Hey hey, I am using tap_facebook, but I think it doesn’t store the state after every day, if I run an insights job over 15 days and it fails on day 14 I have to rerun all 14 days? Right now my solution is to run several jobs one for each day in parallel but then I run into postgres deadlocks.

I think that's right, it uses a window. Is it a problem if it reruns all 14 days? Are you using a target that upserts or are you ending up with duplicate records in the destination?

I see the target upserts so it dedups