I looked over it very briefly. As I remember, it was not in CVS, which
made it a bit more difficult to look at. I do plan on looking at it. Are
there any plans to integrate the HA stuff into the main JBoss branch?
If this is going to be added soon, then we can sync up on ideas about
proxy/handles. How did you handle JNDI interaction? Is is jnp specific?
--jason
On Wed, 2 May 2001, Sacha Labourey wrote:
> Hello Jason,
>
> > I also wanted to share some ideas that I had on how to use the
> > modifications
> > that I made with HA or auto-generated env properties. Basically I think
> ...
> > in a HA aware InitialContextHandle impl that would use jini, raw multicast
> > or whatever to find a suitable container.
> >
> > If anyone else has any ideas on the matter I would like to hear
> > them. If I
> > do not hear anything that I think I will work on implementing the above.
>
> Have you tried my implementation of HA for SLSB?
> (http://194.38.95.241/jboss/)
>
> It already implements a possible solution to this issue.
>
> In some words (nothing would be more accurate than testing it and reading at
> the proxy code ;) ), all possible JBoss instances share HA information such
> as bean location, ... When creating the proxy from the container, we give
> him information about all possible "targets" (i.e. JBoss instances
> containing a replica of the same bean than the current proxy being built).
>
> The proxy is then able to use all these different targets when performing
> invocations (or home lookup, create calls, ...) Furthermore, the current
> proxy implementation is able to "resynch" its state with the JBoss cluster
> (i.e. set of JBoss instances) to update its list of targets (to take in
> account, for example, newly started nodes).
>
> Cheers,
>
>
>
> Sacha
>
_______________________________________________
Jboss-development mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/lists/listinfo/jboss-development
