That was a typo. The nonclustered primary key actually was created on the correct student_attendance table and not the bkup.

If I allow dupes then there will be dupes in the new, partitioned table, but this is not acceptable. The current, nonpartitioned table has no dupes. The new, partitioned table can NOT have any dupes or the data will be invalid.

In my testing I removed the primary key and instead placed a nonunique index on the ID column and after the insert from the original table into the new table of the new table had 514,785 duplicated ID values, resulting in 1,179,720 lost records (some of the ID values were duplicated more than once). 23,436,483 distinct ID values became 22,256,763 distinct values.