Page MenuHomePhabricator

Remove COUNT(*) from datasets when not useful in Superset & Turnilo
Open, MediumPublic


By default, Superset (and Turnilo?) include a COUNT(*) metric. This metric is misleading in pre-aggregated datasets (such as edits_hourly, pageviews_hourly, pageviews_daily) and should be removed.

Event Timeline

fdans added a subscriber: fdans.Jun 18 2020, 4:13 PM

we're not sure this can be removed, let's look into it

fdans moved this task from Incoming to Data Quality on the Analytics board.Jun 18 2020, 4:13 PM
LGoto triaged this task as Medium priority.Jun 22 2020, 4:07 PM
LGoto moved this task from Triage to Tracking on the Product-Analytics board.
mpopov added a subscriber: mpopov.Jun 22 2020, 4:08 PM

It can be removed manually in Superset (and has been for some datasets), but not sure if it gets automatically re-added when new data is ingested or when there are updates to those datasets.