RE: Timestamps in trace files (and other trace file oddities)

It's been about a decade that I have been working with AIX. There must be tools that give you a breakdown by cpu. Else you can do a ps -ef or ps aux and watch your process. If I'mm right you should see it consuming 99% cpu all the time.

I have been looking after Peoplesoft systems for several years and I know those cascading NLs. They can drive up logical reads and cpu usage to astronomical heights. Do you by any chance have optimizer_index_cost_adj or optimize_index_caching changed from their defaults? Care to send me the sql and a 10053 trace of the explain? A warning though. I will be out camping the next three days, so the earliest I will be able to look at it is Monday (july 28th).

At 01:54 PM 7/24/2003 -0800, you wrote:
>Wolfgang,
> There are 4 cpu's, and file# 65 block# 6041 is from the driving
> table of
>the 5 table join (all NL joins). I will take a look at v$bh to see what
>blocks from the other tables are in memory next time I run this. Aside from
>this indirect approach, any other suggestions on confirming your plausible
>hypothesis? Is there a way to breakdown the workload of individual cpu's?

Wolfgang Breitling
Oracle7, 8, 8i, 9i OCP DBA
Centrex Consulting Corporation
http://www.centrexcc.com Received on Thu Jul 24 2003 - 16:21:02 CDT