Azkaban

A small PHP client for the Azkaban workflow manager HTTP API.

Quick start

use Erikwang2013\BigData\BigData;

$azkaban = BigData::driver('azkaban', [
    'base_url' => 'https://azkaban.example.com:8443',
    'username' => 'azkaban',
    'password' => 'secret',
]);

$exec = $azkaban->executeFlow('demo', 'etl');
echo $exec['execid'], PHP_EOL;

echo $azkaban->executionLog((string) $exec['execid']);

Direct instantiation:

use Erikwang2013\BigData\Driver\Azkaban\AzkabanClient;
use Erikwang2013\BigData\Driver\Azkaban\AzkabanConfig;

$azkaban = new AzkabanClient(AzkabanConfig::fromArray([
    'base_url' => 'https://azkaban.example.com:8443',
    'username' => 'azkaban',
    'password' => 'secret',
]));
// Optional HTTP client and PSR-3 logger: new AzkabanClient($config, $http, $logger)

Configuration

AzkabanConfig::fromArray() accepts:

baseUrl is accepted as an alias of base_url.

Authentication

login() posts the credentials to /manager?action=login and returns the session.id. The id is cached on the client and reused; every other call logs in automatically when no session is cached yet.

$sessionId = $azkaban->login();

API

$azkaban->projects();                                  // GET /index?ajax=fetchuserprojects
$azkaban->flows('demo');                               // decoded fetchprojectflows payload
$azkaban->executeFlow('demo', 'etl', ['disabled' => 'a,b']);
$azkaban->execution('42');                             // decoded fetchexecflow payload
$azkaban->executionLog('42', 0, 50000);                // log text
$azkaban->cancel('42');                                // cancelFlow

executeFlow() posts project, flow and any extra $options as form fields. executionLog() targets Azkaban 3.90.x: GET /executor?ajax=fetchExecFlowLogs with offset/length; the server answers {"length": ..., "offset": ..., "data": "..."} and the returned data string (the log text) is what the method returns. The session id is sent as a session.id query parameter on every request.

Logging

AzkabanClient accepts an optional PSR-3 logger as its third constructor argument and defaults to Psr\Log\NullLogger. Debug records contain the HTTP method and URI only — credentials are never logged.

Errors

Transport failures throw Erikwang2013\BigData\Exception\ConnectionException. Any non-2xx HTTP status (including 3xx), invalid JSON, a login response without session.id, an execution-log response without data, or an Azkaban response carrying an error field raises Erikwang2013\BigData\Exception\RequestException, which exposes public $statusCode and $body. Both exceptions extend Erikwang2013\BigData\Exception\BigDataException.

Notes

• The log endpoint is modelled on Azkaban 3.90.x; other versions may differ.
• No automatic retries are performed.