GridPP Meeting, Cambridge, 15 Feb 2002 - PowerPoint PPT Presentation

1 / 14
About This Presentation
Title:

GridPP Meeting, Cambridge, 15 Feb 2002

Description:

Globus MDS. Machines successfully ... MDS produces some errors in the tree which look sensible. PBS is being ... Globus has 'mds-vo-name=local,o ... – PowerPoint PPT presentation

Number of Views:36
Avg rating:3.0/5.0
Slides: 15
Provided by: pdm
Category:
Tags: gridpp | cambridge | feb | mds | meeting

less

Transcript and Presenter's Notes

Title: GridPP Meeting, Cambridge, 15 Feb 2002


1
UCL Testbed 1 status report
  • Paul Mealor

2
Machines
Imperial RB
RAL country GIIS
3
Installation
  • Used the Kickstart mechanism from Andy McNab
  • Started with a very early version of Globus and
    EDG software
  • Upgrading machines (somewhat slowly) to V1.0.0
    (or V1.1.0 when lists become available)

4
Job submission
  • Submitting jobs to the RB at IC
  • Simple jobs successfully run, submitted to a CE
    at IC from one of the three TB1 machines at UCL
  • Teething problems
  • Cant get the results back!
  • This may be because of version differences
  • Uses the EDG software

5
Job submission
  • Using the RB at IC to submit jobs (cont)
  • Work ongoing
  • progress is good so far
  • will hopefully be able to register CEs at UCL
    with the RB

6
Job submission
  • Also tried Globus job submission system
  • RAL and Bristol
  • Not so much success
  • Not concentrating on this anyway, as Globus
    system does not include sandboxes c.

7
Job submission
  • Other tests
  • Java GRAM interface (Janice Drohan)
  • Still under development
  • Attempts to interact with Globus 2 gatekeeper
  • Successfully submits jobs to our local CE

8
Globus MDS
  • Machines successfully registering to the UCL GIIS
  • GIIS successfully registering to the RAL country
    GIIS.
  • When it was up, the Datagrid GIIS at CERN showed
    metadata from UCL!

9
Ftree MDS
  • Machine metadata successfully cached by UCL GIIS
  • GIIS has appeared at the Ftree country GIIS at RAL

10
CE
  • Installed up to the point of telling it where PBS
    was to be found
  • MDS produces some errors in the tree which look
    sensible
  • PBS is being installed as we speak

11
SE
  • No luck yet
  • Original version of TB1 we installed didnt
    really work
  • SE will probably be the last machine to be
    reinstalled

12
MDS
  • Other checks
  • Java info-system browser (Janice Drohan)
  • Looked at various GRISs and GIISs (at UCL, RAL,
    CERN)
  • Some problems since most recent version of TB1
    installed at UCL
  • Extra EDG info seems to confuse queries
  • Globus has mds-vo-namelocal,ogrid naming
    context
  • EDG ogrid naming context has
    mds-vo-namelocal below it

13
MDS
14
END
Write a Comment
User Comments (0)
About PowerShow.com