1:22 p.m.
I'm inside a DeploymentProcessor and I want to execute a management
operation on another subsystem. How can I get a
ModelControllerClient in order to do that?
Stan
1:29 p.m.
On 11/15/2013 01:22 PM, ssilvert(a)redhat.com wrote:
I'm inside a DeploymentProcessor and I want to execute a
management
operation on another subsystem. How can I get a
ModelControllerClient in order to do that?
You shouldn't do that. A deployment operation should never change the
persistent configuration. Non-persistent configuration/resources should
be modifiable in other subsystem-dependent ways.
--
- DML
4:34 p.m.
On 11/15/13 1:29 PM, David M. Lloyd wrote:
On 11/15/2013 01:22 PM, ssilvert(a)redhat.com wrote:
> I'm inside a DeploymentProcessor and I want to execute a management
> operation on another subsystem. How can I get a
> ModelControllerClient in order to do that?
You shouldn't do that. A deployment operation should never change the
persistent configuration. Non-persistent configuration/resources should
be modifiable in other subsystem-dependent ways.
DUPs also run inside a service's start method (the deployment phase
service.) You can't invoke a management op that modifies anything from
inside a service start for a service that in turn was started as part of
a management op. At least not if the DUP is going to block waiting for
that op to complete. The 2nd op won't execute because the first op
holds an exclusive lock; the first op won't release that lock because it
is waiting for the service to complete start.
--
Brian Stansberry
Principal Software Engineer
JBoss by Red Hat
7:37 p.m.
OK. Thanks guys. I'll have to take a very different approach to my
problem then. Guess I've got the weekend to think it over.
On 11/15/2013 5:34 PM, Brian Stansberry wrote:
On 11/15/13 1:29 PM, David M. Lloyd wrote:
> On 11/15/2013 01:22 PM, ssilvert(a)redhat.com wrote:
>> I'm inside a DeploymentProcessor and I want to execute a management
>> operation on another subsystem. How can I get a
>> ModelControllerClient in order to do that?
> You shouldn't do that. A deployment operation should never change the
> persistent configuration. Non-persistent configuration/resources should
> be modifiable in other subsystem-dependent ways.
>
DUPs also run inside a service's start method (the deployment phase
service.) You can't invoke a management op that modifies anything from
inside a service start for a service that in turn was started as part of
a management op. At least not if the DUP is going to block waiting for
that op to complete. The 2nd op won't execute because the first op
holds an exclusive lock; the first op won't release that lock because it
is waiting for the service to complete start.
4381
days inactive
4382
days old
3 comments
3 participants
participants (3)
-
Brian Stansberry -
David M. Lloyd -
ssilvert＠redhat.com