-
Notifications
You must be signed in to change notification settings - Fork 8
Description
This issue comes from forum feedback.
It should be possible to get information on how much data approximately will need to be loaded in and how much have been processed to reach a stable state.
This is important for communication, say answer the question on how long approximately it will take before a stream is stable and can respond to updates.
The negative aspect of this is that getting the count could be costly in some systems, for instance the initial count on the source system might take minutes to run by itself. Therefore this should be an opt-in method,
Possible solution:
Events out are already sent through the metric system from almost all sources.
A new standard metric should be created for approximate expected events out for the initial batch.
It might be neccessary to create a new events out counter for the initial batch since the events out will continue to count for the delta.
But this can then be computed with sum(initial_events_out) / sum(expected_events_initial_load) to get a progress in %.