Laravel queue priority and reserved

Question

I'm getting a bit confused with laravel --queue option and reserved column (using mysql).
1- as laravel documentation has stated :

php artisan queue:listen --queue=high,low

In this example, jobs on the high queue will always be processed before moving onto jobs from the low queue.

lets say i have a high job and multiple low job. job high get fired and because of an issue pushed back to queue to fire again after 60 seconds. what happen in this period? does job low get started or no?

2- Does queue:work --daemon work just like queue:listen ? I mean does it process all jobs like what listen do?

3- what is reserved column for?

Answer 1

1. I can only guess here, but I assume high job would be retried before low ones. That's why you should use tries=3(or whatever) flag when running the worker or add public $tries = 5; to the Job class. Or you could take advantage of InteractsWithQueue trait for even finer control.

2. I believe, later Laravel versions run in daemon mode by default when you use queue:work command. Daemon worker processes are long lived and work faster as they don't need to restart the whole framework for every new job. The queue:listen command isn't even documented in the latest (5.4) version docs. The worker spawned by this command restarts itself before every job poll. Personally, I still use this, as I've got issues with memory when running daemon workers.

3. As I understand, this column is the way to mark when the specific job was started to be executed. In the queue.php config file you can specify when the connection should retry the job. I think the connection decides if the job should be retried by looking at reserved column.

Answer 2

Since there is no accepted answer, and I happened to search and investigate about all of these issues, let me answer them:

1. jobs from low will be processed, at least for 60 seconds. After 60 seconds, the job from high will become available, so it will be processed before of the remaining low jobs.

2. At writing time, March 2019, and since version 5:3, the suggested way to use queues is queue:work, that is the daemonized version of queue:listen

3. In current version there is a reserved_at column representing the time in which a job was assigned to a worker to process it.