Diagnosing Cluster Performance
We are currently running Oracle 10GR2 Standard Edition in a RAC environment. (Windows Server 2003)
Yesterday we had issues with one of our larger web page reports erroring with "user cancelled the operation" after approx. 30 seconds of processing. From the webserver side we have extended the auto timeouts to 1 or 2 hours before IIS kicks out the user.
Through the EM web console I noticed from the performance tab there was a lot more activity from the cluster than usual.(Average Active Sessions graph) Normally it is a small sliver with the occasional spike that quickly drops back to the thin sliver. Yesterday it was the same width as our User IO or CPU and it was a continious section that kept scrolling for the full width of the graph.(this is not normal in our instance). Today the performance graphs appear to be back to normal and the webpage report that was having issues can be run without issue.