Minutes-of-the-17-03-2014-dg-vt-meeting

Minutes of the EGI DesktopGrid Virtual Team teleconference meeting on 03/17/2014

https://wiki.egi.eu/wiki/VT_Promoting_Desktop_Grids

 

Participants:

-          Andrea Manconi

-          Robert Lovas

-          Gergely Sipos (minutes)

-          Muhammad Farhan Sjaugi

-          Tamas Kiss

-          Alexandre Bonvin

-          Jozsef Kovacs

-          Francisco

-          Andrej Zina

-          Todor Gurov

-          Ad Emmen

-          John Kewley

-          Leslie Vers

-          Luciano Milanesi

 

Status updates from the participants:

1)      Alexandre Bonvin (Utrecht University/WeNMR) / Jozsef Kovacs (MTA SZTAKI): WeNMR integration updates

    a)      Haddock portal is sending 1 out of 20 jobs to IDGF. Only this because there is no transparent way to run in EGI and in IDGF

    b)      The use of DGs slows down the execution of Haddock workflows, which is a concern.  (Jozsef: DesktopGrids are not efficient for jobs with short deadline e.g. 6hours)

    c)       WMS does not return the status correctly (Jozsef reported this and has to follow up with WMS experts)

    d)      About 54.000  jobs were run in the last 3 weeks. Out of this 1000 jobs were submitted to EDGeS@home, with ~800 successful completion. Failed jobs were resubmitted. (Notes by Jozsef: on the BOINC side, every job completes successfully. Failed jobs detected by the WeNMR portal comes from two reasons: 1) WeNMR portal sets 6 or 12 hours deadline against BOINC jobs, while BOINC (and the volunteer resources) are not designed for short deadline jobs, therefore there are many deadline miss; 2) The aforementioned gLite-WMS communication problem sporadically causes jobs to remain in ‘running’ state on the WMS however these jobs already in ‘finished’ state on the Computing Element. So the WeNMR portal detects these jobs as ‘running’ forever. Unfortunately, none of the 2 problems can be sorted out by the SZTAKI DG team, so we are asking for help from EGI gLite experts for the WMS issue)

    e)      Worth thinking about connecting DIRAC to IDGF, because the performance through DIRAC seems to be better

    f)       The application can be started to be used in promotion (eg. blog item or other textual description. A video has been prepared recently by Ad Emmen, who is working on a page at the moment where this content will be displayed)

    g)      Tamas and Alexandre should follow up the deployment of a university desktop grid at Utrecht University. This did not get a priority so far

2)      Jozsef Kovacs (MTA SZTAKI): DIRAC integration updates / 1-click demo / new IDGF Operation Center in EGI / others

    a)      Goal: Make DIRAC capable of running under the DG client in a virtualised environment

    b)      BioVeL integration updates

        i)        Two applications have been deployed on the EDGeS@home. They run without gLite, directly through by the BioVeL portal

        ii)       It runs in production level, but the resources are limited, because the application cannot run on windows desktops yet, and the EDGES@Home contains 90% Windows machines

        iii)     Windows versions have arrived, under testing, hopefully finishing before end of March

    c)       work on making IDGF an EGI OC (Operation Center).

        i)        Create a new, international Operations Centre in EGI and put all DG operational activity under this. This would move the accounting of desktop grids from NGI_HU to IDGF OC.

        ii)       Most of the services remain to be operated by SZTAKI (APEL, WMS, ...). A few exceptions are Nagios monitor (by Wigner institute in Hungary).

        iii)     Jozsef just submitted the official request to EGI Operations management to establish the “IDGF” Operations Centre.

3)      Tamas Kiss/Hannu Visti (University of Westminster): Brunel and UoW activities in the UK

    a)      Operate the Westminster DG, and had to carry out

    b)      Brunel DG setup - no progress recently

    c)       Portsmouth University DG has been setup. It's now operational, and now are working on a couple of applications (e.g. a video encoder, Arnold - 3D animation rendering)

4)      Francisco (BIFI):

    a)      They are writing a book on citizen science. Contribution from members of this VT is welcome. Francisco will circulate link to the website for contributors

5)      Muhammad (UTM):

    a)      Two universities in Malaysia deployed DGs

    b)      One is successful and runs applications such as genetic algorithm

    c)       Problem at UPM University: The DG server was deployed from an image, and it crashed, it could not be recovered so had to be rebuilt

    d)      Are planning a summer school and a conference in the autumn. (with hands-on for system admins, and for users) Consider inviting trainers from IDGF

    e)      Put all this information to the 'South-East Asian chapter' on the IDGF website

6)      Luciano and Andrea:

    a)      Connecting desktop grids and GPUs in a single workflow. This could be beneficial for their life science apps (and possibly for others too)

    b)      They use a local DG for testing, will run on a public grid for production execution

7)      Todor:

    a)      They are interested in accessing GPU resources in DGs. Could be used for encryption applications --> Jozsef, Tamas and Todor to organise a separate teleconference on this topic

    b)      Tamas: To test a GPU-DG setup we need an application that we can use. Can Luciano's application or Todor's application can be used for this?

 

Plans:

1)      Ad Emmen (IDGF/AlmereGrid):  EGI campaign plans / DG session & booth plans @ EGI CF

    a)      One session (90 minutes) in the programme: technology; for citizens; applications (to demonstrate the technology). Most focus on the applications --> Contact DIRAC and BioVeL for contribution

    b)      Are currently looking at getting a booth at the forum

    c)       The WeNMR, DIRAC, BioVeL applications are running

    d)      EDGES@Home will be renamed after a famous scientist, donors can vote on the name

    e)      Website of EDGES@Home will be renewed, with more relevance to citizen scientists

    f)       Schedule is not the best, because of clash with NIL meeting (titled EGI Engagement) - Ask a move if you can from Sara Coelho, but beware that some speakers may already arranged their travel.

2)      Robert Lovas (MTA SZTAKI) / Gergely Sipos (EGI.eu):

    a)      1-page promo materials, more detailed plans for the near future

    b)      The document consists of two parts:

        i)        Define the groups in EGI who can benefit from IDGF services and the channels through which they can be reached by IDGF;

        ii)       Define the messages that IDGF should use to promote IDGF services to these EGI groups

    c)       VT members to provide feedback and suggestions for improvement on the email list.

    d)      We want to finalise the content in about one week, then start DG promotion to some target groups. Discuss the first feedbacks at the next teleconference.

 

Next teleconference:

In about two weeks