

It should be possible to specify the Search body as a string (e.g. JSON string). The Search endpoint is currently packing string bodies into the 'q' parameter, which is incorrect and opinionated behavior.

The Search endpoint should simply set the body and let the Serializer decide what to do with it.

Note: If a string is provided, no validation should be performed (e.g. to check for valid JSON, etc).

Related to #3

I've just created my own class which extends from \Elasticsearch\Client to create a method to enable the TTL feature on a given index. I would love to see this functionality in the official bundle or i just did not found the method and created a redundancy?

class ElasticSearchClient extends \Elasticsearch\Client {
    public function enableTtl($params) {
        $index = $this->extractArgument($params, 'index');
        $type = $this->extractArgument($params, 'type');
        $body = json_encode(array($type => array('_ttl' => array('enabled' => true))));
        /** @var callback $endpointBuilder */
        $endpointBuilder = $this->dicEndpoints;

        /** @var \Elasticsearch\Endpoints\Update $endpoint */
        $endpoint = $endpointBuilder('Indices\Mapping\Put');
        $endpoint->setIndex($index)
            ->setType($type)
            ->setBody($body);
        $endpoint->setParams($params);
        $response = $endpoint->performRequest();
        return $response['data'];
    }
} 

If I use:

{
    "query" : {
        "match_all" : {  }
    },
    "facets" : {
        "tag" : {
            "terms" : {
                "field" : "tag",
                "all_terms" : true
            }
        }
    }
}

from the documentation, as a PHP array converted to json by the php-api, the all_terms part doesn't work.

When running the modified version of the Client constructor below, I found that the assigment $this->transport = $this->params['transport']; takes about 40ms.

public function __construct($params = array())
{
$this->setParams($params);
$this->setLogging();

    $start = microtime(true);
    $test = $this->params['transport'];
    var_dump(round(microtime(true) - $start, 2) * 1000 . ' ms');
    var_dump($this->params);

    $this->transport    = $this->params['transport'];
    $this->indices      = $this->params['indicesNamespace'];
    $this->cluster      = $this->params['clusterNamespace'];
    $this->dicEndpoints = $this->params['endpoint'];
}

Have you seen anything like this slow time?

Edit: Ok, I see now how Pimple is using closures to call the Transport class, but in any case, 40ms is still a question mark.

hi folks,

i try to pass a function_score request but it doesn't work.

$json = '{
"query": {
"function_score": {
"functions": [
{
"random_score": {

                   }
                }
             ],
             "boost_mode": "replace",
             "query": {
                "match_all": {}
             }
          }
       }
    }';

    $qry = array(
        'query' => array(
            'function_score' => array(
                'functions' => array(
                    array('random_score' => array())
                ),
                'query' => array(
                    array('match_all' => array())
                )
            )
        )
    );

    $searchParams['body']['query']['function_score']['functions']['random_score'] = array();
    $searchParams['body']['query']['function_score']['query']['match_all'] = array();

    //$searchParams['body'] = $qry;

    $retDoc = $elastic->search($searchParams);

if i pass $query i will get a exception: function_score: malformed query, expected a START_OBJECT while parsing functions but got a FIELD_NAME.

If i pass $json, it works. I'm not sure what i'm doing wrong.

Thanks

Hi,

I'm not able to execute search queries containing aggregations with this client. Though it seems the method used is always GET, that should not be problematic since a direct curl GET query returns the right response with aggregation metrics. Here is a code snippet of the query :

    $params = [
        'index' => 'whatever_index',
        'type' => 'whatever_type',
        'ignore_unavailable' => true,
        'ignore' => [404,500]
    ];
    $params['body'] = <<<JSON
{
    "aggs" : {
        "stats" : {
            "filter" : {
                "bool" : {
                    "must" : [
                        {
                            "term" : {
                                "valid" : true
                            }
                        },
                        {
                            "range" : {
                                "date" : {
                                    "gte" : "{$date_begin}",
                                    "lt" : "{$date_end}"
                                }
                            }
                        }
                    ]
                }
            },
            "aggs": {
                "over_time" : {
                    "date_histogram" : {
                        "field" : "date",
                        "interval" : "day",
                        "time_zone" : "{$timezone}",
                        "format" : "yyyy-MM-dd"
                    },
                    "aggs" : {
                        "types" : {
                            "terms" : {
                                "field" : "_type"
                            },
                            "aggs" : {
                                "unique_hits" : {
                                    "filter" : {
                                        "term" : {
                                            "unique" : true
                                        }
                                    }
                                }
                            }
                        }
                    }
                }
            }
        }
    },
    "size": 0
}
JSON;
    $results = $es->search($params);

The following lines of code docblocks are incorrect in Elasticsearch\Client.php:

• Line 130
• Line 285
• Line 321
• Line 429
• Line 474
• Line 513
• Line 560
• Line 651
• Line 693
• Line 808
• Line 889

I executing a query this is the query
$json='{
"filter": {"term": {
"url": "%s"
}},
"query": {"match": {
"linktext": "%s"
}}
}';
$qStr = sprintf($json,$url,$anchortext);

$params['index'] =get_domain_index($requesturl,"inverted");
$params['type'] ="inverted";
$params['fields'] ="_score,url,referer,linktext";
$params['body'] =$qStr;
$params['sort']="url_date:desc";
$params['size']=100;

try{
    $rc = $GLOBALS['e']->search($params);
}catch (Exception $error){
    echo "Not found\n";
    //echo $error." Error Getting record\n";
}

each record comes back with a NULL score
[98]=>
array(6) {
["_index"]=>
string(10) "inverted-p"
["_type"]=>
string(8) "inverted"
["_id"]=>
string(32) "255718a4b0d3a938360d4d2f92070e40"
["_score"]=>
NULL
["fields"]=>
array(3) {
["linktext"]=>
array(1) {
[0]=>
string(124) "
However when i exectue the same request with Sense I get the score
POST /inver*/_search
{
"filter": {"term": {
"url": "porscheofranchomirage.com"
}},
"query": {"match": {
"linktext": "porsche dealer"
}}
}
"_index": "inverted-p",
"_type": "inverted",
"_id": "be30dd48eea377b389623318186e3141",
"_score": 2.1395235,
"_source": {
"tld": "porscheofranchomirage.com",
"url": "porscheofranchomirage.com",
"referer": "indigoautogroupsocal.com",
"linktext": "Porsche",
"url_date": "2014-03-03"
}
},
{
"_index": "inverted-p",
"_type": "inverted",
"_id": "7582fb1eb94eec4a8fd3f7e0662b3e9d",
"_score": 1.0697618,
"_source": {
"tld": "porscheofranchomirage.com",
"url": "porscheofranchomirage.com",
"referer": "riversidepca.org",
"linktext": "Porsche of Rancho Mirage",
"url_date": "2014-03-03"

Can you please tell how to get the score in my code.

Thanks

Heya cool people of elasticsearch!

Just a random thought i had and thought of doing in my own elasticsearch code before i found this project was to utilize artax (https://github.com/rdlowrey/Artax) to run requests in parallel stuff like searching/count would be pretty neat to be able to have run parallel

The ping method in AbstractConnection is too lazy for me, here is the code:

        try {
            $response = $this->performRequest('HEAD', '', null, null, $options);

        } catch (OperationTimeoutException $exception) {
            $this->isAlive = false;
            return false;

        }

We only catch timeouts, but there is a lot of Exceptions that may occurs here:

• CouldNotResolveHostException
• CouldNotConnectToHost
• OperationTimeoutException
• TransportException

If thinks the 4 of them needs to be catched, if a node goes down (issue or server maintenance) I do not want my whole application to fail.

What do you think about it?

The real issue is, with this code:

        $params = array('hosts' => array (
            '127.0.0.1:9200',
            '127.0.0.1:9201',
        ));
        $this->client = new Client($params);

As the second node does not respond at all, the whole client is crashed with a CouldNotConnectToHost exception - we can't use the first node at all.

Is this the intended behaviour? I'm asking because if I query ES directly, it would return me a JSON object with exists:false property, but elasticsearch-php seems to throw a 404 error instead.

The StaticConnectionPool pings all hosts when it is first initialized, which can cause significant delay as it checks all hosts. For many PHP applications this is a waste of time and resources.

The connection pool should instead only ping hosts before they are used for the first time. The overall flow should look like this:

1. When connection pool initializes, mark all nodes "dead". This forces all nodes to schedule a ping the first time they are used
2. Select node
3. If "alive", return connection
4. If "dead"
   • If node has not been pinged recently, ping node
     • If ping succeeds, return connection
   • If node has been pinged recently (but still dead), add to skip list
5. If all nodes are "dead", force ping on each node in skip list, return if alive
6. If entire skip list fails too, throw exception

Related to #18

Support memcached transport in the future

I'm getting the following error on most calls:

 Missing stream url, the stream can not be opened. This may be caused by a premature call to close().

• PHP Version: 5.5.9
• OS: Ubuntu 12.04.4 LTS
• Nginx: 1.4.6

I'm not using composer for install, did a composer install and then added the files to the repository, including the autoload.php file.
Any ideas what the issue might be?

If i just want to include the php lib and don't have composer for various reasons, how would i do so.

Sorry I'm discovering the php api (I was doing all request manually with curl before today) and I was wondering how I could get _stats data with your api?