feat(deletions): Re-attempt deletions after 6 hours

[armenzg]

This task attempts to flip the column once every 3 hours, thus, it could be more than 24 hours before we re-attempt a deletion.

Changing this to 6 hours it would ensure that the deletion is re-attempted within 24 hours and re-attempted more often.

[cursor]

Bug: Long-running deletions may be incorrectly reset

Reducing the timeout from 24 hours to 6 hours may cause legitimate long-running deletions to be incorrectly identified as stuck. Large deletions that process in many chunks could take more than 6 hours to complete while remaining in_progress=True. When _reattempt_deletions resets these active deletions, they may be picked up again by _run_scheduled_deletions, potentially causing concurrent deletion attempts on the same object or forcing restarts of in-progress work.

 

[armenzg]

Deletion tasks need to finish within 20 minutes:

sentry/src/sentry/taskworker/namespaces.py

Lines 39 to 43 in 498dbfc

	deletion_tasks = app.taskregistry.create_namespace(
	"deletions",
	processing_deadline_duration=60 * 20,
	app_feature="shared",
	)

sentry/src/sentry/deletions/tasks/scheduled.py

Lines 70 to 78 in 498dbfc

	@instrumented_task(
	name="sentry.deletions.tasks.run_scheduled_deletions",
	namespace=deletion_tasks,
	)
	def run_scheduled_deletions() -> None:
	_run_scheduled_deletions(
	model_class=RegionScheduledDeletion,
	process_task=run_deletion,
	)