> -----Original Message-----
> From: [EMAIL PROTECTED]
> [mailto:[EMAIL PROTECTED]]On Behalf Of
> Francisco Reverbel
> Sent: Wednesday, March 20, 2002 4:11 PM
> To: Bill Burke
> Cc: marc fleury; Scott M Stark; [EMAIL PROTECTED]
> Subject: RE: [JBoss-dev] Multiple server configurations
>
>
> Hello Bill,
>
> > > From: Francisco Reverbel [mailto:[EMAIL PROTECTED]]
> ...
> > > Must embed the container's JMX name into my IORs and route IIOP
> > > invocations through the MBean server. This has been in my todo
> > > list for quite a while...
> > >
> >
> > Is this portable? Can other orbs just ignore the extra JMX
> name. I can't
> > remember whether or not you can stuff anything you like in the
> IOR contexts.
>
> Yes, it is fully portable. I will not use an IOR context, will use the
> IOR object id instead.
>
So, you can't add abitrary IOR contexts?
> Every IOR has an "object id" field interpreted only by the server that
> created the IOR. Nobody else looks at it. A server can put whatever it
> wants into the object id field of the IORs it creates.
>
> > Another thing sort of related....I haven't been in the CORBA space for
> > over...But how's the fault-tolerance spec progressing? Can you stuff
> > multiple endpoints into 1 IOR for failover? Do other orbs
> recognize this?
>
> Yes, the spec allows you to stuff multiple endpoints within one IOR.
> If I remember correctly, it introduces the concepts of "object group"
> and "group IOR". But I do not know of any ORBs have implemented this
> feature in a spec-compliant way. Maybe it is just my ignorance, I am
> not following this issue very closely.
>
> Changing subject a little bit...
>
> Have you clustering guys thought about clustering and IIOP? Seems to be a
Yes :). Wanted to know if ORB's supported "object groups". How will we do
failover if Orbix, TAO, and/or JacORB don't support this feature? I was
hoping that JBoss clustering could support failover for even C++ CORBA
clients. I'll ping my friends at Iona. Anybody know the JacORB or TAO
guys? Iona/Jac/TAO are the main dudes, right?
> very good match. I believe IIOP runs a bit faster than JRMP for remote
> invocations. (For local invocations, JBoss optimized calls are
> unbeatable.)
> And you could use the IIOP "location forward" mechanism to do load
> balancing at the IIOP level. You can send a "location forward"
> reply to an
> IIOP client at any time. The client will transparently continue its work
> using the new location.
>
location forward is a good idea.
Bill
_______________________________________________
Jboss-development mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/jboss-development
