Fabric error: "Cannot obtain service: org.osgi.service.url.URLStreamHandlerService(url.handler.protocol=profile)"

[kodiak22]

Hi,

After successfully creating a new fabric (with fabric:create) and restarting the fuse container I get an exception (see later).

Looking at the OSGI service registry the required ProfileUrlHandler service is definitely there.

I've located a recent change in "io.fabric8.agent.service.Agent" that intends to fix bundle loading errors by wating 30 seconds for the ProfileUrlHandler service.

The source comment:

// FABRIC-790, FABRIC-981 - wait for ProfileUrlHandler before attempting to load bundles (in subsystem.resolve())

// (which may be the case with bundle.xxx=blueprint:profile:xxx URLs in io.fabric8.agent PID)

It seems that the fix doesn't work in this case. (Maybe 30 seconds is not enough to wait?)

Can anybody help me solve this, or should I file an issue?

thanks

Peter

Fuse version: 6.2.0.redhat-066 (early access)

Java version: 1.7.0_60 (Oracle HotSpot JVM)

Provision Exception:

• java.lang.IllegalStateException: Cannot obtain service: org.osgi.service.url.URLStreamHandlerService(url.handler.protocol=profile)
•   at io.fabric8.api.gravia.ServiceLocator.awaitService(ServiceLocator.java:125)
•   at io.fabric8.agent.service.Agent.awaitService(Agent.java:350)
•   at io.fabric8.agent.service.Agent.provision(Agent.java:333)
•   at io.fabric8.agent.service.Agent.provision(Agent.java:190)
•   at io.fabric8.agent.DeploymentAgent.doUpdate(DeploymentAgent.java:569)
•   at io.fabric8.agent.DeploymentAgent$2.run(DeploymentAgent.java:216)
•   at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:471)
•   at java.util.concurrent.FutureTask.run(FutureTask.java:262)
•   at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145)
•   at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615)
•   at java.lang.Thread.run(Thread.java:745)

[abhishekkoneru6790]

Hi, I am not a fabric expert but hope our analysis help.

When you create fabric, Create a new Completely new container and use that. Do no add anything on the default container.

This messes up the whole environment.

If you are already doing this, then even I am waiting for an answer

Hope this helps.

[kodiak22]

Hi Abhishek,

I didn't add anything to the new container, I just restarted it.

Besides, the whole point of the fabric container is that you can add new functionalities by assigning profiles to them.

We've been doing that constantly and the root container is no exception.

As for the error, I've managed to eliminate it by compiling a fabric-agent jar with a longer timeout (120s).

But after som further struggling I've concluded that the fabric-agent in this version (redhat-067 EA) is really buggy.

Our application happily runs in standalone mode, but not in a child container, where it doesn't even start properly.

Peter

[grgrzybek]

Hi

[FABRIC-790] Provisioning failure when using profile:xxx references - JBoss Issue Tracker was to fix issues related to provisioning resources with "profile:" url. But you may be right - this timeout should be configurable.

regards

Grzegorz Grzybek

[grgrzybek]

Also I have to mention that setting this await for "profile:" url handler helped a lot with all our automated tests which create child containers.

WIthout it they failed with "java.net.MalformedURLException: Unknown protocol: profile" exception

regards

Grzegorz Grzybek

[kodiak22]

Hi Grzegorz,

Thanks for the reply. As I wrote, I've managed to eliminate this error, but there are others that keep holding Us from using this EA version.

So we decided to wait for a more stable version, until then we're going to use the vanilla karaf-2.4.1 for development.

regards

Peter

[grgrzybek]

I've changed this await for "profile:" url handler and replaced it with more (and longer) retries of downloading the artifacts.

So [FABRIC-790] Provisioning failure when using profile:xxx references - JBoss Issue Tracker is still correct, but there should be no more problems like the one from the beginning of this thread.

regards

Grzegorz Grzybek

[kodiak22]

I will try it in the next build.

thanks

Peter

[grgrzybek]

The fix is present in 1.2.0 branch of Fabric8: Switch from waiting for URL handlers to more download attempts · 2b5d6ef · jboss-fuse/fabric8 · GitHub

regards

Grzegorz Grzybek