Data pump export full RAC database in window single DB by network_link

Question

Hi Experts,
I have a window 32 bit 10.2 database.
I try to export a full rac database (350G some version with window DB) in window single database by dblink.
exp syntax as
exdpd salemanager/********@sale FULL=y DIRECTORY=dataload NETWORK_LINK=sale.net DUMPFILE=sale20100203.dmp LOGFILE=salelog20100203.log
I created a dblink with fixed instance3. It was working for two day and display message as
ORA-31693: Table data object "SALE_AUDIT"."AU_ITEM_IN" failed to load/unload and is being skipped due to error:
ORA-29913: error in executing ODCIEXTTABLEPOPULATE callout
ORA-01555: snapshot too old: rollback segment number with name "" too small
ORA-02063: preceding line from sale.netL
I stoped export and checked window target alert log.
I saw some message as
kupprdp: master process DM00 started with pid=16, OS id=4444
to execute - SYS.KUPM$MCP.MAIN('SYS_EXPORT_FULL_02', 'SYSTEM', 'KUPC$C_1_20100202235235', 'KUPC$S_1_20100202235235', 0);
Tue Feb 02 23:56:12 2010
The value (30) of MAXTRANS parameter ignored.
kupprdp: master process DM00 started with pid=17, OS id=4024
to execute - SYS.KUPM$MCP.MAIN('SYS_EXPORT_FULL_01', 'SALE', 'KUPC$C_1_20100202235612', 'KUPC$S_1_20100202235612', 0);
kupprdp: worker process DW01 started with worker id=1, pid=18, OS id=2188
to execute - SYS.KUPW$WORKER.MAIN('SYS_EXPORT_FULL_01', 'SALE');
In RAC instance alert.log. I saw message as
SELECT /*+ NO_PARALLEL ("KU$") */ "ID","RAW_DATA","TRANSM_ID","RECEIVED_UTC_DATE ","RECEIVED_FROM","ACTION","ORAUSER",
"ORADATE" FROM RELATIONAL("SALE_AUDIT"."A U_ITEM_IN") "KU$"
How to fixed this error?
add more undotbs space in RAC instance 3 or window database?
Thanbks
Jim
Edited by: user589812 on Feb 4, 2010 10:15 AM

Answer 1

I usually increate undo space. Is your undo retention set smaller than the time it takes to run the job? If it is, I would think you would need to do that. If not, then I would think it would be the space. You were in the process of exporting data when the job failed which is what I would have expected. Basically, DataPump want to export each table consistent to itself. Let's say that one of your tables is partitioned and it has a large partition and a smaller partition. DataPump attempts to export the larger partiitons first and it remembers the scn for that partition. When the smaller partitions are exported, it will use the scn to get the data from that partition as it would have looked like if it exported the data when the first partiiton was used. If you don't have partitioned tables, then do you know if some of the tables in the export job (I know it's full so that includes just about all of them) are having data added to them or removed from them? I can't think of anything else that would need undo while exporting data.
Dean