Re: Changing initrans

  • From: Riyaj Shamsudeen <riyaj.shamsudeen@xxxxxxxxx>
  • To: richa03@xxxxxxxxx
  • Date: Thu, 21 Aug 2008 16:14:49 -0500

Hi
I just tested this out (11g though) and itc was correctly updated after rebuild. One possible explanation is that dump block dumps block from the disk, until 11g ( I think). May be, block was not written to the disk yet from cache when you dumped the block again. Can you repeat your test case with checkpoint before dumping the blocks?

Before rebuild:
Object id on Block? Y
seg/obj: 0x11a47  csc: 0x00.3c3c79  itc: 2  flg: -  typ: 2 - INDEX
    fsl: 0  fnx: 0x0 ver: 0x01

alter index t_idx rebuild online initrans 20;


After rebuild;
Block header dump:  0x00415b13
Object id on Block? Y
seg/obj: 0x11a48  csc: 0x00.3c4115  itc: 20  flg: -  typ: 2 - INDEX
    fsl: 0  fnx: 0x0 ver: 0x01

Cheers
Riyaj
www.pythian.com
personal: http://orainternals.wordpress.com
richa03@xxxxxxxxx wrote:
Thanks all, but it still isn't 100% clear to me.

Tested on 64-bit 10.2.0.3 Solaris x86 - block size 8KB.

create table t ( x ) as select rownum from dba_objects;

create index t_idx on t(x);

[from a treedump on the object id of t_idx]
branch: 0x10001ac 16777644 (0: nrow: 31, level: 1)
   leaf: 0x10001ad 16777645 (-1: nrow: 485 rrow: 485) - block number 429
   leaf: 0x10001ae 16777646 (0: nrow: 479 rrow: 479)
... snipped ...

...
--
//www.freelists.org/webpage/oracle-l


Other related posts: