Retry clear_locks! when ActiveRecord::Deadlocked is occurred

[nishidayuya]

Sometimes, ActiveRecord::Deadlocked is occurred in Delayed::Backend::ActiveRecord::Job.clear_locks!. And some jobs are still locked.

F, [2019-06-20T11:30:53.459511 #123099] FATAL -- : Mysql2::Error: Deadlock found when trying to get lock; try restarting transaction: UPDATE `delayed_jobs` SET `delayed_jobs`.`locked_by` = NULL, `delayed_jobs`.`locked_at` = NULL WHERE `delayed_jobs`.`locked_by` = 'delayed_job.1 host:XXXXX pid:123099 (ActiveRecord::Deadlocked)

worker_name is generated with PID. So, Delayed::Backend::ActiveRecord::Job.clear_locks! is retryable method. This pull-request makes Delayed::Backend::ActiveRecord::Job.clear_locks! to retry when ActiveRecord::Deadlocked is occurred.

Before pull-request

• When ActiveRecord::Deadlocked is occurred in Delayed::Backend::ActiveRecord::Job.clear_locks!, some jobs are still locked.

After pull-request

• When ActiveRecord::Deadlocked is occurred in Delayed::Backend::ActiveRecord::Job.clear_locks!, retry it (max 10-times) and clear locked jobs.