MinutesJun7

Introduction

• Last meeting: MinutesMay31; Tuesday's VTB meeting: MinutesVTBJun5
• Previous meetings: MeetingMinutes
• ITB list archive
• VTB list archive
• Coordinates: 2:30pm Central, 1-510-665-5437 #1212

Attending

• Meeting attendees: Terrence, Rob, Suchandra, Stu, John W, Alain, Anand, John R, Rob Q, Horst, Jeff, Burt, Michael, Doug
• Apologies: none

VTB status (Suchandra)

• Change to a new syslog-ng which wont require an injector.
• Changes happening to configure-osg - options for logfile forwarding.
• VTB cache is more or less ready.
• John R - failed, tried to use VTB:ce. Now doing VTB:ce-1.7.0.
• John W - has installed, problems with Condor probe.
• Jeff - noticed Gratia and WS gram components missing. Note - no WS gram for an SGE, yet.
• Should check the new syslog-ng independently.

VDT

• Trying to get new Globus 4.0.5, Condor 6.8.5, Gratia, GUM, Pegasus, trying for new VOMS for VDT 1.7.1
• Will include Glue 1.3? Yes. Will this affect config of CEMon. Will need a round a testing.

WS Gram testing (Jeff)

• In the middle of a new set of tests.
• Found an improved response, but ran into config errors on a new submit host.
• But now new errors after 400 jobs - debugging with Martin. Using UNL to uct3-edge5.
• Suchandra has changed the configuration to allow more jobs to schedule onto the worker nodes.
• Jeff will setup an IM to communicate with Suchandra.

Site availability and validation (RobQ)

• Scheduling mechanism using Condor is in progress
• Probes are being tested by Dan Yokum at Fermilab
• Gratia will be used as a collector to send information to central host.
• There is a tarball package with the available probes that can be tested by hand.
• Few things to be decided - how to run probes locally.
• Test results will be available in the WLCG SAM monitoring; probes are run in a standard way. SAM mechanisms read this data in a standard way.
• Probes are written in perl. Will not be available for Nagios compatible in the first release.
• More on probes from Arvind:

Here's where you can download the current version of the probes:

http://peart.ucs.indiana.edu/docs/osg/OSG_probes.tar.gz

Here's the README file URL if you or folks want to check it out before downloading the probes

http://peart.ucs.indiana.edu/docs/osg/OSG_probes_README.txt

(The above README file also has a link to the WLCG probe standards document)

Like I mentioned in the concall, I'll be making changes to the probes to conform to the latest WLCG stds in the next week or two. But the functionality the probes check for underneath should remain the same for the best part. The output fields may look different in the future version.

Cheers, Arvind

Update on storage validation (John R)

• Managed to get two more sites passing, but two are now failing (7 passing, 6 failing).
• One site is being updated.
• Two with GIP problems.
• http://dev2-scan.grid.iu.edu/cgi-bin/show_results?grid=2
• http://gip-validate.grid.iu.edu/

AOB

• Terrence notes that ATLAS' use of static OSG_WN_TMP (relying on leased local storage on the compute node to be available beyond the duration of the original job) is outside the scope of the original purpose of the variable (cf discussion thread). Long discussion.
• Alain will likely make the default setting for workernode temp to be dynamic, at least for Condor job managers. Horst reports success doing this with LSF, and Steve Timm with PBS.
• Alain reminds us to not forget to register for the OSG site admins meeting in July, see: https://indico.fnal.gov/conferenceDisplay.py?confId=866

-- RobGardner - 06 Jun 2007