[cddlm] Notes

[Steve Loughran]

Milojicic, Dejan S (HP Labs) wrote:

> 
> Stuart: Focusing internally on cleanup of the implementation that is not
> working. Still needs some time to sync-up with Jun. Still can't parse
> all test cases and when they fix some of them other problems pop up.
> Deployment APIs is up and running. 20% passing. Started testing on other
> end-point with others, but gave up, still focusing on internal. Still
> not have endpoints, working with IT to make it happen. For the next
> week, if he can not digest CDL, he will post it to Jun and the group.

Stuart, if you want access to the logs of my run, I can help there. But 
focus on internal first.


> Satish: component model and CDL complete 100%. Received CDL from
> Guilherme and will complete tests against UFCG implementation. 30% have
> been completed against HP's portal, but non component model so far.
> Regarding the system load on Tomcat, after creating 8 or 9 systems, the
> JVM can not take any more load on Tomcat and after initializing 3 or 4
> systems, the system has been fully loaded. Need to work on how to
> resolve this issue. 

Satish,

Do you want me to pause in my client too? I could add a configurable 
sleep between postings.

I think my unit tests all do a <wsrf:Destroy /> afterwards, the 
assumption being that will trigger cleanups. But I think I will 
ultimately tweak my test server to do a cleanup of deployments after a 
well-defined period. Alternatively, using the deployment api, a client 
app could enum all deployed systems and delete them one by one. I don't 
think I would do that on my system as one deployed application is the 
deployment endpoint itself.

-Steve