UvA Web University of Amsterdam - UvA
Faculty of Science
WS-VLAM
Home

Documentation

Publications

Presentations

FAQ

Applications

Demos

Distribution

Ongoing development

Developers

VL-e Links

WS-VLAM on the PoC ...

=========================================================================

content

Bullet what the PoC? [Details]

Bullet How is the WS-VLAM is deployed on the PoC infrastructure? [Details]

Bullet How to save files on remote storage systems? [Details]

Bullet How to check logs on the worker nodes using globus tools? [Details]

========================================================================

Bullet What is the PoC? [Table of Contents]

The poc infrastructure is composed of a number of clusters located either at SARA or NIKHEF [Poc Web page] . Each cluster is running the PoC distribution which contains a set of third party tools and software developed within the VL-e project. The current PoC distribution is PoC-R3 [Poc-R3]

Here is quick start for the new PoC users [Poc quick start]

To be able to use the WS-VLAM on PoC infrastucture you need:

  1. to belong to Virtual Organisation. To become a member of VO just follow the instructions in the following link [VO_membership]

  2. to be added to grid-mapfile on the Service node (ws1.grid.sara.nl). Send a request to the PoC help [PoC-help-page]

  3. to have a login on a UI node of the PoC. This will help you to do some debugging of workflow when needed Send a request to the PoC help desk [PoC-help-page]

  4. you have to know the name of the Service Element (SE) where you can submit your workflow components. From your account on the UI node , you can use the following lcg command to get the available SE:
    $ lcg-infosites --vo your-vo-name closeSE

    Note: the name of hthe queues for globus is bit different, if you get:

    ...
    Name of the CE: gazon.nikhef.nl:2119/jobmanager-pbs-qlong
    tbn18.nikhef.nl
    srm.grid.sara.nl
    ...

    then you should set the hostname of the workflow component in the WS-VLAM composer to

    gazon.nikhef.nl:2119/jobmanager-pbs -q qlong

========================================================================

Bullet How WS-VLAM is deployed on the PoC infrastucture? [Table of Contents]

WS-VLAM web services:

  1. The two WS-VLAM services are running the service node (ws1.grid.sara.nl). So when you start the WS-VLAM composer, you have to set the correct location of the service. This can be done via the Options menu item in the File menu of the WS-VLAM composer

  2. RTSM service URL should be set to https://ws1.grid.sara.nl:8443

  3. RM service URL should be set to https://ws1.grid.sara.nl:8443

WS-VLAM repository:

  1. The WS-VLAM repository is a location where you have packed your workflow modules as tar files to be deployed automatically at runtime on the worker nodes. This location could be on any host where a gsiftp is running. You can define the URL of the repository using the option item in the File menu of the WS-VLAM composer

  2. Repository URL should be set to the location where you have saved your workfow components (example: gsiftp://fs2.das3.science.uva.nl/home/vleroot/wsvlam/repository)

  3. NOTE: modules written in C/C++ have to be compiled on the PoC instructure (use your UI login for that). For modules written in java, or any other scription languages you need to check if the scripting language is supported by the PoC infrastructure and the available PoC version is compatible with the one you used to develop your modules.

========================================================================

Bullet How to save the files generated by the workflow? [Table of Contents]

  • WS-VLAM provide a standard module called FileWriter, you can use this module to save files on remote storage resources using various protocols gsiftp, ftp, http, and srb. In the parameter of the fileWriter module sepecify the location and the name of the file to be created

    For example to created a file REMOVE-ME on the fs2.das3.science.uva.nl using gsiftp put in the parameter value of the fileWriter the following line:

    gsiftp://fs2.das3.science.uva.nl/tmp/REMOVE-ME

=======================================================================

Bullet How to check logs on worker nodes using globus tools ? [Table of Contents]

  • login on the ui.grid.sara.nl

  • you need to know the name of the host on which you have submitted your workflow. This information can be retreive from the WS-VLAM composer in the Module panel of any module composing your workflow (if for example the host name is set to:
    gazon.nikhef.nl:2119/jobmanager-pbs -q qshort
    then in the following commands you need to replace host-name with gazon.nikhef.nl

  • to see what is your current home dir on the worker nodes use:
    $ globus-job-run host-name /bin/pwd

  • to see the file created on the worker nodes use:
    $ globus-job-run host-name /bin/ls -la

  • to check if your workflow components have been deployed at runtime on the worker nodes use:
    $ globus-job-run host-name /bin/ls .wsvlam/modules/ -la

  • to see the content of the file on the worker node use:
    $ globus-job-run host-name /bin/cat path/name-of-file

==================================================================


University of Amsterdam UvA
Kruislaan 403 NL-1098 SJ
Amsterdam, The Netherlands
Tel: +31(0)20 525 7514
Fax: +31 (0)20 525 7490
gvlam-dev_no-spam_@lists.vl-e.nl

Last update:
25 March 2008