Re: Bug in dba_free

From: Yong Huang <yong321_at_yahoo.com>
Date: 7 May 2002 10:25:00 -0700
Message-ID: <b3cb12d6.0205070925.63663cdc@posting.google.com>

"Chuck" <chuckh_nospam_at_softhome.net> wrote in message news:<ab8lmj$g0h5i$1_at_ID-85580.news.dfncis.de>...
> Just added 64k to that datafile and an additional 100m magically appeared in
> dba_free_space. Don't know where I picked up the idea that you could only
> add that 64k at file creation time but it's obviously wrong. Thanks for the
> help Jonathan.

I think you got the impression from the statement that an Oracle extent has to be contiguous. But the word contiguous is as far as Oracle is concerned, i.e. datafile blocks are contiguous. As far as file system or OS is concerned, that's not the case. File allocation blocks (of size 512 bytes) could be anywhere. On UFS, the only tool I know that can tell you where each block is is Richard McDougal's filestat program (see one of the last few pages of his Solaris Internals). Unfortunately I can't find his program from any Web site. I got a copy from him directly.

# filestat system01.dbf

Inodes per cyl group:       6208
Inodes per block:           64
Cylinder Group no:          52
Cylinder Group blk:         5395712
File System Block Size:     8192
Device block size:          512

Number of device blocks: 665968

Start Block End Block Length (Device Blocks)

-----------    -----------  ----------------------
    5391520 -> 5391551      32
    5391808 -> 5391839      32
    5393328 -> 5393455      128
    6146240 -> 6148239      2000
    6150928 -> 6152127      1200
    6153360 -> 6162271      8912
    6530848 -> 6530879      32

...

Yong Huang Received on Tue May 07 2002 - 12:25:00 CDT

Re: Bug in dba_free_space?