HTTP transport for Elastic PHP clients
This is a HTTP transport PHP library for communicate with Elastic products, like Elasticsearch.
It implements PSR-7 standard for managing
HTTP messages and PSR-18 for sending HTTP requests.
Moreover, it uses the PSR-17 for building PSR-7
objects
like HTTP requests, HTTP responses, URI, etc.
It uses the HTTPlug library to automatic discovery a PSR-18 client, a PSR-17 factory and the HttpAsyncClient interface with Promise for asyncronous HTTP requestes.
The architecture of the Transport is flexible and customizable, you can configure it using a PSR-18 client, a PSR-3 logger and a custom NodePoolInterface, to manage a cluster of nodes.
Quick start
The main component of this library is the Transport class.
This class uses 3 components:
- a PSR-18 client, using ClientInterface;
- a Node pool, using NodePoolInterface;
- a PSR-3 logger, using LoggerInterface.
While the PSR-3
and PSR-18
are well known standard in the PHP community, the NodePoolInterface
is a new interface proposed in this library. The idea of this interface is to provide a class
that is able to select a node for a list of hosts. For instance, using Elasticsearch, that is a
distributed search engine, you need to manage a cluster of nodes. Each node exposes a common
HTTP API and you can send the HTTP requests to one or more nodes.
The NodePoolInterface
is a component that can be used to manage the routing of the HTTP
requests to the cluster node topology.
In order to buid a Transport
instance, you can use the TransportBuilder
as follows:
use Elastic\Transport\TransportBuilder;
$transport = TransportBuilder::create()
->setHosts(['localhost:9200'])
->build();
This example shows how to set the transport to communicate with one node located at localhost:9200
(e.g. Elasticsearch default port).
By default, TransportBuilder
will use the autodiscovery feature of HTTPlug
for the PSR-18 client, the SimpleNodePool
as NodePoolInterface
and the NullLogger
as LoggerInterface
.
The Tranport
class itself implements the PSR-18 and the
HttpAsyncClient interfaces,
that means you can use it to send any HTTP request using the Tranport::sendRequest()
function
as follows:
use Http\Discovery\Psr17FactoryDiscovery;
$factory = Psr17FactoryDiscovery::findRequestFactory();
$request = $factory->createRequest('GET', '/info'); // PSR-7 request
$response = $transport->sendRequest($request);
var_dump($response); // PSR-7 response
The sendRequest
function will use $request
to send the HTTP request to the localhost:9200
node specified in the previous example code. This behaviour can be used to specify only the URL path
in the HTTP request, the host is selected at runtime using the NodePool
implementation.
NOTE: if you send a $request
that contains already a host the Transport
will
use it without using the NodePool
to select a node specified in TransportBuilder::setHosts()
settings.
For instance, the following example will send the /info
request to domain
and not localhost
.
use Elastic\Transport\TransportBuilder;
$transport = TransportBuilder::create()
->setHosts(['localhost:9200'])
->build();
$request = new Request('GET', 'https://domain.com/info');
$response = $transport->sendRequest($request); // the HTTP request will be sent to domain.com
echo $transport->lastRequest()->getUri()->getHost(); // domain.com
Asyncronous requests
You can send an asyncronous HTTP request using the Transport::sendAsyncRequest()
as follows:
use Http\Discovery\Psr17FactoryDiscovery;
$factory = Psr17FactoryDiscovery::findRequestFactory();
$request = $factory->createRequest('GET', '/info'); // PSR-7 request
$promise= $transport->sendAsyncRequest($request);
var_dump($promise); // Promise
var_dump($promise->wait()); // PSR-7 response
The $promise
contains a Promise object.
A promise is an object that does not block the execution of PHP. This means the promise does not
contain the HTTP response. In order to read the HTTP response you need to use the wait()
function.
Another approach to use a promise is to specify the functions to be called on success and on faliure
of the HTTP request. This can achieved using the then()
function as follows:
$promise->then(function (ResponseInterface $response) {
// onFulfilled callback, $reponse is PSR-7
echo 'The response is available';
return $response;
}, function (Exception $e) {
// onRejected callback
echo 'An error happens';
throw $e;
});
For more information about the usage of Promise objetcs you can read the documentation from HTTPlug.
Set the number of retries
You can specify the number of retries for any HTTP requests. This means if the HTTP request will fail the client will automatically try to perform another request (or more).
By default, the number of retries is zero (0). If you want you can change it using the Transport::setRetries()
function, as follows:
use Elastic\Transport\TransportBuilder;
$transport = TransportBuilder::create()
->setHosts([
'10.0.0.10:9200',
'10.0.0.20:9200',
'10.0.0.30:9200'
])
->build();
$transport->setRetries(1);
$factory = Psr17FactoryDiscovery::findRequestFactory();
$request = $factory->createRequest('GET', '/info');
// If a node is down, the transports retry automatically using another one
$response = $transport->sendRequest($request);
This feature can be interesting as retry mechanism especially useful if you have a cluster of nodes.
You can read the following section about Node Pool
to understand how to configure the selection
of nodes in a cluster environment.
Node Pool
The SimpleNodePool
is the default node pool algorithm used by Tranposrt
.
It uses the following default values: RoundRobin as
SelectorInterface
and NoResurrect as ResurrectInterface
.
The Round-robin algorithm select the nodes in order, from the first node in the array to the latest. When arrived to the latest nodes, it will start again from the first.
* NOTE: the order of the nodes is randomized at runtime to maximize the usage of all the hosts.
The NoResurrect option does not try to resurrect the
node that has been marked as dead. For instance, using Elasticsearch
you can try to
resurrect a dead node using the HEAD /
API. If you want to use this behaviour you can use the
ElasticsearchResurrect class.
Use a custom Selector
You can specify a SelectorInterface
implementation when you create a NodePoolInterface
instance.
For instance, imagine you implemented a CustomSelector
and a custom CustomResurrect
you can
use it as follows:
use Elastic\Transport\NodePool\SimpleNodePool;
use Elastic\Transport\TransportBuilder;
$nodePool = new SimpleNodePool(
new CustomSelector(),
new CustomResurrect()
);
$transport = TransportBuilder::create()
->setHosts(['localhost:9200'])
->setNodePool($nodePool)
->build();
Use a custom PSR-3 loggers
You can specify a PSR-3 LoggerInterface
implementation using the TransportBuilder
.
For instance, if you want to use monolog library
you can use the following configuration:
use Elastic\Transport\TransportBuilder;
use Monolog\Logger;
use Monolog\Handler\StreamHandler;
$logger = new Logger('name');
$logger->pushHandler(new StreamHandler('debug.log', Logger::DEBUG));
$transport = TransportBuilder::create()
->setHosts(['localhost:9200'])
->setLogger($logger)
->build();
Use a custom PSR-18 clients
You can specify a PSR-18
client using the TransportBuilder::setClient()
function.
For instance, if you want to use Symfony HTTP Client
you can use the following configuration:
use Elastic\Transport\TransportBuilder;
use Symfony\Component\HttpClient\Psr18Client;
$transport = TransportBuilder::create()
->setHosts(['localhost:9200'])
->setClient(new Psr18Client)
->build();
As mentioned in the introduction, we use the HTTPlug library to automatic discovery a PSR-18 client.
You can use the TransportBuilder::setClient()
to specify the client manually, for
instance if you have multiple HTTP client library installed.
By default, if the PSR-18 client implements the
HttpAsyncClient
it will use it when using Transport::sendAsyncRequest()
. If you want you can override
this setting using the Transport::setAsyncClient()
function. That means you can use
a PSR-18 client for the syncronous requests and
a different HttpAsyncClient
client for the asyncronous requests.
Copyright and License
Copyright (c) Elasticsearch B.V.
This software is licensed under the MIT License. Read the LICENSE file for more information.