Index Extent sizes while using parallel option

[1dbrown]

We recently had an "interesting" happenstance.
I wonder if anyone else has any experience with this.

We have 2 instances DEV for development and PRD for production. (Obvious, I know ......)

Both instances are:
Oracle7 Server Release 7.3.4.0.0 - Production
With the distributed, replication and parallel query options
PL/SQL Release 2.3.4.0.0 - Production

I needed to copy some data from DEV to PRD so I wrote a script that:
1) truncated the necessary PRD tables
2) dropped the indexes on the PRD tables
3) copied the data between the instances
4) created the indexes
5) analyzed the tables

Some of our maintenance scripts reported the oddest thing.
The indexes that were recreated had the correct initial extent,
but the next extents would vary in size.

Sample of the create index portion of the script:
CREATE INDEX IDX008003 ON
CKTLOCT(BDGIDC)
TABLESPACE IFP_IRML PCTFREE 0
STORAGE(INITIAL 10485760 NEXT 10485760 PCTINCREASE 0 )
PARALLEL(DEGREE 4)
;

The only thing that is different from all of our other processing, is that
I used the PARALLEL parameter.

I can find no doumentation on this side effect anywhere
(or I may be mis-interpreting what I'm reading).

[Halo]

Is there a difference in block size between the two dbs??

[1dbrown]

Both instances are on the same NT server and have the same block sizes.

I don't think this should matter, as there "should be" no inter-instance communication
when the indexes are created. You can let me know if I'm mistaken there.

The copy was accomplished via SQL*Plus COPY,
not a remote view or DB link.

Further to the original:

All extents follow the sizes set in the storage clause of the create index
command, except the final extent is small and "full",
which implies that the last extent is sized by something outside of
the parameters of the create index command.

I've tried leaving out the storage clause and relying on the tablespace
defaults, but the same symptoms arise.

I understand that the parallel option makes use of the direct load facilities.
I don't see that this would cause this side effect, as direct loads
with SQL*Load and pCTAS create tables with consistant extent sizes.

[jmodic]

This behaviour is documented in Oracle8i Concepts. Look at Chapter 26 Parallel Execution -> Parallel DDL -> Space Management For Parallel DDL