Understanding Data Deletion: Impact on $insert_id Events

when we initiate a data deletion, does it delete all events with the same $insert_id or only the ones that match the deletion query?

5 comments

· Sorted by Oldest

Julianne Bozzini
APP
·
·
Hi Vlad- Julianne from Mixpanel Support here! Mixpanel expects the $insert_id to be unique to each event. Are the events duplicates that share the same $insert_id, or are they completely separate events? Can you share more on how the deletion is targeting the events? The deletion targets events based on the criteria specified, so having the same $insert_id as an event targeted for deletion wouldn't automatically mean another event with the same $insert_id would be deleted. However, if you're targeting based on deletion criteria that both events meet, then both would be deleted.
Vlad S.
·
·
they are essentially the same event sent twice, but on occasion they may have a minor custom property be different; if a deletion query used that property, would that leave behind events with the same insert_id, event name and timestamp, but a different value for that property?
Julianne Bozzini
APP
·
·
Ah- that makes sense. One issue you may run into is that, due to the query time de-duplication, you may not be able to target the specific instance of the event for deletion. That said, let me double-check on the behavior in that specific case.
🙌1
Julianne Bozzini
APP
·
·
Heard back from engineering on this, and there isn't a guaranteed behavior since it would depend on which event meets the criteria, as well as whether we had already done compaction-time deduplication before the deletion occurs. Since Mixpanel doesn't support upserts, ideally, there shouldn't be non-duplicate duplicates to begin with. Given that there isn't a guaranteed consistent behavior, I'd recommend fully deleting the event data and, once the deletion is processed, uploading the corrected event data.
👍1
Vlad S.
·
·
awesome, thanks for looking into this!
1