Expdp fails with ORA-39125, ORA-04031

Problem Description
I was performing data pump export operation and it fails with ORA-39125, ORA-04031 as below.
SQL> $expdp full=y dumpfile=fulldb.dmp directory=d userid=system/a

Export: Release 10.2.0.1.0 - Production on Tuesday, 16 December, 2008 17:07:44

Copyright (c) 2003, 2005, Oracle. All rights reserved.

Connected to: Oracle Database 10g Enterprise Edition Release 10.2.0.1.0 - Production
With the Partitioning, OLAP and Data Mining options
FLASHBACK automatically enabled to preserve database integrity.
Starting "SYSTEM"."SYS_EXPORT_FULL_02": full=y dumpfile=fulldb.dmp directory=d userid=system/********
Estimate in progress using BLOCKS method...
Processing object type DATABASE_EXPORT/SCHEMA/TABLE/TABLE_DATA
ORA-39125: Worker unexpected fatal error in KUPW$WORKER.GET_TABLE_DATA_OBJECTS while calling DBMS_METADATA.FETCH_XML_CLOB []
ORA-04031: unable to allocate 28 bytes of shared memory ("shared pool","SELECT /*+rule*/ SYS_XMLGEN(...","sql area","ub1[]: qkexrXformVal")

ORA-06512: at "SYS.DBMS_SYS_ERROR", line 95
ORA-06512: at "SYS.KUPW$WORKER", line 6235

----- PL/SQL Call Stack -----
object line object
handle number name
64E077B4 14916 package body SYS.KUPW$WORKER
64E077B4 6300 package body SYS.KUPW$WORKER
64E077B4 9120 package body SYS.KUPW$WORKER
64E077B4 1880 package body SYS.KUPW$WORKER
64E077B4 6861 package body SYS.KUPW$WORKER
64E077B4 1262 package body SYS.KUPW$WORKER
64CB4398 2 anonymous block

Job "SYSTEM"."SYS_EXPORT_FULL_02" stopped due to fatal error at 17:08:03

Cause of the Error
The ORA-04031 error returns whenever oracle attempts to allocate a large piece of contiguous memory in the shared pool but fails to allocate. Oracle first flushes all objects that are not currently in use from the pool and the resulting free memory chunks are merged. If there is still not a single chunk large enough to satisfy the request the ORA-04031 error is returned.

Since in 10g automatic memory management in enabled and this allows the dba to reserve a pool of shared memory that is used to allocate the shared pool, the buffer cache, the java pool and the large pool with the single parameter SGA_TARGET. So simply increase of SGA_TARGET likely solve the problem.

Solution
Let's see the sga_target value.

SQL> show parameter sga_t

NAME TYPE VALUE
------------------------------------ ----------- ------------------------------
sga_target big integer 100M

And SGA_MAX_SIZE value,
SQL> show parameter sga_max
NAME TYPE VALUE
------------------------------------ ----------- ------------------------------
sga_max_size big integer 100M

We see these two parameter settings are low. So we increase it and restart database. As setting of SGA_TARGET is depend on SGA_MAX_SIZE and to change SGA_MAX_SIZE we need to bounce the database in order to effect.

SQL> alter system set sga_max_size=300M scope=spfile;

System altered.

SQL> alter system set sga_target=300M scope=spfile;

System altered.

SQL> startup force
ORACLE instance started.

Total System Global Area 314572800 bytes
Fixed Size 1248768 bytes
Variable Size 96469504 bytes
Database Buffers 209715200 bytes
Redo Buffers 7139328 bytes
Database mounted.
Database opened.

SQL> show parameter sga_target

NAME TYPE VALUE
------------------------------------ ----------- ------------------------------
sga_target big integer 300M

Now the export operation runs smoothly as shared pool finds enough memory to do the operation.
SQL> $expdp full=y dumpfile=fulldb.dmp directory=d userid=system/a