Parallel Query: lots of wait time in event "PX qref latch"

From: Herman de Boer <h.de.boer_at_itcg.nl>
Date: 1 Nov 2001 07:13:26 -0800
Message-ID: <bde5777e.0111010713.10d774b8@posting.google.com>

Dear *,

I am investigating a site running RDBMS 8.1.6.2.0, on HP V2500, UX11, 2Gb memory, a 4-cpu box hosting a ~400 Gb database with Parallel Query.

Most time is spent in event PX qref latch, with parallel degree 8 (so 16 slaves are running), with about 50% idle time (as given by top). There is a certain amount of disk I/O, but far below the bandwidth. Complex sql-statements, mostly pctas.

Anyone useful hints or advice or information what this event is about, and why so much time is spent in this one?

Kind Regards,

Herman de Boer
IT Consultancy Group bv Received on Thu Nov 01 2001 - 09:13:26 CST