Inconsistent behavior on Laravel Queue Job - laravel

Question

I have a job that looks like this:
<?php
namespace App\Jobs;
use App\User;
use Illuminate\Bus\Queueable;
use Illuminate\Contracts\Queue\ShouldQueue;
use Illuminate\Foundation\Bus\Dispatchable;
use Illuminate\Queue\InteractsWithQueue;
use Illuminate\Queue\SerializesModels;
use Illuminate\Support\Facades\Redis;
class FakeJob implements ShouldQueue
{
use Dispatchable, InteractsWithQueue, Queueable, SerializesModels;
/** #var User */
private $user;
public $tries = 30;
/**
* Create a new job instance.
*
* #param User $user
*/
public function __construct(User $user)
{
$this->user = $user;
}
/**
* Execute the job.
*
* #return void
*/
public function handle()
{
try {
Redis::throttle('key')->allow(1)->every(60)->then(function () {
// Job logic...
throw new \Exception('fake exception');
$this->logInfo($this->user->name);
});
} catch(\Illuminate\Contracts\Redis\LimiterTimeoutException $e) {
$this->logInfo('inside LimiterTimeout catch, ' . $e->getMessage());
$this->release(1);
}
}
public function failed(\Exception $exception) {
// Send user notification of failure, etc...
$this->logInfo('inside failure, ' . $exception->getMessage());
}
public function logInfo($message) {
$path = storage_path('logs/FakeJob.log');
$logText = time() . ' ' . $message;
file_put_contents($path, $logText.PHP_EOL , FILE_APPEND | LOCK_EX);
}
}
Now you'll notice that I have a Redis throttler in the handle function, and that I immediately throw an error inside the callback. This is what I'm testing, and that error is what's going wrong.
You see, I'm trying to distinguish between LimiterTimeoutExceptions and exceptions thrown inside the Throttler callback. When I run the job through the sync connection, everything works as expected: FakeJob::dispatch($user)->onConnection('sync'); Errors that happen because I made a request within 60 seconds of a previous request go into the LimiterTimeoutException catch block, and errors that happen inside the Throttle callback go to the failed function.
But when I schedule the job through my default scheduler, which is Database, it seems as though EVERY error is going through the LimiterTimeoutException catch block, until I hit the retry limit - that error goes into the failed function, but not the 'fake exception' error.
I feel very confused about this.
Of note: even though my 'fake exception' ends up being caught by the LimiterTimeoutException catch block, the $e->getMessage() function does NOT return fake exception there for some reason. But I'm 100% certain that it is that error that is causing it to go there, because it's definitely not due to the throttler in those tests.

Answer 1

OK so actually what's happening is that it won't end up in the failed() function until it hits the retry limit, no matter what, regardless of any exception thrown. If I want to deal with LimiterTimeoutException errors in a different way from every other error, I'll have to wrap my Redis::throttle call in a try-catch and just check the type of error in the catch.