SitesCoordOct23

Introduction

  • Attendees: Rob Gardner, Suchandra, Michael Thomas, Jim Blasney, Jeff Porter, Anand, Rob Quick, Karthik, Xin, Ted, George, Para, Sam Hoover
  • Coordinates: Thursday, Oct 23, 2008, 2:30pm Central; 510-665-5437, #1212

Next face-to-face meeting

New wn-client

  • Added LFC client to existing wn-client package
  • Need to update vdt-version package before updating LFC client tools
  • See email sent to osg-sites with detailed update instructions

configure-osg.py updates

  • Update notice sent out by goc with instructions
  • To update pacman -update OSG:osg-config
  • Fixes consist of

Usefulness of these meetings

  • What can we do to make these site coordination meetings more useful so that site more administrators want (and do) attend?
  • Feedback from admins
    • Michael Thomas
      • Have a "featured talk" at each meeting (e.g. a verbal presentation on security, HA, etc.)
    • Karthik
      • Discussions on site experiences / feedback on particular areas
    • Xin
      • Meeting summarizes discussions from mailing lists
      • Continue biweekly format
    • George
      • Tutorials or featured speaker at each meeting
    • Anand
      • Presentations on various topics would be useful
    • Sam Hoover
      • Discussions on monitoring (new software updates, recommendations on which packages sites should be installing and when) which admins would be interested in

Feedback from support

    • Jim
      • More site representation
      • More security feedback
    • Ted
      • Important for ITB and testing coordinations
    • Parag
    • Rob
      • Need general communications from site admins and feedback from the admins
    • Jeff
      • Admin feedback is helpful
    • Suchandra
      • Agree with responses about getting admin feedback
      • Tutorial talks are interesting as well

Carryover topics

  • Will follow up on Globus issues at ITB meeting

Globus infrastructure performance/improvements (Charles)

Pre-WS GRAM performance

  • Follow up last status:
    • A proposal to improve the performance and robustness of GT2 GRAM via a configuration change to use the WS-GRAM SEG (scheduler event generator).
    • http://www.globus.org/toolkit/docs/4.0/execution/prewsgram/admin-index.html#s-gram-admin-seg
    • Protect gk from rogue users that don't have grid monitor enabled. Based on events rather than constant polling.
    • Issue: no condor logfile rotation.
    • Interaction with gridmonitor not known.
    • Not well tested - will test on ITB w/ a user submitting workflow w/o condor-g.
  • Will work with Suchandra and Ben Clifford to try out on UC ITB. Will use Swift - high volume non-condorg submitter.

WS GRAM delegation error

Globus 17 & 43 errors (Steve)

  • Follow-up last status:
    • Large number of Globus errors 17 & 43: Errors being investigated as a Globus-Condor issue: update - there were a couple of fixes in the condor gahp server - Steve Timm is testing a version from Condor.
    • No time to test. There is a patch, will test before next meeting.
    • From Steve: new gahp server available from Jamie. Running for a couple of weeks. Will come with a pre-release of Condor-G (possible will be available in Condor 7.0.6). Effects all scheduler sites.
    • Alan Sill reporting Globus 17 probs, has LSF job manager.

Backlog of Globus patches (Alain)

  • Alain would like test all these on the ITB

OSG 1.0 bugs (Jeff)

  • Follow-up last status:
    • Configure-osg doesn't pick up VO names correctly (footprints 5572 & 5573)
      • Suchandra has sent a new package which should fix these bugs; - sent to Steve, but not yet tested.
    • Configure-osg error at STAR-WSU (footprings 5419) fixed? no longer in footprints
      • Suchandra has sent a fix for these.
    • Once these are tested he'll release it to Rob Q for the OSG.
    • Fixed in just released configure-osg update
  • Brian reported vo-map.txt file if empty, configure-osg causes problems w/ GIP - needs a warning (Fixed in just released update to configure-osg.py)
  • Bestman - SBGrid site problems. Unique problems, no bug reports yet. Fermilab client needs an additional flag.
  • RSV on a non-standard port - not handled well. Requires Suchandra and Arvind's coordination.

AOB

  • Parag - Looking for feedback on priority on work to add support for mpi
    • CEMon plugin available to allow advertising mpi support through ReSS?
    • Purdue testing this
    • Needs to be further discussion and conversation on continuing this
  • Michael wanted suggestions on what to do about failing VOMS servers
    • GOC doesn't currently actively pursue this but will with OIM
    • GOC tickets can be filed and the GOC will follow up and investigate


-- RobGardner - 21 Oct 2008

Topic revision: r5 - 13 Aug 2009 - 19:25:17 - MarcoMambelli

Hello, TWikiGuest!
Register

 
TWIKI.NET

TWiki | Report Bugs | Privacy Policy

This site is powered by the TWiki collaboration platformCopyright by the contributing authors. All material on this collaboration platform is the property of the contributing authors..