--
Do the following while connected as sysdba.
Please keep a spool of the output.
The procedure you are adopting is to clean up stranded 2pc entries in the database. Stranded entries means RECO process is not able to clean up the entries even after failure due to some reason and In which ever database you are doing it identify the local_tran_id from dba_2pc_pending and use the procedure to purge it The below steps would resolve the issue and you can commant out the
1. select * from dba_2pc_pending;
2. alter system disable distributed recovery ;
3. alter system set "_smu_debug_mode" = 4 SCOPE=SPFILE SID='*'; and restart the instance for oracle 10G
4. execute DBMS_TRANSACTION.PURGE_LOST_DB_ENTRY('
5. select * from dba_2pc_pending;
6. alter system enable distributed recovery;
