> During migration (Step 2), we created a 2-week temporary Lambda that listens for S3 bucket events, transforms them into Parquet, and publishes them to S3. Due

> During migration (Step 2), we created a 2-week temporary Lambda that listens for S3 bucket events, transforms them into Parquet, and publishes them to S3. Due to the fact that both the source and destination buckets and prefixes were assigned to the same bucket, it triggered millions of times. I wouldn't so much call it recursive (as in subdivide and conquer), but rather endlessly processing work queued for yourself.