Old deployments of Gitlab might have a big number of old events to be
deleted. Such numbers cause the worker to timeout.
I've limited the amount of rows that should be destroyed at once to
10000, and increased how often pruning shall take place to 4 times a
day.
