Hey Stefan,
sure. There is the chance however that binds are already in the existing
trace. Then we don't need more tracing .. ;-)
Regards
Lothar
On 16.09.2016 16:03, Stefan Koehler wrote:
Hey Lothar,
What you need is the statistic of the execution that went wrong. It would beYou get this statistic, if you dump with "all_executions" and the SQL is only
good to have the bind variables for that execution.
executed 27 times in this case - so no problem. Regarding binds you are
right - that's the reason why i have suggested the option
"plan_stat=all_executions,wait=true,bind=true" :)
Best Regards
Stefan Koehler
Freelance Oracle performance consultant and researcher
Homepage: http://www.soocs.de
Twitter: @OracleSK
Lothar Flatz <l.flatz@xxxxxxxxxx> hat am 16. September 2016 um 15:56
geschrieben:
Well, I would say "who cares which executions stats it is?".
What you need is the statistic of the execution that went wrong. It
would be good to have the bind variables for that execution.
Given that, you can repeat your bad execution any time and create
runtime stats.
Regards
Lothar