July 2005 - cddlm-wg - lists.ogf.org

In preparation for the CDDLM/ACS joint session at GGF14
by Keisuke Fukui 19 Jul '05

19 Jul '05

Folks in the cddlm-wg, As you know we will have a joint session at GGF14 between CDDLM and ACS. This is in preparation for the session from acs-wg. After the GGF13 inside the acs-wg, we studied and discussed about the possible interactions between CDDLM and ACS, especially in terms of "File upload" section and AddFile() in the deployment API document. Sequence diagrams in the attachment describe our interpretation about how CDDLM works, and our proposal for the possible interactions in the case that the ACS co-exists in the system. We believe our proposal goes along with what the current set of CDDLM specifications define and doesn't require change in the original definitions. We are looking forward to discuss about this at the joint session for CDDLM/ACS at GGF14. It is very much appreciated if we get responses before the joint session at GGF14 in case that important overlook in our understanding is found. Please feel free to make comments or questions. FYI, At GGF14 joint session, we'd like to start our discussion with this diagram and if we don't find critical issues, we may go down to the detail of the interface and/or advanced interaction. We'd also like to hear requirements on ACS in terms of the event notification and asynchronous invocation of the interfaces, if the time permitting. Thanks in advance for you efforts on this! Best Regards, Keisuke Fukui ACS-WG

2 9

OGSA-WG and fellow WG August '05 F2F Survey
by Hiro Kishimoto 12 Jul '05

12 Jul '05

Hi all, OGSA-WG and fellow WGs are planning to have August '05 F2F meeting. A schedule in week of August 29 at San Diego was proposed at GGF14. However, I got several requests for another date (week of August 15) and/or city (SF bay area). Thus I took action item to ask all expected attendees for their availability for F2F meeting. Please take the time to fill in the following online survey by this Friday (July 8) or at latest next Monday (July 11). We will decide date and city early next week. Post the survey on a Web site - use this URL: http://www.zoomerang.com/survey.zgi?p=WEB224GECBLCHX Thanks, -- Hiro Kishimoto

1 2

Notes from CDDLM sessions at GGF14
by Milojicic, Dejan S 07 Jul '05

07 Jul '05

---------------------------------------------------------------- CDDLM #1 Reference implementations, note taker Stuart Schaefer Reference Implementations - How do we go about it? Presentations from HP Labs, NEC, HP-Brazil and Softricity HP Labs - Steve Loughran - Parser mostly complete - cdl:extends only one level deep - Converts CDL to graph, then downconverts to SmartFrog - WSRF/WSDM o Specifications in flux o Postponed touching until the end - hopefully things will stabilize - Created many test cases - these test cases need validation NEC - Jun Tatemura - Not finished with client - Not finished with WSDM - Need to implement Component Model - Need to do conformance testing for CDL Parser - Want to provide open source implementation plus front-end application for Business Grid HP-Brazil - Building a full reference implementation - A joint effort between HP, Intel, and University of Brazil Softricity - Will have complete implementation, non-conformant - Complete by end of August - Will work on interop after September Schedule for reference implementations - Jun will compile conformance test for CDL v0.1 for mid-August ACS Integration / Compatibility not likely due to scheduling ---------------------------------------------- CDDLM #2, 6/29, 2-3:30pm, Windsor, Interop interop demo: opinions? what kind of applications should be demoed? interop level of CDL? CDDLM framework? interop of CDL parsers - is that a part of conformance? - if conformant, they should interoperate - do we need to demonstrate as interop? CDL is a unit test, deployment API is a system test - component model is a functional test = walkthrough of component model functionalities only two touch points we can verify - client perspective - component can be handled in different implementations invalid CDL case can be also tested within Deployment API as fault in initialize() support environment interoperability - fault if not deployable test 1: one single component on one single machine a component has one query-able property - walk through lifecycle state transition + query properties test 2: multiple component - lazy reference resolution - test semantics (e.g. ordering of deployed components) test 3: cascaded deployment? --> not included what is not covered: deployment of a system with multiple components across multiple implementations (i.e., sharing one deployment graph on multiple platforms) what is cascaded deployment? - a deployed component is also a client to another system --> not be included assumption: same client deploys to different CDDLM implementations assumption: interop includes WSRF/WSDM interoperability assumption: do a number of conformance test that - succeed (with predictable outcome) - fail capture outcome in a log and compare logs for different reference implementations? --> no --> deploy same executables witch are self-checking NUnit/JUnit on sourceforge, aim is to automate interop testing define test components test case document - three parts individual test drafts by the mid-august the test case document aims for GGF15 ---------------------------------------- Notes from CDDLM #4 session, demos Note taker Dejan Milojicic. Jun presents the CDL demo, which also includes the component model (initialization, starting, terminating, and destroying) the application (JBOSS and MySQL). At the moment, there is a separate component model, but the plan is to eventually switch to Stuart's component model. However, Jun used pieces of Steve's deployment APIs. Tom asked to be walked though some of the CDLs. Dejan asked if Jun believes that the version with component model would be easier to develop. This is TBD. Steve presented CDL parser with a number of Junit tests. He then did a demo of starting a Web server. Steve has then gone over the code. He did not implement either component model or deployment APIs, but rather did a hack to interface to SmartFrog which deployed the Web server. This was possible to do only because of the same semantics between CDL and SmartFrog. Conclusions about the conversions between the two languages: there is a possibility to do a small amounts of compatibility ------------------------------------------ CDDLM Session with WSDM was walk over the deployment APIs and then component model ------------------------------------------ CDDLM session with other WG has not happened in a coherent fashion since a couple of chairs dropped by, we instead used that time to resync on our commitments. We never held the last session on the specs because there was no need for it. Thanks, Dejan.

1 0

EMS and CDDLM
by Milojicic, Dejan S 06 Jul '05

06 Jul '05

Hi, We just held the meeting between representatives of EMS of OGSA and CDDLM-WG with the objective to identify the role of CDDLM in the OGSA. Attendees: Steve Loughran, Mark Morgan, Ayla, Jun Tatemura, Chris, Hiro Kishimoto, Keisuke Fukui, Stuart Schaefer, Michael Behrens, Fred Maciel. Agenda and notes: - introductory presentation on CDDLM (intro, CDL, component model, deployment APIs), Hiro will post it into the GridForge shortly. - discuss how to proceed: CDDLM reads architecture document (EMS section) on deployment and configuration work - plan F2F, one day in the week of August 29th in CA, San Diego Supercomputing center, or Santa Clara. The goal is to identify how does CDDLM fit with provisioning, resource allocation, etc. how to start activity on the Grid. There are currently such diagrams in the OGSA architecture specification, as well as in the CDDLM Foundation document. Next phone conference will take place on the 3rd of August, prior to that we'll start email exchange, check boxes on resource allocation and how CDDLM fits them (resource scheduling). Thanks, Dejan Milojicic, co-Chair CDDLM-WG.

1 0

FW: Meeting to discuss CDDLM and OGSA (EMS) in more detail
by Milojicic, Dejan S 06 Jul '05

06 Jul '05

-----Original Message----- From: Milojicic, Dejan S Sent: Wednesday, July 06, 2005 1:59 PM To: 'Subramaniam, Ravi'; 'csmith(a)platform.com'; 'behrens(a)R2AD.com'; 'Keisuke Fukui (on the road)'; 'Fred.Maciel(a)hds.com'; 'mmm2a(a)cs.virginia.edu' Cc: 'Jun Tatemura'; Loughran, Steve; 'Stuart Schaefer'; 'Hiro Kishimoto'; 'Andrew Grimshaw' Subject: RE: Meeting to discuss CDDLM and OGSA (EMS) in more detail Hi All, Attached is the presentation. The session will start in a a couple of minutes. Thanks, Dejan. > -----Original Message----- > From: Milojicic, Dejan S > Sent: Tuesday, July 05, 2005 7:03 AM > To: 'Subramaniam, Ravi'; 'csmith(a)platform.com'; 'behrens(a)R2AD.com'; > 'Keisuke Fukui (on the road)'; Fred.Maciel(a)hds.com; > 'mmm2a(a)cs.virginia.edu' > Cc: Jun Tatemura; Loughran, Steve; Stuart Schaefer; 'Hiro Kishimoto'; > 'Andrew Grimshaw' > Subject: Meeting to discuss CDDLM and OGSA (EMS) in more detail > > Hi Chris, Fred, Mark, Mike, Keisuke, and Ravi, (cc-ed to Andrew, Hiro, > key CDDLM members, and CDDLM mailing list) > > Per the OGSA meeting at GGF, you have been identified as a potential > candidates to discuss how CDDLM (Configuration Description, Deployment > and Lifecycle Management) fits into the broader OGSA picture and in > particular with EMS. > > We have agreed on the following course of actions. We will meet > tomorrow (Wed 7/6, at 2pm PST) and briefly go over the CDDLM so that > you get more familiar with it. We'll have intro and 3 slides per spec > (language, component model, and deployment API). We will identify some > key areas that need addressing integration of CDDLM into OGSA and then > we shall plan the F2F some time in August/September time frame. > > Can you please confirm that you will be able to participate in the > meeting tomorrow. The meeting is open to all CDDLM members. > > Thanks, > > Dejan. > > We shall use the following conference number: > > +1 866-639-4732 (toll free while in US) > Code 2362906 (followed by a '#' sign) > +1 574-948-0379 (toll & international)

1 0

Updated Component Model Specification
by Stuart Schaefer 06 Jul '05

06 Jul '05

Dejan and team, Included are the updates to the Component Model document to resolve the typos and errata found by Ayla's team. I have also enclosed updated wsdl and schema documents as they have changed. They are all up to date in Source Forge. There are many updates. I apologize for the scope of these updates at this late date. 1) Flow control elements can take a Boolean value to help them be overridden if inherited. 2) Delegates can also (Jun's request). 3) Event registrations as well. 4) Two new data types are used to simplify the use of these values. Stuart

1 0

Deployment API final release
by Steve Loughran 06 Jul '05

06 Jul '05

This is the final release; we believe that Muse is happy with and all is well. I'm submitting the PDF for standardisation, not the .doc file, because -the doc file has embedded visio content -even tools that support .doc have fun with complex bits All the attached xml docs are the actual files that have been pasted into the document. I will tag everything in CVS shortly. steve

1 0

Notes from the 7/6 meeting
by Milojicic, Dejan S 06 Jul '05

06 Jul '05

Meeting on 7/6/2005, Attendees: Ayla, Jun, Stuart, Steve, and Dejan. Agenda ------ GGF14: We gave an update, everyone seems to be happy with the progress (Stuart, Jun, and Dejan) Ayla asked if WSDM in the repository is the changed one Meeting today with EMS: Dejan will briefly present overview then everyone will provide a quick summary of his own area. Thereafter we'll agree how to best proceed with the integration of CDDLM with OGSA roadmap and other Reference implementation: Ayla asked a number of detailed questions about component model. Stuart will follow up on some of the questions. Update on plans: Dejan asked a few lines of plans from Stuart/Steve/Jun/Ayla from now until GGF16. It should be no more than 5-10 lines with the task item and the time by which it will be completed. Examples include: reference implementation completed, functional testing performed, interoperability testing, etc. Future meetings: no meetings until 8/3. If there is a need, you should arrange on your own or send me an email that I facilitate through someone else. Documents: Stuart will update the component model and send it to Dejan. He will start from the document uploaded on GridForge (Stuart please let me know if you need the URL). Steve will send the Deployment APIs before the end of the day. Thanks, Dejan.

1 0

RE: [cddlm] Doubts about ComandPath
by Stuart Schaefer 06 Jul '05

06 Jul '05

Guilherme, I checked in a change to implement this correctly, by adding minOccurs and maxOccurs specifications to the sequence. Stuart -----Original Message----- From: Stuart Schaefer Sent: Tuesday, July 05, 2005 6:07 PM To: Guilherme Mauro Germoglio Barbosa / Projeto Ourgrid - Projeto Ourgrid/LSD Cc: Marcelo Iury Oliveira / Projeto Ourgrid - Prof. Walfredo; cddlm-wg(a)ggf.org Subject: RE: [cddlm] Doubts about ComandPath Guilherme, You are actually right. The sequence element is missing a minOccurs and maxOccurs specification. It is supposed to be optional, not required. This will be fixed. Stuart -----Original Message----- From: Guilherme Mauro Germoglio Barbosa / Projeto Ourgrid - Projeto Ourgrid/LSD [mailto:guiga@dsc.ufcg.edu.br] Sent: Tuesday, July 05, 2005 4:31 PM To: Stuart Schaefer Cc: Marcelo Iury Oliveira / Projeto Ourgrid - Prof. Walfredo; cddlm-wg(a)ggf.org Subject: Re: [cddlm] Doubts about ComandPath Stuart, If this example is correct, the component model schema has problems defining the <cmp:CommandPath> element as it must present a <cmp:path> element and zero or more <cmp:args> elements. Defined by: / <xsd:element name="CommandPath" type="commandPathType"> <xsd:annotation> <xsd:documentation>The location or object within the assets of the component to use in activating the component.</xsd:documentation> </xsd:annotation> </xsd:element> [...] <xsd:complexType name="commandPathType" mixed="true"> <xsd:annotation> <xsd:documentation>A command path can contain a command and a list arguments.</xsd:documentation> </xsd:annotation> <xsd:sequence> <xsd:element name="path" type="xsd:anyURI"/> <xsd:element name="args" type="xsd:anyType" minOccurs="0" maxOccurs="unbounded"/> </xsd:sequence> </xsd:complexType>/ I am correct? The other example from Iury's mail follows this schema. Is this schema the same you are refering to? Guilherme Stuart Schaefer wrote: >Iury, > >I think you missed a bit of the documentation. The example you are >looking at does not include the <cmp:CodeBase> element. > ><WebApplication> > <cmp:CodeBase>http://server/file.jar</cmp:CodeBase> > <cmp:CommandPath>com.exns.ApacheDeployer</cmp:CommandPath> > <port>80</port> > <hostname>www.example.org</hostname> ></WebApplication> > >The command path is supposed to be used to locate the actual executable >instance, class file, etc... The command path is declared normatively >in the schema as mixed content, which allows inline content. This is >done in lieu of an <xs:any> element, which could cause problems with >inheritance. > >I will look in the Component Model document, and see if this can be >clarified. > >Regards, > >Stuart > >-----Original Message----- >From: Marcelo Iury Oliveira / Projeto Ourgrid - Prof. Walfredo >[mailto:iury@lsd.ufcg.edu.br] >Sent: Tuesday, July 05, 2005 2:33 PM >To: cddlm-wg(a)ggf.org >Subject: [cddlm] Doubts about ComandPath > >Hello all! > >I'm developing a cddlm engine with Ayla, and now we are writing a >factory that deploys and instantiates a given deployment component, >using a CodeBase and a CommpandPath. >Looking at the component model specification, I got a little bit >confused about the CommandPath, specially considering the given example, > >which is in the following: >... ><cmp:CommandPath> > <cmp:path>http://server.com/app.jar</cmp:path> > <cmp:args>...</cmp:args> * ></cmp:CommandPath> >... > >As we have arguments, I think we need a "command" element (that can be >optional) to pass these arguments. As we do not have the "command" >element, we thought it could be in the "path" element, or we could have >there executable files, but in the example presented we have a .jar >file, which is not executable by itself. > > >Can anyone help us in making this more clear? > > Iury > > > >

1 0

RE: [cddlm] Doubts about ComandPath
by Stuart Schaefer 05 Jul '05

05 Jul '05

Guilherme, You are actually right. The sequence element is missing a minOccurs and maxOccurs specification. It is supposed to be optional, not required. This will be fixed. Stuart -----Original Message----- From: Guilherme Mauro Germoglio Barbosa / Projeto Ourgrid - Projeto Ourgrid/LSD [mailto:guiga@dsc.ufcg.edu.br] Sent: Tuesday, July 05, 2005 4:31 PM To: Stuart Schaefer Cc: Marcelo Iury Oliveira / Projeto Ourgrid - Prof. Walfredo; cddlm-wg(a)ggf.org Subject: Re: [cddlm] Doubts about ComandPath Stuart, If this example is correct, the component model schema has problems defining the <cmp:CommandPath> element as it must present a <cmp:path> element and zero or more <cmp:args> elements. Defined by: / <xsd:element name="CommandPath" type="commandPathType"> <xsd:annotation> <xsd:documentation>The location or object within the assets of the component to use in activating the component.</xsd:documentation> </xsd:annotation> </xsd:element> [...] <xsd:complexType name="commandPathType" mixed="true"> <xsd:annotation> <xsd:documentation>A command path can contain a command and a list arguments.</xsd:documentation> </xsd:annotation> <xsd:sequence> <xsd:element name="path" type="xsd:anyURI"/> <xsd:element name="args" type="xsd:anyType" minOccurs="0" maxOccurs="unbounded"/> </xsd:sequence> </xsd:complexType>/ I am correct? The other example from Iury's mail follows this schema. Is this schema the same you are refering to? Guilherme Stuart Schaefer wrote: >Iury, > >I think you missed a bit of the documentation. The example you are >looking at does not include the <cmp:CodeBase> element. > ><WebApplication> > <cmp:CodeBase>http://server/file.jar</cmp:CodeBase> > <cmp:CommandPath>com.exns.ApacheDeployer</cmp:CommandPath> > <port>80</port> > <hostname>www.example.org</hostname> ></WebApplication> > >The command path is supposed to be used to locate the actual executable >instance, class file, etc... The command path is declared normatively >in the schema as mixed content, which allows inline content. This is >done in lieu of an <xs:any> element, which could cause problems with >inheritance. > >I will look in the Component Model document, and see if this can be >clarified. > >Regards, > >Stuart > >-----Original Message----- >From: Marcelo Iury Oliveira / Projeto Ourgrid - Prof. Walfredo >[mailto:iury@lsd.ufcg.edu.br] >Sent: Tuesday, July 05, 2005 2:33 PM >To: cddlm-wg(a)ggf.org >Subject: [cddlm] Doubts about ComandPath > >Hello all! > >I'm developing a cddlm engine with Ayla, and now we are writing a >factory that deploys and instantiates a given deployment component, >using a CodeBase and a CommpandPath. >Looking at the component model specification, I got a little bit >confused about the CommandPath, specially considering the given example, > >which is in the following: >... ><cmp:CommandPath> > <cmp:path>http://server.com/app.jar</cmp:path> > <cmp:args>...</cmp:args> * ></cmp:CommandPath> >... > >As we have arguments, I think we need a "command" element (that can be >optional) to pass these arguments. As we do not have the "command" >element, we thought it could be in the "path" element, or we could have >there executable files, but in the example presented we have a .jar >file, which is not executable by itself. > > >Can anyone help us in making this more clear? > > Iury > > > >

1 0