Notes of the meeting on 8/3/2005
Attendees: Steve, Ayla, Dejan. 1. Update on the progress with reference implementations Ayla: did a parser and service, but Tomcat dies after sequential deployments. Working in parallel to finish service, have prototype of the portal of the system, but only using the properties of Muse it works. There are some problems with updating properties. They are implementing more tests. Still confident that the first prototype will be done by end off August. Steve suggested using Jetty instead of Tomcat. Steve: Hasn't done much since GGF because of other commitments. Has completed requirements document and put into the CVS. No information from Jun and Stuart. Hopefully they will join in next week. 2. Preparation for the meeting with OGSA Steve will join (potential 3. Defining agenda for the F2F with OGSA (I need to provide it to Hiro shortly) A) Introduction (learn more about EMS broader picture, in the past wee overviewed CDDLM) B) Defining architectural relationship C) Indebt overview of the CDDLM interfaces (if needed) D) Integration roadmap E) Next steps 4. Planning next activities Meeting are back on the weekly basis. Thanks, Dejan.
Hi Dejan and Team, I am sorry I was absent because I was confused with Today's meeting with OGSA... I wanted to ask if we can give some Grid deployment usecases to OASIS SDD TC (they are now aggregating use cases to identify requirements). If it is okay, I will give you an initial draft in our next meeting as well as a status report on our reference implementation. I will be joining a conference call with OGSA today (although I may be late). Best Regards, Jun Tatemura Milojicic, Dejan S wrote:
Attendees: Steve, Ayla, Dejan.
1. Update on the progress with reference implementations
Ayla: did a parser and service, but Tomcat dies after sequential deployments. Working in parallel to finish service, have prototype of the portal of the system, but only using the properties of Muse it works. There are some problems with updating properties. They are implementing more tests. Still confident that the first prototype will be done by end off August. Steve suggested using Jetty instead of Tomcat.
Steve: Hasn't done much since GGF because of other commitments. Has completed requirements document and put into the CVS.
No information from Jun and Stuart. Hopefully they will join in next week.
2. Preparation for the meeting with OGSA
Steve will join (potential
3. Defining agenda for the F2F with OGSA (I need to provide it to Hiro shortly)
A) Introduction (learn more about EMS broader picture, in the past wee overviewed CDDLM) B) Defining architectural relationship C) Indebt overview of the CDDLM interfaces (if needed) D) Integration roadmap E) Next steps
4. Planning next activities
Meeting are back on the weekly basis.
Thanks,
Dejan.
Hi Dejan and all, we have analyzed the problems with Tomcat and we've concluded that in fact our tests were stressing it, because we had at about 30 deployments and undeployments of 11MB each, one after the other in a set of 94 tests. We've reduced each component size verifying common libs from Muse that could be removed from each deployed component and it worked. Maybe the same problem can happen with Jetty, but we had a meeting yesterday and concluded that a lot of code, including unit tests would need to be rewritten and we would not be able to have a prototype at the end of this month if we make this change. Maybe in a future version we can use Jetty. In relation to the update of properties, it is working well using Muse, we are just implementing more JUnit tests that verify the integration of the CDL parser with the module that deals with the update of properties in components. We hope to have the first prototype by the end of the month, but we fear not to have enough time to assure that all the tests from the Test Plan sent by Steve are passing, because we are suposing Muse will actually provide some of the functionality that is demanded there and won't give us problems difficult to find as the ones we had when the specification needed adjustments. Ayla Milojicic, Dejan S wrote:
Attendees: Steve, Ayla, Dejan.
1. Update on the progress with reference implementations
Ayla: did a parser and service, but Tomcat dies after sequential deployments. Working in parallel to finish service, have prototype of the portal of the system, but only using the properties of Muse it works. There are some problems with updating properties. They are implementing more tests. Still confident that the first prototype will be done by end off August. Steve suggested using Jetty instead of Tomcat.
Steve: Hasn't done much since GGF because of other commitments. Has completed requirements document and put into the CVS.
No information from Jun and Stuart. Hopefully they will join in next week.
2. Preparation for the meeting with OGSA
Steve will join (potential
3. Defining agenda for the F2F with OGSA (I need to provide it to Hiro shortly)
A) Introduction (learn more about EMS broader picture, in the past wee overviewed CDDLM) B) Defining architectural relationship C) Indebt overview of the CDDLM interfaces (if needed) D) Integration roadmap E) Next steps
4. Planning next activities
Meeting are back on the weekly basis.
Thanks,
Dejan.
Hello, we are assuming something here about the properties of a component and we want to know if there is any problem in this assumption: we have noticed that the Muse implementation of WSRF only allows the inclusion of a property previously defined in the WSDL that describes the component. So, if in the CDL we declare a property and its value and this property is not included in the properties of the component (declared in the schema referenced by the WSDL or in the WSDL itself in the types part), we get an error indicating that the property is unknown. Is this the expected behavior? So, if the user wants to configure a property through the CDL, this property must be defined in the schema, am I right? Thanks for your attention, Ayla
participants (3)
-
Ayla Debora Dantas de Souza - Projeto Ourgrid -
Jun Tatemura -
Milojicic, Dejan S