Hey Marcel,

I finally got a chance to try your suggested workaround. After provisioning the 
fragment individually, restarting the host bundle does not resolve the bundle 
fragment. However, the fragment does become resolved when both the host and 
fragment are provisioned together. I can use this approach to work around the 
issue, but do you think this the “provisioning a fragment by itself” issue is 
something that should be addressed? If so, how would you suggest we proceed 
from here?

Thanks,
Bree


On May 25, 2016, at 1:45 PM, Marcel Offermans 
<[email protected]<mailto:[email protected]>> wrote:

Hello Bree,

On 23 May 2016 at 19:16:01, Bree Van Oss 
([email protected]<mailto:[email protected]>) wrote:

Quick question regarding fragment bundles. We have a fragment that contains an 
artifact which uses a separate build process. We are using fragments as a way 
to loosely couple the artifact from the code the operates on it. When I 
provision the fragment to the target by itself (host bundle already 
provisioned) ACE is unable to resolve the bundle fragment and it stays in the 
“installed” state. When I restart ACE agent, the bundle fragment correctly 
makes it into the “resolved” state.

The problem you are running into is that a fragment will only “attach” to a 
host if it is present when the host is resolved. In this case the host was 
already active, so installing the fragment does nothing.

This feels like a bug to me, but wanted to put this out to the list to see if 
anyone has another explanation.

It’s a corner case we don’t explicitly cater for (from memory we do detect that 
a bundle *is* a fragment and we don’t bother to start it, as that won’t work 
either). But yes, this is a bug, probably needs to be fixed in Felix.

Please try restarting the host manually to confirm that this indeed resolves 
your issue. Then our next step is to get this bug recorded and fixed.

Greetings, Marcel

Reply via email to