Hello Kaniz,

It's working till 1896-01-01 00:00:00 which rules out the spark's supported date range issue.

Secondly I don't want the accurate timestamp but time delta subtracted in to_utc_timestamp should be the exact value being added in from_utc_timestamp or vice versa but in the snippets you can see there is 5 minutes +- issue.

 P.S. It works correctly on job cluster and sql warehouse. This issue is only in interactive cluster.