Details
Major Description
Below is a sample usage of a managed service factoriy with factory-pid = "rg.apache.karaf.someservice"
<osgix:managed-service-factory factory-pid="org.apache.karaf.someservice" update-strategy="container-managed" auto-export="all-classes">
<bean class="org.apache.karaf.SomeService"/>
</osgix:managed-service-factory>
From the karaf shell I use the config command to create a new instance of that service say "instance1":
karaf@root> config:edit org.apache.karaf.someservice-instance1
karaf@root> config:propset somekey somevalue
karaf@root> config:update
As a result a new instance of the service is created and the configuration is persisted in the file etc/org.apache.karaf.someservice-instance1.cfg
The problem is that its assigned a random service.pid of the form org.apache.karaf.someservice.b95c9421-77e8-474d-8083-7665a0cfcedf and when I try to update the configuration
karaf@root> config:edit org.apache.karaf.someservice-instance1
It doesn't work. Instead I have to use the assigned pid (which requires me to perform a config:list and then use correlate):
karaf@root> config:edit org.apache.karaf.someservice.b95c9421-77e8-474d-8083-7665a0cfcedf
We need to change the config commands in a manner that will either:
a) Enforce the use of a PID: org.apache.karaf.someservice-instance1 or
b) Provide a flag so that when editing existing configuration the filename can be used.
If you can think of other alternatives please discuss.