Hi ,
Can anyone explain me how CAI synchronizes it’s sessions , subsessions , adapters and bounded data to sessions (like data for server side grids) in cluster environment.Is there any point to pay attention while working in clustered environment ?
Thanks
Hi Serhat,
CAI is session sticky - once a brwoser connects to a server the browser must always go back to the same server. Inside the server then the corresponding CAI session (internally holding subsession and adapters) is kept.
Most (all?) application servers use the JSessionid for routing a request to its correct server. Though the CAI session management does not build on http sessions, you can of course use them. There is a default way of forcing CAI to internally create http sessions, by switching attribute…
createhttpsession=“true”
…in cisconfig.xml.
Hope this helps…
Bjoern
Hi Bjoern ,
Thanks for the aswer.But I didn’t understand in which case should I set “createhttpsession” parameter to true.Does CAI uses httpsessions instead of it’s own when I set this parameter to true?
By the way some j2ee servers (including tomcat) supports automatic failover. ( [url]http://tomcat.apache.org/tomcat-5.5-doc/cluster-howto.html[/url] , [url]Oracle Java Technologies | Oracle ) .Do you think that we can support automatic failover with CAI ?
… we do not internally use http sessions but “force” the servlet engine to create one. As consequence a JSessionId is assigned and communicated (just normal as cookie).
We currently do not support the automatic failover - the failover assumes that all objects that are kept in a session are serialized (and of course it creating a “hell” of effort with serailization).
Bjoern
Thanks a lot for the information