PQ spinning on wait event "flush prewarm buffers"
Oracle RAC 10.1.0.3
Google and Metalink did not turn up anything much - and nothing about this specific wait state.
When doing an index rebuild in parallel on RAC, I sometimes (maybe about 5% of the time?) get one or three PQs on one node spinning on:
enq: PW - flush prewarm buffers
The remaining 100 or so PQs across the rest of the RAC patiently waits on a PX Deq: Execution Msg as they're done and wanting to terminate.
The Large Hammer is current being used - I kill one of the stuck PQs and the remaining 1 or 2 gets unstuck. An ORA-12801 (error signalled in PQ) is thrown.
Any ideas what this event is about? Any comments/suggestitions as to what can be causing it, potential work-arounds, etc?
Yeah, I'm planning to upgrade to 10.2 (and OFCS 2), but that is easier said than done. :-)