> If I had the problem you describe, shouldn't the
> cluster work at first and then collapse after a
> couple of minutes? If so, this is not the case here.
> In my case it seems more as if the nodes don't join
> the same party for some reason.
That is exactly what happened. Data session replication would initially
work for about 15 minutes.
However, as you can make replication work with multiple nodes on a single box,
nodes not being able to join the same party must be a multicasting/networking problem.
>
> WRT low level problems, I have tried a pair of small
> C-programs (I can post here if required) that create
> a multicast group and print messages (server side)
> that clients posted after joining the mcast group.
> That seems to work so far.
>
> Cheers
> Mobi.
[Message sent by forum member 'updmst' (updmst)]
http://forums.java.net/jive/thread.jspa?messageID=348927