Hi,

Thank you Tervor & Richard, After posting the mail, I run a dumpdb (to 3590
tape), format, loaddb, auditdb last night. It toke 6 hours totally. And db
size can be reduced.

In what case can the auditdb be omitted? This step toke more than 3 hours.
Before the dump, I have full backup the db, "halt quiesce" or just "halt",
but in each case, it prompts for auditdb fix=yes after loaddb.

From output of auditdb, there are messages like:
ANR4112W AUDITDB: Storage pool BACKUP_D2, specified as a destination for
copygroup STANDARD in management class CP2ACP_D2,
 policy set ACTIVE, domain DM_7_SQLBT, does not exists.
ANR4073I AUDITDB: Invalid file aggregation information encountered for node
 RS66TEST - information will be corrected.

I don't understand why there are ANR4073I messages, as the db was orginally
 restored from a v2 server and delete filespace have been performed.
For nodes with ANR4073I, I think all filespaces are deleted and under those
 nodes backup have performed in the v3 server. With only migration
but not reclaimation nor move data have been performed in the v3 server.

To repeat the test, I will fix all the ANR4112W before the next dumpdb,
format, loaddb. Hope that auditdb can be skipped.

Regards,
Eric Tang