Bad plan during incremental ETL

Hi All-

We are using OBIA for some of our analytics. The data warehouse database version is 19.7 and is running on RH Linux 7. We see very poor performance (> 24 hours, before I cancelled it) on one of the steps. I've identified the SQL which is having the issue. Note, it is a select statement running on the source EBS database, which is 12.1.0.2 & Solaris 10.

SELECT STATEMENT  Cost = 16846262178695278	    
FILTER  	    
  HASH JOIN  	 2442M
    TABLE ACCESS FULL GL_PERIOD_STATUSES 	  5K
    HASH JOIN  	 2162M
      TABLE ACCESS FULL GL_LEDGERS 	  19 
      NESTED LOOPS OUTER 	 2162M
        HASH JOIN RIGHT OUTER 	 2162M
          TABLE ACCESS FULL GL_JE_BATCHES 	  2M
          MERGE JOIN  	 2162M
            HASH JOIN  	 2162M
              TABLE ACCESS FULL GL_CODE_COMBINATIONS 	  1M
              TABLE ACCESS FULL GL_JE_LINES 	 2162M
            SORT JOIN 	  10M
              TABLE ACCESS FULL GL_JE_HEADERS 	  10M
        TABLE ACCESS BY INDEX ROWID GL_ENCUMBRANCE_TYPES 	  1 
          INDEX UNIQUE SCAN GL_ENCUMBRANCE_TYPES_U1 	  1 
  VIEW  	 2162M
    UNION-ALL  	    
      FILTER  	    
        INDEX FULL SCAN GL_JE_LINES_N5 	 2162M
      NESTED LOOPS  	  1 
        NESTED LOOPS  	  1 
          TABLE ACCESS BY INDEX ROWID GL_JE_HEADERS 	  1 
            INDEX UNIQUE SCAN GL_JE_HEADERS_U1 	  1 
          INDEX UNIQUE SCAN GL_JE_LINES_U1 	  1 
        TABLE ACCESS BY INDEX ROWID GL_JE_BATCHES 	  1 
          INDEX UNIQUE SCAN GL_JE_BATCHES_U1 	  1