The problem is that the Group By is applied prior to the aggregate (“Max”) function, and the Max function is thereby run on the result of the Group By, and you are including the “Stamp_Date” column in your Group By clause.

Essentially, every unique occurrence of the Group By fields is being constructed in the result set, and then the Max is applied to that result.

What you need to do is first obtain the max(stamp_date) value for the Field_Index values, then get the record data that match that combination. This requires a little more complex query:

select
	al1.Stamp_Date, al1.log_index, al1.log_comment, al1.stamp_uid
from action_log al1
where
	al1.log_index = '1007' 
	and al1.doc_id = '49107052FD224F68BF64DA79E36E583E'
	and not exists (
		select 1 from action_log al2
		where al2.log_index = al1.log_index
			and al2.doc_id = al1.doc_id
			and al2.Stamp_Date > al1.Stamp_Date
		)

This will retrieve the latest entry for each combination of log_index and doc_id, by finding the one with the highest Stamp_Date.

If I only select the stamp_date sql returns just the max stamp_date. but if I add in the other select fields I get all rows with log_index = ’1007′ for the doc_id