Larry,
You're right - I'm out of date.
Oracle 8.1.7.4 (which is what I've been
fiddling with for the last couple of hours)
seems to be doing something a lot smarter
than earlier versions.
I'm still trying to figure out what happens
when - but I'm having a lot of trouble because
the 10046 level 8 traces are not reporting anything
like the correct wait states.
In the test case I have, which generates about
30 extents in the finished table - the QC is handing
out 13 blocks at a time to the PX slaves. Each
slave is building its own mini-table for most of
the time.
However, somewhere near the end of the process,
I think the QC starts tell slaves to start filling holes
in each other's closing extents (this may be the
meaning of a 'hwn_brokering' hint I have found in
the executable). I have one set of trace files which
shows 3 out of 4 PX slaves all contributing to a
single extent near the end of the final table.
My problem, though, is the PX slave traces
are only showing about 25% of the writes
that they should be showing, and they are
missing about 30% of the reads that should be
showing, and repeating reads about 20% of the
time. This doesn't make it easy to figure out
what happens when. The upshot, though, is
that I keep seeing every extent filled except
the last one when doing:
create table t2 nologging parallel (degree 4)
as
select /*+ parallel(t1,4) */ * from t1;
Thanks for passing giving me the heads-up on
this one.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Coming soon one-day tutorials:
Cost Based Optimisation
Trouble-shooting and Tuning
Indexing Strategies
(see http://www.jlcomp.demon.co.uk/tutorial.html )
____UK_______March 19th
____USA_(FL)_May 2nd
Next Seminar dates:
(see http://www.jlcomp.demon.co.uk/seminar.html )
____USA_(CA, TX)_August
The Co-operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
----- Original Message -----
To: <[EMAIL PROTECTED]>
Sent: 26 February 2003 17:53
> When doing a parallel insert like that, doesn't he keep all that in
temp
> segments in the table's TS and then merge all those as opposed to
leaving a
> lot of half-empty extents? Yes, he will try to allocate the initial
and next
> for each slave. But you would only end up with a single potentially
> partially populated extent? Maybe I'm thinking wrong but I'm pretty
sure
> that's what we see.
>
> Regards,
>
> Larry G. Elkins
> [EMAIL PROTECTED]
> 214.954.1781
>
--
Please see the official ORACLE-L FAQ: http://www.orafaq.net
--
Author: Jonathan Lewis
INET: [EMAIL PROTECTED]
Fat City Network Services -- 858-538-5051 http://www.fatcity.com
San Diego, California -- Mailing list and web hosting services
---------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB ORACLE-L
(or the name of mailing list you want to be removed from). You may
also send the HELP command for other information (like subscribing).
