SAPSQL_ARRAY_INSERT_DUPREC¶

Duplicate record when inserting into database table.

A bulk INSERT (array insert) tried to write a row whose primary key already exists in the table, and the code didn't handle the conflict.

Symptom¶

ST22 shows:

Runtime Errors         SAPSQL_ARRAY_INSERT_DUPREC
Short text             Duplicate record when inserting into database table.

The dump details include the table name and the offending key values.

Cause¶

ABAP's INSERT ... FROM TABLE is a single SQL statement that inserts all rows at once. If any row violates the primary key constraint, the whole statement fails and dumps — unless you handle it explicitly.

Common root causes:

Running the same program twice (reprocessing without cleanup).
Two parallel jobs or users inserting the same key at the same time.
A dataset that already contains duplicate keys internally (the first duplicate row was inserted fine; the second hits the existing one).
Missing a DELETE / cleanup step before re-inserting in a load program.

Reproduce¶

DATA lt_data TYPE STANDARD TABLE OF ztable.

" Insert the same key twice
APPEND VALUE #( key1 = 'A' key2 = '001' value = 'X' ) TO lt_data.
APPEND VALUE #( key1 = 'A' key2 = '001' value = 'Y' ) TO lt_data.

INSERT ztable FROM TABLE lt_data.   " <-- dumps on second row

Fix¶

Use INSERT with SY-SUBRC checkUse MODIFY (insert or update)Deduplicate before insertingCatch the exception (7.40+)

For single-row inserts, check SY-SUBRC and handle duplicates:

INSERT ztable FROM ls_record.
IF sy-subrc = 4.
  " duplicate — skip or update
ENDIF.

If your intent is "insert if new, update if exists", use MODIFY:

MODIFY ztable FROM TABLE lt_data.
" SY-SUBRC is always 0; SY-DBCNT = rows affected

MODIFY does a full row overwrite

All non-key fields are overwritten. If you only want to update specific columns, use UPDATE ... SET instead.

Remove duplicates from the internal table before the bulk insert:

SORT lt_data BY key1 key2.
DELETE ADJACENT DUPLICATES FROM lt_data COMPARING key1 key2.
INSERT ztable FROM TABLE lt_data.

In newer releases you can catch database exceptions:

TRY.
    INSERT ztable FROM TABLE lt_data.
  CATCH cx_sy_open_sql_db INTO DATA(lx_db).
    " log lx_db->get_text( ) and continue
ENDTRY.

Note: this catches the dump but no rows are inserted — the whole array insert is atomic at DB level. You'll need to fall back to row-by-row inserts with individual error handling if partial success is needed.

Prevention¶

Always deduplicate (SORT + DELETE ADJACENT DUPLICATES) before a bulk insert.
In load/migration programs, add a DELETE FROM ztable WHERE ... or TRUNCATE before re-inserting.
Use MODIFY instead of INSERT when idempotency is more important than detecting unexpected duplicates.
For concurrent scenarios (parallel jobs), use database locking (ENQUEUE) or a sequence/GUID key that is inherently unique.