SigCle wrote:Explain with basic logic exactly how you could possibly Page In (pi) many more blocks than you Page Out (po).
one of my client Oracle DB Production ADDM said :
"Significant virtual memory paging was detected on the host operatingsystem"
OS : Sun Solaris Spark 64 bit
Oracle DB : 10.2.0.5
OEM always on 24 hours
I haven't get vmstat, glance but hope could get the information for next 6 hours since I writing this.
Based on two links below :
Virtual Memory PAging 11G
First I must analyze about vmstat 1 1000 and glance from the OS for gathering information about excessive of Oracle RAM usage which could suffer in swap.
and then after that decide whether should reduce : SGA (shared_pool_size and java_pool_size) and PGA and sort_area_size parameter
in order to giving breath for external OS services processes.
My Question :_
1. vmstat indicator to reduce sga and pga :
IF pi column compare with po column, if pi quite bigger than po and high wa so I should reduce SGA and PGA size ?
2. Is OEM could annoy Production System Process if OEM view performance data set to refresh every 15 seconds ?If you suspect that OEM is source of the problem, then shut it down since it is NOT required
jgarry wrote:Indeed, I've noticed this a LOT since upgrading one of my 'streams' to 18.104.22.168 (from 10.2.0.5). We hit a known bug which is incorrectly reporting swapping statistics in a v$ view.
I don't know about your OS, but on mine it can be a completely bogus message (and funnily enough, it spit out a bogus complaint just a couple of hours ago).