BICC Extract job timestamp used to extract data when there is a failure

Summary

What extract timestamp is used for each incremental job after a failure

Content

When there is a failure in BICC incremental job which timestamp is used to start the next scheduled extract job?  Does BICC use the extract timestamp less the prune minutes that would have been used on the scheduled job that failed or does it use the completed date of the failed job less the prune minutes.  Is the end time of the prior run used whether it was successful or failed?

Is there somewhere to see the extract timestamp that is being used to compare against each incremental object (last_update_date) in the datastores that is being analyzed to determine if there are new or updated records to include in the extract?