[12.x] Don't fail database queue jobs on deadlock

[jackbayliss]

The index added in #59111 seems to have caused a lot of deadlocks, at least for me, which means I now have a lot of failed jobs, and logs due to concurrency errors, because I have a lot of workers listening on the same queues, rather than just one.

Rather than reverting that PR, thought I'd try this as I guess the index is useful still. But up to you 👍🏻

Currently I have a bunch of jobs in the failed_jobs table, where concurrency issues happen during marshalJob / markJobAsReserved

This PR returns null on concurrency errors so the worker moves on quietly, leaving the job available for the next poll and stops spamming logs.

Added a test for regression too.

Images / Trace here

1213 Deadlock found when trying to get lock; try restarting transaction (Connection: mysql, Host: xxxx), Port: 3306, Database: zzzz, SQL: update `jobs` set `reserved_at` = 1773439353, `attempts` = 1 where `id` = 112812452)

/srv/app/my-secret-app/x/vendor/laravel/framework/src/Illuminate/Queue/DatabaseQueue.php:422                                                          
/srv/app/my-secret-app/x/vendor/laravel/framework/src/Illuminate/Queue/DatabaseQueue.php:408                                                                                    
/srv/app/my-secret-app/x/vendor/laravel/framework/src/Illuminate/Queue/DatabaseQueue.php:294

[github-actions]

Thanks for submitting a PR!

Note that draft PRs are not reviewed. If you would like a review, please mark your pull request as ready for review in the GitHub user interface.

Pull requests that are abandoned in draft may be closed due to inactivity.