Hi there
> [To HG] > I said the problem deal with cluster means: > assume we have 3 servers (S1, S2, S3) form a cluster and one Struts > plugins deal with home object caching (P1) > > when the application start up, P1 with get all home interfaces from > the cluster and also create all home objects and cache them. > > so if P1 get an object (O1) from S1 and cache it, this O1 is indeed an > instance created by S1 and if S1 go down, O1 would still reference to > S1 and become NULL > > The question comes to if i call some method to O1 at this time, will > the cluster manager redirect the request to another machine?? I don't > think it will hapen because you locally cache the object reference. > But if you get the home object from the JNDI tree again, I am sure the > cluster manager will preform a failover and redirect all request > target to S1 to another machine > > do I make any sense? > It makes perfectly sense. I haven't tried it out since I am not a cluster expert, but it must be the responsibility of the J2EE server to provide the clustering transparent to the client. Maybe this is a too optimistic thought, but nevertheless... It has to be done.. One of the main reasons to have a J2EE server, is that it manages and provides these services for you. I think you are right..When you refetch the home from JNDI, you get the home from another server i the cluster if S1 is down. Then you recreate the object from that home, and you are up running. Are there some cluster experts out there who can answer these questions...I know they somehow are quite off topic...but the discussion evolved...and isn't evolution great ? Glad I could help..at least I thought I could help.. :-) --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
