[wildfly-dev] Confusion around deployment runtime-name

[Jason Greene]

On Jun 13, 2013, at 9:12 AM, Bartosz Baranowski <bbaranow at redhat.com> wrote:

> Hi guys.
> 
> Im trying to fix one of EAPs BZ and this led me into deep state of confusion.
> Said BZ: https://bugzilla.redhat.com/show_bug.cgi?id=964446
> 
> 
> Ive looked into CLI and this is what it offers:
> 
> 
> [standalone at localhost:9990 /] deploy --help
> SYNOPSIS
> 
>    deploy ((file_path | --url=deployment_url)
>               [--script=script_name] [--name=deployment_name]
>               [--runtime-name=deployment_runtime_name]
>              ...
> 
> 
> file_path         - filesystem path to the application to deploy.
>                     If the deployment has to be added or replaced in the
>                     repository, either the file_path or the --url argument
>                     is required.
>                     The path can be either absolute or relative to the current
>                     directory.
> 
> --url             - ...
> 
> --name            - the unique name of the deployment. If the file path
>                     argument is specified the name argument is optional with
>                     the file name been the default value. If the file path
>                     argument isn't specified then the command is supposed to
>                     enable an already existing but disabled deployment, and in
>                     this case the name argument is required.
> 
> --runtime-name    - optional, the runtime name for the deployment.
> 
> 
> 
> A bit vague description of --runtime-name. --name gives a bit more info - it tells me that ${name} will be part of PathAddress in AS7 model.
> I kept digging and found this bit in DMR file:
> 
> Name by which the deployment should be known within a server's runtime. This would be equivalent to the file name of a deployment file, and would form the basis for such things as default Java Enterprise Edition application and module names. This would typically be the same as 'name', but in some cases users may wish to have two deployments with the same 'runtime-name' (e.g. two versions of \"foo.war\") both available in the deployment content repository, in which case the deployments would need to have distinct 'name' values but would have the same 'runtime-name'."
> 
> 
> Ok now to the point: 
> 1. ${runtime-name} is used as 'root' of deployment. It becomes a part of service name in jboss MSC. However model name contains ${name}/archive name as root of PathAddress.
> 
> Is this correct? I've been told that this is wrong and the ${runtime-name} should be also part of PathAddress.
> #1 is essentially a root of BZ. Operation seek service which contain ${name} as part of service name, while MSC has service which has the ${runtime-name}:
> lookup value:        jboss.deployment.subunit."jboss-as-ejb-in-ear-ear.ear"."jboss-as-ejb-in-ear-ejb.jar".component.GreeterEJB.START
> actual msc service:  jboss.deployment.subunit."nnn.ear"."jboss-as-ejb-in-ear-ejb.jar".component.GreeterEJB.START
> 
> 
> 2. From above description of runtime-name it seems I should be able to execute:
> deploy --runtime-name=nnn.ear /x/archive-1.0.ear
> deploy --runtime-name=nnn.ear /x/archive-1.1.ear
> 
> In fact I'm able to do that, but I've been told that this is wrong. Which is it?


You can only have one of a runtime name in a server instance (JVM). However, you might have two different server groups, which both have an account-manager.ear, but both are completely different. Both might also rely on that name (e.g. global JNDI names are created based on the deployment name). All servers share the same content repository, so to handle that situation they need different names to establish uniqueness in the repository, but identical runtime-names to allow them to execute correctly.

--
Jason T. Greene
WildFly Lead / JBoss EAP Platform Architect
JBoss, a division of Red Hat