Last year, I have said that I will increase AUL version by resolving LOB or compress block. Today I made a great step of supporting LOB type, so I released AUL 4.0.0 Beta.
You can download the beta release to test LOB recovery. Some new set option introduced to support LOB recovery, I will introduce then in later chapters. Why beta? I have been hoping to support LOB or compress block for a long time, and I have not made enough test on this feature, so I called it beta release. The following is some limits:
- No enough test made.
- Tested on on 10g version.
- Not all character set conversion supported for CLOB.
- Using index full scan to access LOB Index.
- Ignore chunk size now, cheat it as one block size always.
Compare other tools such as jDUL or Oracle DUL, AUL can recovery lob data into DMP format.
If you want to recover LOB data, AUL dictionary must exists (get from system tablespace or manually create them). I will try to do more test and make some improvement in the coming two or three months. Hope I can release it in two months (not beta version).
Comments (7)
Just to put the record straight ;-)
jDUL can :
- store LOBs in DMP format - including LONG/LONG RAWs larger then 0xFFFF bytes
- take care of chunk versions for read consistency
- take care of different chunk sizes
- doesn't need a full LOB index scan unload LOBs
Nevertheless - I'm impressed ;-)
Posted by koert | December 19, 2006 1:20 AM
jDUL made an improvement recently, great!
But jDUL did not put a demo program for test.
Posted by anysql | December 19, 2006 10:14 AM
There's a 'demo' version available that will extract only the first 10 rows of the first block of each table.
However - it's only freely available for people who have experienced a crash and is certainly not available for download.
Posted by koert | December 19, 2006 6:42 PM
Did you ever test AUL?
Posted by anysql | December 19, 2006 8:15 PM
I hit a problem on partitioned lob table to get the segment header of the LOB Index. I do not want to unload LOB$, PARTLOB$ table.
And another limit is that the CHUNK size of same table's LOB must keep same.
Posted by anysql | December 19, 2006 8:19 PM
Ran an old version once ... I'm too busy with my own things.
Why don't you want to unload LOB$ ? In case of a missing SYSTEM tablespace ? For partitioned lob tables also check out lobfrag$...
Posted by koert | December 20, 2006 7:01 AM
Currently I just unload USER$, OBJ$, TAB$, IND$, ... and COL$ as AUL dictionary information.
AUL cannot regenerate the table creation script for customer. I saw jDUL has this feature, that's great!
Posted by anysql | December 20, 2006 9:26 AM