zenstruck
Fix single server locking
The existing behaviour auto-released the lock once the schedule:runcommand finished execution.
For short running tasks (~1 second or less), auto-releasing the lock can lead to a case where if a cron starts slightly later on another server, the lock from the first invocation has already been deleted, and the task is executed for a second time. For a cluster of 3 app servers running in the same datacenter using Redis to store locks, I've noticed this happening fairly frequenty on ~1 second jobs (~30% of the time).
This change relies on the DEFAULT_TTL constant to tidy-up locks rather than doing so in this extension. I'm not entirely sure this is the best way of handling things, so I'm open to other ideas!
I'm glad you found this, because this saved me a couple of hours of figuring out why it wasn't working. Now i'm hoping this gets merged soon.
