[jboss-as7-dev] AS7 -b 0.0.0.0

[Brian Stansberry]

On 9/22/11 8:50 AM, Jim Tyrrell wrote:
>
> Form the core, If I can add something to this discussion or ask a question?
>
> I agree with adding -b to 7.0.2, good catch and fix, and IMHO there are
> a whole host of options that "used" to be there, that need to just work
> for making peoples lives easier.
>
>  From my clustering labs I have used a command line like this:
> ./run.sh -c node1 -g A -u 224.0.0.0 -m 1110 -b 192.168.200.2
> -Djboss.Domain=A -Djboss.jvmRoute="node2" -Djboss.messaging.ServerPeerID:0=2
>
> Something like this: -Djboss.messaging.ServerPeerID:0=2 is a "macro"
> that replaces values in the config file, does a feature that like work
> today?
>

In 7.0.x expressions (fka system property substitution) are supported in 
some places related to interface and socket configuration. For 7.1 we 
need to add a lot more.

(For devs interested in best practices on how to do that, look at the 
usage of MAX_POOL_SIZE in [1].)

> Potentially I would hope all of those options could be considered for
> migration,

1) Re: -c -- anyone interested, let's have a debate.

In AS7 -c foo.xml would be a shorthand for --server-config=foo.xml

I don't like that because:

a) It's only a rough translation of what -c meant in AS < 7. People 
would have to read docs to figure out how to use it, so is there much 
benefit?
b) There's no clean equivalent for domain.sh. It could be a shorthand 
for --domain-config=foo.xml but it could also be a shorthand for 
--host-config=foo.xml. So which?

(I could be convinced to support -c=foo.xml as a shorthand for 
-server-config=foo.xml and --domain-config=foo.xml if I feel people 
understand the issues and still want it. I've had conversations with 
people about it before, but they've gotten sidetracked when other things 
came up. And I could be much more easily be convinced by a nice patch.)

2) -g -- In AS < 7 this sets system property jboss.partition.name. I'll 
let Paul Ferraro comment on this one if he wants; he knows better than I 
if that system property fits into what he's doing with clustering.

3) -u -- IMHO it makes sense to use this and use the resulting system 
property in our standard config files for the sockets that have a 
multicast address.

4) -m -- Same as -u.

5) Any -D, it's up to the subsystem leads to determine if it makes sense 
to use an expression for some attributes in our default configs.

> I am not smart enough to know all of the possibilities, but
> guessing we had a whole lot more.
> How do we identify them?

./run.sh -h shows all the cmd line switches.

For stuff previously set via -D, subsystem leads should know their own 
subsystems; i.e. if they had system property substitution somewhere 
before, check if it still makes sense.

> How do we figure out if they should be in the future release/
>
> I would also hope things like this get into a migration guide, although
> it sounds like they are getting closer.
>

[1] 
https://github.com/jbossas/jboss-as/blob/master/ejb3/src/main/java/org/jboss/as/ejb3/subsystem/StrictMaxPoolResourceDefinition.java


-- 
Brian Stansberry
Principal Software Engineer
JBoss by Red Hat