Currently on the PI extractor configuration file you can specify the end timestamp for the backfilling of a timeseries by using the “to” parameter on the backfill section of the configuration file.
What we would expect is to have something relatively close to this set date. However, the docs say that it can overshoot by the amount of datapoints specified on the filling chunk size, this results in over-ingestion of datapoints.
We would like to have a way to control the backfilling more precisely since we currently cannot limit the overshooting of the backfilling without also affecting the front-filling given that both depend on the same data point chunk size parameter (cdf-chunking>data-points).