Fix ORA-01555: Snapshot Too Old in Oracle EBS Concurrent Requests

How to fix ORA-01555 in a concurrent request

1. 1Read the request log fully - ORA-01555 usually appears with rollback segment too small or names the undo segment; note which SQL statement in the program actually failed.
2. 2Check current UNDO_RETENTION and undo tablespace size with SELECT tablespace_name, value FROM v$parameter WHERE name = 'undo_retention'; and confirm the undo tablespace has autoextend on with enough headroom.
3. 3If the program is a long-running report against volatile tables with high DML during the run, increase UNDO_RETENTION and the undo tablespace size so older read-consistent images survive longer.
4. 4Review the program itself - if it is custom PL/SQL, look for COMMIT statements executed while a cursor opened earlier in the same procedure is still being fetched; this is the single most common root cause of ORA-01555 outside of undersized undo.
5. 5Where possible, restructure the loop to BULK COLLECT the needed rows first, then commit in batches after the fetch completes, rather than committing between fetches on an open cursor.
6. 6Check whether the concurrent program is scheduled to overlap with heavy batch updates, for example running during month-end postings, and consider rescheduling it to a quieter window.
7. 7If the error is intermittent and only under heavy load, consider Guaranteed Undo Retention only after confirming the undo tablespace is sized to absorb it, since it can cause DML failures (ORA-30036) if undo space runs out.

The undersized-undo cause

ALTER SYSTEM SET undo_retention = 10800 SCOPE=BOTH;

SELECT tablespace_name, autoextensible, bytes/1024/1024 mb FROM dba_data_files WHERE tablespace_name = (SELECT value FROM v$parameter WHERE name = 'undo_tablespace');

Common pitfalls

• !Increasing UNDO_RETENTION without also sizing the undo tablespace to support it - retention alone does nothing if the tablespace fills up.
• !Assuming ORA-01555 is a database bug and reopening a Service Request before checking for commit-inside-cursor-loop patterns in custom code.
• !Enabling Guaranteed Undo Retention without headroom, which trades ORA-01555 for ORA-30036 (out of space in undo tablespace) under heavy DML.
• !Scheduling a large custom report to run during month-end or year-end posting windows, when undo turnover is at its highest.
• !Fixing the symptom by simply re-running the failed request repeatedly instead of addressing the undersized undo or the code pattern causing it.