Sample() stage

[ktoso]

An operation which "drops every n-th element", useful when for example you just want to log some progress every 1000 elements etc: broadcast to the logging stream then sample().log()

Another use case: I want to migrate my data from X to Y, the migration is resumable, writes are idempotent, I sometimes need to write my offset, every 100 writes or so. sample().map(e => store(e.offset) (also known as mini-spark)

[drewhk]

Rather, take every n-th element, or take an element in every T period (both are useful).