Parallel thread spinning

All,

Hoping you can help.

Oracle 9.2.0.3, HP-UX 11i, Superdome, EMC drives.

Running a query against a partitioned table in parallel.

The child parallel process can spin out of control. As evidenced by top -h output which shows the ora_p032_xxx at 100%+ cpu usage.

During this time no one (including sys/sysdba) can login. No messages written to the alert log, etc. Nada.

I kill -9 this one spinning parallel thread and everything is back to normal.

(There is lots more details but...)

So 2 questions:

1. Anyone else having this problem?
2. Anyone have a way to track:

Oracle Parent SID = Unix PID => Child SID's => Child Unix PID's

*without* having the Oracle system tables available? Remember I can't login during the hang.

Thank you,

Dave Received on Fri Sep 12 2003 - 13:38:45 CDT