It seems Telerik OpenAccess have this behavior when it comes to releasing internal resources. If the connection timeout of telerik is set to 0 all resources is released when the last object scope is disposed. However, releasing the last scope and freeing internal resources sets the connection timeout to infinite (-1).
This results in a memory leak, since telerik querybuilders and caches references object scopes internally and thus prevent them from being garbage collected although they're disposed.
The current solution we have is to reset the connection timeout before creating a scope and after disposing a scope.
But this is giving us issues in long running transaction with an error "Not connected to oracle".
Does orm 2011.2 have better solution for the above problem?
Thanks
1 Answer, 1 is accepted
Thank you for pointing us to this behavior. I have added this to our backlog. We will try to address this as soon as possible. In the meantime, you can dispose the database manually to work around this. Please accept our apologies for any inconvenience caused.
Another thing came to my mind when working with Oracle databases. With the Q3 2011 release, we introduced 2 new connection pool modes. Since you are using Oracle as your backend, we would like to recommend using the 'Open Access Connection Pool with OpenAccess Connection Management' setting. This is because you will gain performance improvements when you use the internal driver connection pool. See this blog post for details.
Hope this information is helpful for you.
Feel free to ask if you have any other question.
Ralph
the Telerik team