Question

So I have the following query which gets a Duplicate PK error
what am I doing wrong?

////////
EDIT
The PK = ITEM & LOC

MERGE INTO item_loc_traits il 
USING   (
        SELECT  item, loc, reward_eligible_ind 
        FROM    (
                SELECT  dc_vert.item , dc_vert.loc, dc_vert.actie, 
                        dc_vert.reward_eligible_ind, 
                        MAX(dc_vert.actie) over 
                            (PARTITION BY dc_vert.item, dc_vert.loc) actie_max
                FROM    dc_item_loc_vert dc_vert
                ) 
        WHERE   actie = actie_max
        ) dc_vert
ON      il.item = dc_vert.item
        AND il.loc = dc_vert.loc
WHEN MATCHED THEN
       UPDATE   
       SET      il.deposit_code = reward_eligible_ind
        ,       il.last_update_datetime = SYSDATE;
WHEN NOT MATCHED THEN
       INSERT (item, loc, deposit_code, last_update_datetime, 
               last_update_ID, create_datetime)
       VALUES (dc_vert.item, dc_vert.loc, dc_vert.reward_eligible_ind,
               SYSDATE, 'CNVOBJ_SRC', SYSDATE);

Answer 1

There are probaby multiple rows in the merge source for a particular item, loc. You could test this with the following query:

SELECT  item, loc
FROM    (
        SELECT  item, loc, reward_eligible_ind 
        FROM    (
                SELECT  dc_vert.item , dc_vert.loc, dc_vert.actie, 
                        dc_vert.reward_eligible_ind, 
                        MAX(dc_vert.actie) over 
                            (PARTITION BY dc_vert.item, dc_vert.loc) actie_max
                FROM    dc_item_loc_vert dc_vert
                ) 
        WHERE   actie = actie_max
        ) sub
GROUP BY
        item, loc
HAVING  count(*) > 1

If this returns any rows, the when not matched part of your merge will try to insert multiple rows with the same item, loc combination. That would result in a primary key violation.