*** Kudos go to James Letts and Ian Fisk at UCSD for having *** *** the 1st fully installed VDT site as of the end of Today! *** Well done! 1. Quick Status Report: a. Florida: Rick - Some hardware and OS config problems, but nothing serious b. San Diego: Ian and James - no problems at San Diego c. Fermilab: Greg, Lothar, Shahzad, and Jim - Temporarily using raman and droidf for USCMS Test Grid (working just fine). The 3 new dedicated nodes will be installed with VDT by end of March. d. Caltech: Suresh - Fibre connection, gigabit switch in place. e. Wisconsin: Peter could not make the telecon due to Condor Week preparations. 2. VDT (part 1) a. Globus 2.0 beta: [x] Florida [x] San Diego [x] Fermilab [x] Caltech [x] Wisconsin Result of discusion: -------------------- Globus is fully installed and working at each site. b. GDMP 3.0 alpha 3 (and GDMP Heartbeat) [x] Florida [ ] San Diego [x] Fermilab (but not on USCMS Test Grid cluster?) [ ] Caltech [x] Wisconsin (but not on USCMS Test Grid cluster?) Result of discusion: -------------------- Fermilab, Wisconsin, and Florida have GDMP and GDMP Heartbeat runnining (please see: http://yamashita.fnal.gov/~moacyr/gdmp.html). San Diego and Caltech will install GDMP 3.0 alpha 3 and GDMP Heartbeat today. Shahzad agreed to be available today to assist James and Suresh with the install and commissioning of GDMP. We decided not to install GDMP 2.1 as: - GDMP 3.0 will be officially released towards the end of March - Shahzad will provide support for GDMP 3.0 alpha in the mean time. - the EDG testbed is expected to move to GDMP 3.0 towards the end of March Lothar and Greg indicated the importance of using GDMP for the data transfers of the Spring 2002 production (at least among U.S. sites) A last minute discussion with Ian produced the question of which port to use for GDMP 3.0 alpha 3. Previously we had decided to use port 2000 for GDMP 2.1 and port 2050 for GDMP 3.0 alpha 3. However, as we have decided only to install GDMP 3.0 alpha 3, it seems better to use the standard port numbers. *** Hence, we decided to use port number 2000 for GDMP 3.0 alpha 3 *** However, Florida and Fermilab should STAY with the current port numbers and configuration until AFTER Greg's MOP demonstration at CERN next week. 3. MOP a. GRAM jobmanager (http://listserv.fnal.gov/scripts/wa.exe?A1=ind0202&L=uscms-grid-testbed#1) b. MOP information form (http://listserv.fnal.gov/scripts/wa.exe?A2=ind0202&L=uscms-grid-testbed&F=&S=&P=1630) c. MOP site contacts i. Fermilab Jim Amundson (replacement?) ii. Wisconsin Peter Couvares iii. Florida Rick Cavanaugh iv. San Diego James Letts v. Caltech Suresh Singh d. Discuss short-scale timeline for beginning MOP operations Result of discusion: -------------------- The current status of MOP is that Fermilab is the Master MOP site while Florida and Wisconsin are currently configured to accept and process MOP jobs submitted from Fermilab. Ian asked about what Objectivity DB IDs to use for MOP. He suggested that we allocate a small number of IDs at the end of ID range alreadly assigned to the Tier 2 sites. Jim is working with Greg to get IMPALA and MOP fully synchronized (will be finished today). As next week is Condor Week, Peter and Suresh will be unavailable for most of the week and so Wisconsin and Caltech will not participate in MOP before the end of next week. Hence, the immediate plan is to have Greg demonstrate MOP between Fermilab (Master) and Florida (Remote). Since San Diego plans to install GDMP and submit the MOP site information form to Jim today, there is a possibility that San Diego might be included in Greg's MOP demonstration if there is enough time during CMS week to set everything up robustly. Greg, Ian (at CERN) and James (at UCSD) will work out the details. Caltech will be available for MOP operations after the LHC Grid week. So to recap: - MOP demo at CERN, 8 March: Fermilab, Florida, possibly San Diego - Start MOP operations, 11 March: Fermilab, Florida, San Diego, - Full MOP operations, 18 March: Fermilab, Florida, San Diego, Wisconsin, Caltech 4. VDT (part 2) a. Condor 6.3.1 [x] Florida [x] San Diego [ ] Fermilab [ ] Caltech [x] Wisconsin Result of discusion: -------------------- Fermilab plans to install Condor by the end of March Caltech plans to install Condor by 18 March b. Condor-G 6.3.1 [ ] Florida [x] San Diego [x] Fermilab [ ] Caltech [x] Wisconsin Result of discusion: -------------------- Florida plans to install Condor-G by 11 March Caltech plans to install Condor-G by 18 March c. ClassAds 0.9 [ ] Florida [x] San Diego [ ] Fermilab [ ] Caltech [x] Wisconsin Result of discusion: -------------------- Florida plans to install ClassAds by 11 March Fermilab plans to install ClassAds by end of March Caltech plans to install ClassAds by 25 March 5. Discuss short-scale timeline for being "VDT Certified" (full VDT installed on all 5 sites) Result of discusion: -------------------- End of March :-) (UCSD is *already* fully VDT installed!) 6. Simple Monitoring Tools for Grid Status Result of discusion: -------------------- Short term: Following conversations with Paul and Ruth, Rick suggested a *simple* Grid Status monitoring tool to advertise the USCMS "Grid Status". Lothar suggested adopting GridView (developed for the ATLAS testbed). The GDMP Heartbeat page could be incorporated (at a minimum it will be used as a stand alone web page immediately). A Florida CS Grad Student, Suchindra Katageri, will perform the necessary MDS work to configure GridView (or something similiar) for our needs. Longer term: Ian suggested that the USCMS Test Grid is an ideal place for Iosef to deploy and test his monitoring package. This was certainly welcomed. 7. Site description web pages. Result of discusion: -------------------- Rick asked for web page contacts for each of the different grid sites: - Florida : Rick - San Diego: James - Fermilab : Greg will delegate - Caltech : Suresh (will try to delegate) - Wisconsin: Peter (will try to delegate?) To make this as painless as possible, Rick will make a simple html file and distribute it to everyone. Each site can then tailor it (read: slightly modify it or create something completely different :-) to the local site.