You didn't miss it, it wasn't there. You can load balance by routing subsequent requests back to the original server. This is a perfectly valid way of handling requests. Cocoon will work with this approach quite easily.You are under a slight misperception. In a clustered environment all requests must go to the cluster, but to any server in the cluster - not just the one the user logged into.
Oh - maybe I misread the email.. I must have missed the part
about them running a cluster. Just that they had machines
running at different locations and that a request may
go to any of these servers.
However, you can also have your servlet container replicate sessions between servers. If you use this method, then you've got to code your Cocoon site quite carefully to make sure that it works correctly. I don't believe that flow will work correctly, as it doesn't share its continuations, as they aren't shared within sessions.
Therefore, if you want to use stateful flow, you've got to use the first method.
As I say, replicated sessions won't work for stateful Cocoon flow. (So I believe).Cocoon needs to make sure that it either
a) stores what it needs in the session so that the session can be replicated
b) always handles null values by reloading the correct data.
True - but this goes for any web development.. to replicate
sessions you should probably store data into it :) But,
I've only had experience with session replication under
iPlanet.. and we just used the JDBC session manager. (had
to use that since the support staff would pull machines
out of service during normal operating hours)
Regards, Upayavira
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
