Not binding to localhost by default in jboss-head was: Re: [jboss-dev] Clustering bootstrap still taking a long time

[Dimitris Andreadis]

>> However, "-b localhost" should be the default!
> 

Yes, that should be changed. The code in this regard should resemble that main() in Branch_4_2.

> Yeah, it's not.  If you don't specify -b or otherwise specify how 
> JGroups binds, JGroups picks the first non-loopback interface.
> 
> The only issue here is how much we care about whether we can still say 
> "go to two nodes and './run.sh -c all' and you'll have a cluster". TBH, 
> I don't much care about that any more and have no objections to having 
> the AS set a localhost default.  Would be interested in opinions of 
> others though.

You mean deep copy 'all' to all2 then:

run -c all  -b x.y.z.1
run -c all2 -b x.y.z.2

That should work, like it does in AS 4.2, yes.

One glitch, is the "Post Office" of the new messaging. It prints out a warning/stacktrace 
when the 2nd instance in the cluster boots. I think that should work out of the box without 
problems.

> 
> (BTW, clustering JBM requires assigning a unique server peer id to each 
> node, so "go to two nodes and './run.sh -c all' and you'll have a 
> cluster" doesn't work anymore anyway. JBM guys might find a solution to 
> that though.)