MinutesSep6

Introduction

• Attendees: Suchandra, Arvind, William, Neha, Ted, Alan Sill, Jeff, Burt, Anand, Steven Timm, Abhishek, Michael Thomas, Charles, Karthik
• Apologies: Xin, Rob
• Coordinates: Thursdays, 2:30pm Central; 510-665-5437, #1122
• Previous meetings, MeetingMinutes

VO Validation (Jeff, Charles, Abhishek)

• SDSS only VO that has done initial work on validation
• Abhishek and Charles have created VO Validation table
• VOs are encouraged to test SE elements
• SE testing is limited to srm since other protocols (dcap, etc.) may not be required by all VOs. Should try gridftp (if enabled), dcap for on-site reads, and srm access.
• Need to get timeframe for release to coordinate with VOs and user groups
  • May take 2 or more weeks for VO validations to complete this is fine - RG

Problems Found

• Condor-g / gridftp problems with ws-gram
  • installing grid-ftp server can be done by installing Globus-Base-Data-Server on a client installation
  • new condor (6.8.6) installed with vdt 1.8.1 should help a bit (will ignore prima callouts if setup by default)
  • two problem cases:
    • running on standalone client (should be fixed with condor 6.8.6 in vdt 1.8.1)
    • running on CE using system wide gridftp (need to resolve the system wide mapping issues)
      • can be worked around by using a personal gridftp server
  • Alan will check with Jaime about the fix in condor 6.8.6
  • Abhishek will check to see if ws-gram problems will affect VO validations
• VOMS file
  • Host DN for LIGO had extra characters and need to be fixed
  • Charles will follow up with Rob Quick to get this updated
• Alan is waiting for new GUMS in VDT 1.8.1

BDII Information, Information Services (Burt)

• Burt sent patch to list and William to fix dropped information from Indiana to interoperability aggregrator
• Burt's site and Purdue are the only sites currently reporting correctly
• Currently running sites should apply patch and new patch should be applied and configure_gip should be rerun
• Action Item: try to get admins to deploy patch and correct information reporting
• Delays due to people at CHEP and vacactions, will see more testing as people get back
• Testing against Burt's site was successful
• Gratia is seeing most of the sites
• ReSS? is operating correcting

Site Status Report (Site Admins)

• BNL:
  • still working on ws-gram validation
  • all other CE services have been validated
  • srm-dCache SE works with 12TB space, ready for VO validation
• FNAL
  • site up and no reports of problems
  • LIGO jobs are being held
• UC
• CIT ITB
  • VORS entry updated so information should be reported correctly
  • limited vo activity, some problems with VOs using remote_initialdir
  • configure_gip patch applied and rerun
  • Alain will send information on nfslite patch so that it can be reported
• CIT LIGO
  • 2 VOs run, LIGO and usatlas1 running successfully
• TTU
  • ready for testing overall, problems with gridftp returns for wsgram
  • considering deploying squid
  • VOMRS is running and successful on site
  • SE being deployed and should be available soon (using srm/bestman)
  • VOs will be notified that an SRM/Bestman will be available for testing
• UCSD
  • Production and testbed SE's, both versions of SRM protocol, available for validation
  • Support all VOs, should be ready soon
  • Charles reporting that site (CE) not showing up in VORs, will follow up with Terrence

Doc (Anne)

• currently reorganizing voms install info
• Forrest is working with Ted on SE info
• osg-client needs a little work for condor-g (see my red comments; suggested test command failed for me)
• Status? Alan will provide instructions for SE install guide for Bestman
• I need guidance on category "where do these go" (bottom of doc table)
• Charles B is assigned to work on consolidating troubleshooting info

Questions

• getting host cert doc. Some docs point to OSG RA, some to vdt and some to the CEConfigurePKI. Should be one place. I suggest OSG RA, and ditching CEConfigurePKI. Comments?
• from last week (answer?): do we want to point to any doc on how to set up a cluster? RG - I think this is beyond the current scope. Perhaps we can follow-up on this post-ITB
• Edu web has some general answers to (user) questions: how to start, monitor, cancel, get results from a job? Documentation/EndUserDocHub#Using_the_Grid?. Are these right for 080? (globusrun-ws -submit ..., globus-job-run..., condor_submit..., condor_submit_dag...). How to handle? RG: I think these should be incorporated into the client validation processes.

AOB

• VOMs problem due to extra token
  • two solutions
    • upgrade all servers
    • remove token (the solution that will be used, no harm in doing this)
  • Alain is following up with VOMS developers (no fix currently, not sure if there will be a fix soon)
• We need to test glexec on at least one site in the ITB.
• Do we have a site that has tested squid?

-- SuchandraThapa - 05 Sep 2007