Http Session Replication in WL8.1 - Is This What It Does ?
Is someone able to confirm my understanding of in-memory http session
replication in WL8.1 :
* Session replication only occurs when the session data changes i.e.
when the user clicks on the page.
* Session replication doesn't occur at other times - specifically,
session data isn't passed on by a server shutting down or requested by
a server starting up
* The only record of which nodes hold data for a particular session
are held on the client-side ( typically in the cookie ). There's no
location information held on the server-side and WebLogic is therefore
reliant on the cookie to find session data. Since the cookie contents
can't be influenced between http requests, there's no value ( under
the current scheme ) in copying session data around
* The consequence of this is that if the particular nodes marked as
primary/secondary are restarted between requests, that session would
be lost - even if there were many other nodes in the cluster which
were up and running the whole time.
I presume this approach is to avoid either :
* having a central session-locator on the server-side ( single
point-of-failure/performance issue )
* having the performance/effort of asking which of all the nodes in
the cluster now has the latest version of the session data
Re: Http Session Replication in WL8.1 - Is This What It Does ?
Apologies - I didn't notice there was a more specific newsgroup. I've
now re-posted my message on