Unable to establish connection tcp://127.0.0.1:6001 when trying to publish job using memory driver

[acelaya]

I have started to experiment with RoadRunner 2.11, and the HTTP server was pretty easy to set-up and worked almost out of the box.

However, I'm now trying to get jobs working too, but when I try to check the availability of Jobs (or skip that check and dispatch a task directly) I get a Spiral\Goridge\Exception\RelayException exception with message Unable to establish connection tcp://127.0.0.1:6001.

This is my RR dev config:

version: '2.7'

server:
  command: 'php ./bin/roadrunner-worker.php'

http:
  address: '0.0.0.0:8080'
  pool:
    num_workers: 3
    supervisor:
      max_worker_memory: 100

jobs:
  pool:
    num_workers: 2
    max_worker_memory: 100
  timeout: 300
  consume: ['myqueue']
  pipelines:
    myqueue:
      driver: memory
      config:
        priority: 10
        prefetch: 10

logs:
  mode: development
  channels:
    http:
      level: debug # Log all http requests, set to info to disable
    server:
      level: debug # Everything written to worker stderr is logged
    metrics:
      level: debug

The roadrunner-worker.php file looks like this:

<?php

declare(strict_types=1);

use Mezzio\Application;
use Psr\Container\ContainerInterface;
use Psr\Http\Message\ServerRequestFactoryInterface;
use Psr\Http\Message\StreamFactoryInterface;
use Psr\Http\Message\UploadedFileFactoryInterface;
use Spiral\RoadRunner\Http\PSR7Worker;
use Spiral\RoadRunner\Jobs\Consumer;
use Spiral\RoadRunner\Worker;

(static function (): void {
    $rrMode = getenv('RR_MODE');
    /** @var ContainerInterface $container */
    $container = include __DIR__ . '/../config/container.php';

    if ($rrMode === 'http') {
        // This was spin-up as a web worker
        $app = $container->get(Application::class);
        $worker = new PSR7Worker(
            Worker::create(),
            $container->get(ServerRequestFactoryInterface::class),
            $container->get(StreamFactoryInterface::class),
            $container->get(UploadedFileFactoryInterface::class),
        );

        while ($req = $worker->waitRequest()) {
            try {
                $worker->respond($app->handle($req));
            } catch (Throwable $e) {
                $worker->getWorker()->error((string) $e);
            }
        }
    } else {
        // This was spin-up as a task worker
        $consumer = new Consumer();

        while ($task = $consumer->waitTask()) {
            try {
                var_dump($task);
                $task->complete();
            } catch (Throwable $e) {
                $task->fail($e);
            }
        }
    }
})();

At some point, in some HTTP request I do this (simplified):

use Spiral\RoadRunner\Jobs\Jobs;
use Spiral\RoadRunner\Jobs\Serializer\JsonSerializer;

$jobs = new Jobs(null, new JsonSerializer());
if (! $jobs->isAvailable()) {
    // It always ends up here.
    // I debugged the isAvailable code and noticed it always ends inside the catch block with the error reported in the title
    return;
}

$queue = $jobs->connect('myqueue');
$task = $queue->create(SomeEvent:class, [...]);
$queue->dispatch($task);

If I remove the $jobs->isAvailable() check, then the Unable to establish connection error is thrown when $queue->dispatch($task); is called.

According to the docs, if the task is dispatched inside a RoadRunner worker (which is the case) I do not need to pass the connection settings, so I'm not sure what's missing.

[rustatian]

Hey @acelaya 👋🏻
You need to add the RPC plugin to the configuration because Jobs works over RPC.

https://github.com/roadrunner-server/roadrunner/blob/master/.rr.yaml#L19

[rustatian]

I mean, internally, it works via in-memory; however, pushing Jobs from the outside requires some transport. In our case, this transport is RPC 😃

[rustatian]

if (! $jobs->isAvailable()) {
    // It always ends up here.
    // I debugged the isAvailable code and noticed it always ends inside the catch block with the error reported in the title
    return;
}

Please, update spiral/roadrunner-jobs package to the latest version; here are some deprecation notes about this method. You may safely delete this check.

[acelaya]

Yeah, I have seen in the docs that this is not needed anymore. I will get rid of it.

BTW, adding the rpc config worked like a charm. Thanks a lot :)

For some reason I thought I had to add that only if I wanted to change the default address.

[rustatian]

My pleasure 🤝 Enjoy RR ❤️