But how one can assign the startingVersion in production, it should pick the data from where the job got failed.I am encountering a similar issue where the checkpoint location is consistently updated with new offset values every 15 minutes, yet the s...