Diagnosing an Oracle crash
561854Jan 23 2008 — edited Apr 29 2008We recently deployed a new version of a web app that uses Oracle 10g running on windows server and Oracle has been crashing regularly about every 10-20 hours since the deployment.
The Oracle instance is used only by this app is not subjected to much load, a maximum of a 8-10 requests per minute. A typical request runs four or five queries.
I say 'crashing' but I'm not sure if that is technically what is happening and I would like some guidance as to how to figure out what actually is happening.
The symptom we see is that application begins logging 'can not connect' exceptions. When this happens we see that the Oracle service is no longer running. Restarting the service gets everything going again for 10-20 hours. Then the same scenario repeats.
Where do I look and what do I look for?
What sort of scenario would lead to an Oracle to crash? The application is a lightly used web app but does have a new feature that uses AJAX. This may send bursts of requests that run a query with a 'like' clause but could this be enough to crash the db?
Thanks for any advice or guidance,
-=bill