Re: Weird extent sizes in locally managed tablespace

But are you sure no object was ever allocated to the tablespace, perhaps for a special one-time task, and then dropped or moved to another tablespace?

It is interesting that some odd size holes exist, but unless you are absolutely sure that every object ever allocated to the tablespace is still allocated to it there is no way to be sure if the odd sizes are due to some quirk in the logic or to reuse of free extents left by deleted objects.

Oracle does not guarentee that all allocated extents will be 64k, 1M, 8M, or 64M only that these are the extents sizes it is going to try to use. The logic does grab odd extent left at the end of datafiles, backfills toward the front of the file, and when unable to get the desired next extent size drops down and grabs a smaller extent size.

There is no real way to say for sure why you find odd sizes but the odds are the sizes used were consistent with the logic at the time allocated.

HTH -- Mark D Powell -- Received on Thu Aug 04 2005 - 09:15:25 CDT