Mixpanel Integration fails in syncing Historical backfill data
Setup
We created a Mixpanel connector pointing to our Mixpanel project, configured with:
Historical backfill start date: 10 Aug 2025 (approximately one year of data)
Replication schedule: daily
What we observed
We triggered a manual run to load the historical data. That run continued for 24 hours without completing the historical sync, and was still in progress when the next scheduled run was triggered.
The following day's run did not resume from where the previous run stopped. It started again from the beginning and attempted to load the entire year of historical data from scratch. This repeated for several days. On some days the Mixpanel API errored partway through, which also forced the next run to restart from the beginning.
Impact
Across those few days, more than one billion rows were loaded, the large majority of which are duplicates of the same historical data re-ingested on each run. The connector never advanced far enough to reach current data.
What we would like to understand
Why does the connector behave this way — specifically, why is progress from an incomplete run not retained, so that each subsequent run restarts from the original start date rather than resuming?
What is the recommended way to configure this connector so the historical backfill completes reliably and the pipeline then stays current — without re-ingesting the same data on every run?
