Hi, The xml example is made from information in RDB tables used by other NAREGI services, for example, by the meta-scheduler to find candidate set of resources for job execution. I think that our basic requirements are met in current glue 2 except for our particular needs in terms of advanced resource reservation and job monitoring. I'm in Japan for the preparation for new version of our middleware and I'll look at the new version of GLUE2.xsd and change the translator for new glue2 and NAREGI version in a month. Yuji Sergio Andreozzi wrote:

Hello Yuji,

Yuji Saeki ha scritto:

...

I tried to make a glue2 xml example of our NAREGI resource/service description based on the recent glue2.xsd (rev1240) linked from the glue group's wiki page.

The attached file is the output of a JAXB application program which retrieves information about our middleware test environment ( 5 small clusters) from NAREGI information service, translates it from current NAREGI schema to the glue2 draft schema and makes an xml document according to the glue2.xsd. I haven't applied it to the grid of multiple admin domains yet.

thanks for this implementation experience. This is really valuable for validating what we are doing. There is a new version of the XML rendering available in the repository, you may want to have a look.

As regards, your experience, can you comment on the suitability of current GLUE for the NAREGI needs?

BTW, are you at OGF? If so, we can talk about it.

Cheers, Sergio

...

regards, Yuji Saeki

Balazs Konya wrote:

...

Hi,

Recently, NorduGrid started to look at the potential xml rendering of the current glue2 draft. We tried to follow INFN's initial attempt (linked from the glue group's wiki page: http://forge.ogf.org/sf/wiki/do/viewPage/projects.glue-wg/wiki/GLUE2XMLSchem...), nevertheless we run into several questions.

Please find our questions/suggestions collected on a newly created wiki page:

http://forge.ogf.org/sf/wiki/do/viewPage/projects.glue-wg/wiki/FeedbackGLUE2...

Sergio, I suggest to go through our items during one of the scheduled phone meetings.

Tomorrow or the day after i'll make our glue2 xml examples of a homogeneous cluster, desktop machine, a multi-service computer box available too (i have to clean them a bit, first).

Meanwhile i encourage every potential glue2 users that they start putting together small XML/LDAP/SQL examples of their resource/service descriptions. The glue2 draft has reached that level it is feasible to start creating examples. This is the best way to provide feedback to the glue group.

regards, Balazs Konya _______________________________________________ glue-wg mailing list glue-wg@ogf.org http://www.ogf.org/mailman/listinfo/glue-wg

------------------------------------------------------------------------

_______________________________________________ glue-wg mailing list glue-wg@ogf.org http://www.ogf.org/mailman/listinfo/glue-wg

-- Sergio Andreozzi INFN-CNAF, Tel: +39 051 609 2860 Viale Berti Pichat, 6/2 Fax: +39 051 609 2746 40126 Bologna (Italy) Web: http://www.cnaf.infn.it/~andreozzi

Hi, These are our particular needs in terms of (1) Advance reservation (and bulk job submission): NAREGI grid-level scheduler can make advanced reservation of computing resources in the job execution management. Users can submit their jobs without resource reservation as well as with it, where not all of LRMS are equipped with advanced reservation function. So the NAREGI grid-level scheduler has to know which of batch queues can accept advanced reservation requests or don't accept jobs without reservation (reservation necessary). And it also supports bulk job submission, where not all of LRMS are equipped with array job submission function. So we require our information model to include the propterties: Reservability and ReservasionNecessary to a batch queue, BulkSubmissionAvailable to a LRMS. We can deal with them as a NAREGI specific extension and put them in the key-value pair. (2) Job monitoring: When an application program, which consists of multiple components, starts to execute on computer systems belonging to multiple sites in NAREGI environment, each of the components has to get information about the systems where other components run to transfer data among them. Yuji Sergio Andreozzi wrote:

Hi Yuji,

Yuji Saeki ha scritto:

...

The xml example is made from information in RDB tables used by other NAREGI services, for example, by the meta-scheduler to find candidate set of resources for job execution. I think that our basic requirements are met in current glue 2 except for our particular needs in terms of advanced resource reservation and job monitoring.

it would be interesting for the group to have more details on what is not covered, specifically advance reservation and job monitoring. Can you provide more details?

...

I'm in Japan for the preparation for new version of our middleware and I'll look at the new version of GLUE2.xsd and change the translator for new glue2 and NAREGI version in a month.

this is still subject to changes. I would recommend you to wait for the end of the month before starting a new revision.

Cheers, Sergio

...

Yuji

Sergio Andreozzi wrote:

...

Hello Yuji,

Yuji Saeki ha scritto:

...

I tried to make a glue2 xml example of our NAREGI resource/service description based on the recent glue2.xsd (rev1240) linked from the glue group's wiki page.

The attached file is the output of a JAXB application program which retrieves information about our middleware test environment ( 5 small clusters) from NAREGI information service, translates it from current NAREGI schema to the glue2 draft schema and makes an xml document according to the glue2.xsd. I haven't applied it to the grid of multiple admin domains yet.

thanks for this implementation experience. This is really valuable for validating what we are doing. There is a new version of the XML rendering available in the repository, you may want to have a look.

As regards, your experience, can you comment on the suitability of current GLUE for the NAREGI needs?

BTW, are you at OGF? If so, we can talk about it.

Cheers, Sergio

...

regards, Yuji Saeki

Balazs Konya wrote:

...

Hi,

Recently, NorduGrid started to look at the potential xml rendering of the current glue2 draft. We tried to follow INFN's initial attempt (linked from the glue group's wiki page: http://forge.ogf.org/sf/wiki/do/viewPage/projects.glue-wg/wiki/GLUE2XMLSchem...), nevertheless we run into several questions.

Please find our questions/suggestions collected on a newly created wiki page:

http://forge.ogf.org/sf/wiki/do/viewPage/projects.glue-wg/wiki/FeedbackGLUE2...

Sergio, I suggest to go through our items during one of the scheduled phone meetings.

Tomorrow or the day after i'll make our glue2 xml examples of a homogeneous cluster, desktop machine, a multi-service computer box available too (i have to clean them a bit, first).

Meanwhile i encourage every potential glue2 users that they start putting together small XML/LDAP/SQL examples of their resource/service descriptions. The glue2 draft has reached that level it is feasible to start creating examples. This is the best way to provide feedback to the glue group.

regards, Balazs Konya _______________________________________________ glue-wg mailing list glue-wg@ogf.org http://www.ogf.org/mailman/listinfo/glue-wg

------------------------------------------------------------------------

_______________________________________________ glue-wg mailing list glue-wg@ogf.org http://www.ogf.org/mailman/listinfo/glue-wg

-- Sergio Andreozzi INFN-CNAF, Tel: +39 051 609 2860 Viale Berti Pichat, 6/2 Fax: +39 051 609 2746 40126 Bologna (Italy) Web: http://www.cnaf.infn.it/~andreozzi

_______________________________________________ glue-wg mailing list glue-wg@ogf.org http://www.ogf.org/mailman/listinfo/glue-wg

-- Sergio Andreozzi INFN-CNAF, Tel: +39 051 609 2860 Viale Berti Pichat, 6/2 Fax: +39 051 609 2746 40126 Bologna (Italy) Web: http://www.cnaf.infn.it/~andreozzi

Hi,

...

(2) Job monitoring: When an application program, which consists of multiple components, starts to execute on computer systems belonging to multiple sites in NAREGI environment, each of the components has to get information about the systems where other components run to transfer data among them.

this is recorded in the following tracker: http://forge.ogf.org/sf/go/artf6161 the use case is not enough clear to us. If we understand correctly, you consider the fact that a set of interrelated tasks (e.g., workflow) run in different computing services.

Yes, it is a use case about communication among co-allocated tasks executed with resource reservation.

What are the required information that you need to discover in order to perform the data transfer among them?

The properties we need are global job id, VO name, application name(, executable) and URL of communication buffer (list of host names and directory paths where a task puts files for another tasks). For example in the XML format, <?xml version="1.0" encoding="UTF-8"?> <Reservation xmlns="http://www.naregi.org/infoservice/namespaces/sbc"> <Job> <GlobalJobId>34fcb207-0814-4e9e-8cc5-7ec3ed49a05a</GlobalJobId> <SubmittingUserName>/C=JP/O=National Research Grid Initiative/OU=CGRD/CN=XXX YYY</SubmittingUserName> <VOAttributeName>vo0</VOAttributeName> <JobType>MPI</JobType> <Executable>./bin/mode</Executable> </Job> <Site> <TargetHost>pbg1044.naregi.org</TargetHost> <IMPIClientId>-1</IMPIClientId> <SitesCoallocated>3</SitesCoallocated> </Site> <Node> <HostName>pbg1045.naregi.org</HostName> <HostName>pbg1046.naregi.org</HostName> <HostName>pbg1047.naregi.org</HostName> <HostName>pbg1048.naregi.org</HostName> </Node> <Extend> <SynchronousConnection> <Connection protocol="GRID">MD GA</Connection> <Connection protocol="GRID">GA MO</Connection> <Connection protocol="GRID">MO GB</Connection> <Connection protocol="GRID">GB MD</Connection> <Buffer appname="MD" executable="mode">MDMO_NII_beta2/work/sbc</Buffer> <Buffer appname="GA" executable="ga">MDMO_NII_beta2/work/sbc</Buffer> <Buffer appname="MO" executable="mo">MDMO_NII_beta2/work/sbc</Buffer> <Buffer appname="GB" executable="gb">MDMO_NII_beta2/work/sbc</Buffer> </SynchronousConnection> </Extend> </Reservation> We deal with a part of them as an extention in our schema as observed above because such an application is infrequent even in NAREGI. Yuji