Go to the Indico Home Page Oldest event Older event Up to category Newer event Newest event Printable version
  • Compact style
  • Indico style
  • Indico style - inline minutes
  • Indico style - numbered
  • Indico style - numbered + minutes
  • Indico Weeks View
    • Switch to the management area of this event
    Login

    EGEE-SA1-SWE Meeting

    → Europe/Zurich
    Forest Virtual Room (http://www.vrvs.org)

    Forest Virtual Room (http://www.vrvs.org)
    Pacheco, Andreu
    Description
    * Present: Rafa Marco (IFCA), Francisco Bernabé (CESGA), Carlos Borrego (PIC), Javi Fontan (CESGA), Marc Rodriguez (PIC), Gonzalo Merino (PIC), Dani Cano (IFCA), Javi Lopez (CESGA), Manuel Sanchez (USC), Pablo Rey (CESGA), Celso Martinez (IFCA), Javier Calonge (CIEMAT), Kay Neuffer (INTA-CAB), Eduardo Huedo (INTA-CAB), Farida Fassi (IFIC), Pepe Salt (IFIC), Javier Sanchez (IFIC), Santiago Gonzalez de la Hoz (IFIC), Alvaro Fernandez (IFIC), Jorge Gomes (LIP), Mario David (LIP), Carlos Fernandez (CESGA) * Apologies received from Ricardo Graciani (UB)
      • 12:00 → 12:15
        Briefing of activities 15m VRVS

        VRVS
        Speaker: Pacheco, Andreu (IFAE/PIC)
        more information
      • 12:05 → 12:20
        Operational status from sites 15m VRVS

        VRVS
        * From February 1st 2005 all sites must report by mail their operational status each friday if they want Gonzalo Merino to report with accuracy their status to EGEE Operations Meeting * Last friday an old RPM used during the last year's Spanish Certification authority upgrade aused a problem. The problem is solved by uninstalling the old RPM package. New action is created to Rafa Marco to follow-up the problem. Correct configurated sites do not have this problem! Problem is tracked also in the way CRLs are being checked. _
        Speaker: Merino, Gonzalo (PIC)
        more information
        • CESGA 15m
          * Javi Fontan reported. * The new LCG 2.3.0 site is running OK. We had some sporadic problems with RGMA and the resource broker during the week. * The old LCG 2.2.0 site has been removed from GOCdb. * We are now adding the WNs of the old site to the new site. * Next week they will add SWETEST VO and a new one for them. * New site is fully running with Scientific Linux.
          Speaker: Fontan, Javier (CESGA)
        • CIEMAT-CMS 15m
          * Javier Calonge reporting. * New release LCG-2_3_0 is installed with RH7.3. * Apel upgraded to new version 3.4.41-1 * Next week we will continue our planning and we will replace OpenPBS with TORQUE/MAUI and we will start installation of our new Storage Element and additional Worker Nodes. * Test jobs don't run properly in our cluster. They are reporting GSS authentication failure but my own submited jobs works fine and I can get back output from them). They think that the problem is related to the CA issue.
          Speaker: Calonge, Javier (CIEMAT)
        • CNB 15m
          * Reported by mail. * LCG-2_3_0 is being installed (problems with some profiles, queu batch system), the next week I assume that I´ll finih it.. All nodes have RH7.3 * The PCM cluster will be installed the next week using SL3...
          Speaker: Merino, Angel (CNB)
        • IFCA-CMS 15m
          * Dani Cano reporting. * Complete migration to new network. * Complete migration to LCG-2_3_0 and SL3 using generic installtion. * All machines now running SL3, the rest of worker nodes will be added next week. * More problems than initially expected (CE pbs mainly). * The MON node crashes periodically, memory full. It appears as if the pager is full and doesn't drain. I'm investigating possible hardware problems. * Installation of Torque/Maui, with fair share policies active, but in very early stage, this tools have lots of parameters. * Supported VO's are: CMS, ATLAS, LHCB, BIOMED, DTEAM, SWETEST when all WN are added (so next week on moday I expect). * In the GOC web page (RGMA Access) the new domain ifca.org.es is not there. Javi Lopez will do it.
          Speaker: Cano, Dani (IFCA)
        • IFIC-ATLAS 15m
          * Javi Sanchez reporting. * Upgraded to LCG-2_3_0 * added swetest VO support * installed torque/maui * queue reconfiguration (one queue for VO) short: 0:15:0 ALL VOs ific: 300:00:0 VO ific atlas: 300:00:0 VO atlas dteam: 300:00:0 VO dteam swetest: 300:00:0 VO swetest * The initial maui.cnf is based on RAL example. * The VO swetest can run on 10 machines if there is no competition on resources but only 5 if the farm is busy. * Two nodes are reserved for short jobs. * Fair share is enabled (70% atlas, 30% ific initially) * swetest directory created on SE (CASTOR) and software area. * Registering of files for the swetest is not possible though, until a RLS service for this VO is provided. * The SO remains rh7.3. Configuration is still done using LCFGng. Upgrade to SLC3.3 will be made in near future. * We had a hang of our CE (lcg2ce) on friday our 12:00 pm. It was imposible to log into the machine and even the three finger salute didn't work. Display shows a lot of "Out of Memory: Killed process" messages. Gstat @ sinica shows 800 jobs in the queue around 9:00 am, so perhaps is related to this. They will keep an eye on this issue... * Gonzalo reports that keeping lots of queued jobs may exaust CE memory. Each queued job creates globus-job-manager processes and allocated some memory space in the MAUI scheduller. This was seen at PIC and it looks like this happened at IFIC.
          Speaker: Sanchez, Javier (IFIC)
        • INTA-CAB 15m
          * Kai Neuffer reporting. * We are running LCG 2.3.0 over RH7.3 (SL3 on the SE) with no problems. * Next week, we plan to re-install the CE and WNs with SL3 usign Quattor/YAIM. It is possible that we only use the eight P4 machines (1 CE, 1 SE and 6 WNs), currently as WNs. * We are still waiting for new nodes to add more WNs to the production testbed and to prepare the pre-production one. * Once the upgrade is finished we will add SWETEST and BIOMED VOs.
          Speaker: Neuffer, Kai (INTA)
        • LIP 15m
          * Mario David reporting. * LCG version: lcg-2.3 with SCL3 * Status: No special news. * Some problems with RGMA.
          Speaker: David, Mario (LIP)
        • PIC 15m
          * Carlos Borrego reporting. * Problem last week with the MySQL database of the Resource Broker. The job table was full. The table was increased and the problem now is OK. All information of all the job are kept (14 GB of disk space occupied). * Mario reports that some sites related to the Crossgrid project purge the resource broker database. Mario will send this information to the list. * Working on a new cluster testing features of Torque/maui. * LCG version: lcg-2.3 running mixed RH7.3/SCL3
          Speaker: Borrego, Carlos (PIC)
        • UAM-ATLAS 15m
          * Reported by mail. * Running LCG 2.3.0 * Added support for two new VOs: zeus and gvuam.
          Speaker: Pardo, Juan Jose (UAM)
        • UB-LHCB 15m
          * The new air conditioning system is been installed. * This implies an extra delay in the upgrade to LCG-2_3_0, but a new downtime will not be necessary. * GOC downtime has been extended to Friday 11th.
          Speaker: Graciani, Ricardo (UB)
        • UPV-GRyCAP 15m
          * Reported by mail from I. Blanquer. * We are running LCG 2.3.0 over Fedora Core 2. * Installed torque/maui * Problems with RGMA (No RGMAAccess entries in lcgic01.gridpp.rl.ac.uk) * APEL upgraded to 3.4.39 * Supported VO's: BIOMED, DTEAM
          Speaker: Blanquer, Ignacio (UPV-GRyCAP)
        • USC-LHCB 15m
          * Manuel Sanchez reporting (USC) * APEL upgraded to 3.4.39 * The new rack of machines has been set up. They will start testing the machines next week. They will start with RH7.3 and then they will upgrade to SCL3. * They would like to use dcache/srm. * Gonzalo Merino reports a dcache storage element deployed at CERN. Place to start with... * Caros Borrego reports that there is an script in YAIM to install an storage element with dcache.
          Speaker: Sanchez Garcia, Manuel (USC)
        • Savannah EGEE SA1 SWE open tickets for sites 15m
          * Mario David (LIP) reporting. * One ticket issued to UPV to publish spanish-written glue schema tags. * One ticket issued to CNB because RGMA refused connection to servers. * Tickets issued to INTA and UPV to check versions of spanish CAs. Not clear who can close it. * 2 tickets on IFCA- Incorrect information in the GIS. Corrected and closed. * One ticket issued to CIEMAT because not reporting accounting to GOC database. * One ticket issued to UB because of not including profiles for RGMA.
          Speaker: David, Mario (LIP)
      • 12:10 → 12:25
        New accounting web developped at CESGA 15m VRVS

        VRVS
        * This is a presentation of the features available in the web page: http://www.egee.cesga.es/EGEE-SA1-SWE/accounting/reports/ * During the last week some options and features were added. * Some discrepancies between GOC and CESGA accounting data were found but they are related to GOC problem. * Now the values shown are: number of jobs, cpu and elapsed time, memory and other custom queries per site, VO, etc ... * All accounting data is downloaded from GOC and stored at CESGA in a local database. * Andreu Pacheco suggests to create web accessible reports on a monthly basis of standard plots to allow referencing. * Gonzalo mentions the issue of having different normalizations of the CPU time as a function of the type or types of CPUs in the working nodes. * It is assumed in Apel that all sites report CPU time homogeneously. * Please visit the web site and send comments. .
        Speaker: Rey Mayo, Pablo (CESGA)
        more information
      • 12:15 → 12:30
        Action lists 15m VRVS

        VRVS
        Speaker: Pacheco, Andreu (IFAE)
        • 5.1.17.4 Andreu Pacheco to update the SWE Execution Plan (PENDING) 15m
          * CESGA sent updates * IFCA sent updates
          Speaker: Pacheco, Andres (PIC)
          more information
        • 5.1.17.5 All sites except LIP and IFIC to support SWETEST Virtual Organization (PENDING) 15m
          * Done for IFIC after upgrade
          Speaker: Pacheco, Andres (IFAE)
        • 5.1.17.6 CNB and UB to upgrade to LCG 2.3 (PENDING) 15m
          * Done for IFIC, CIEMAT * In progress for CNB and UB.
          Speaker: Merino, Gonzalo (PIC)
          more information
        • 5.1.17.7 CIEMAT, CNB, IFIC, INTA, PIC, UAM and USC to upgrade worker nodes to Scientific Linux 3.0.2 (PENDING) 15m
          * Done for CESGA, LIP and UPV. * UPV upgraded to Fedora Core 2!
          Speaker: Merino, Gonzalo (PIC)
        • 5.1.17.8 CNB, IFCA, UAM, UB to upgrade Apel version 3.4.39-1 (PENDING) 15m
          * New version just available Apel 3.4.41-1 * Done for USC, UPV and CIEMAT. * Ciemat upgraded to the very latest Apel version 3.4.41-1
          Speaker: Lopez Cacheiro, Javier (CESGA)
        • 5.1.24.2 Farida Fassi to deploy the SWE support site (PENDING) 15m
          * Due to some compromises and the review of Crossgrid may delay this action. * All sw components available. * Farida will send an email with a possible date to implement the suport site ~3 weeks.
          Speaker: Fassi, Farida (IFIC-Valencia)
        • 5.1.31.1 Request to support BIOMED from SWE sites (NEW) 15m
          * Done for IFCA, CNB and UPV!
          Speaker: Merino, Angel (CNB)
        • 5.1.31.2 Follow problems adding support for BIOMED VO (CLOSED) 15m
          * This was just a problem of the let's say "non standar location and name" of some variables. Please take this into account when accepting this VO. While for HEP VO's: VO_CMS_SGM=ldap://grid-vo.nikhef.nl/ou=lcgadmin,o=cms,dc=eu-datagrid,dc=org VO_CMS_USERS=ldap://grid-vo.nikhef.nl/ou=lcg1,o=cms,dc=eu-datagrid,dc=org for BIOMED: VO_BIOMED_SGM=ldap://vo-biome.in2p3.fr/ou=lcgadmin,o=biomedical,dc=lcg,dc=org VO_BIOMED_USERS=ldap://vo-biome.in2p3.fr/ou=lcg1,o=biomedical,dc=lcg,dc=org please note o=biomedical, not o=biomed
          Speaker: Merino, Gonzalo (PIC)
          document
        • 5.2.7.1 Rafa Marco (IFCA) will send a report to LCG-ROLLOUT about the Spanish CA Issue (NEW) 15m
          * Last friday an old RPM used during the last year's Spanish Certification authority upgrade caused a problem. The problem is solved by uninstalling the old RPM package. New action is created to Rafa Marco to follow-up the problem. * Correct configurated sites do not have this problem! * Problem is tracked also in the way CRLs are being checked.
          Speaker: Marco, Rafa (IFCA)
          more information
        • 5.2.7.2 Javi Lopez to clarify the CPU normalization being stored in the accounting (NEW) 15m
          * An issue was raised during the meeting about the number of CPU reports being accounted for. * The purpose of this action is to clarify which is the exact meaning of the CPU time stored in the accounting and its relationship with the real kSI2000 being delivered by the sites in the case of worker nodes with different types of CPUs.
          Speaker: Lopez, Javier (CESGA)
        • 5.2.7.3 Implement monthly reporting templates in the CESGA accounting web site for SWE (NEW) 15m
          * The purpose of this action is to create pre-filled templates with standard plots per months for reporting. * These reports should be recreated automatically in case of change of accounting data. e.g. site sending old accounting data. * Each report should have a unique url
          Speaker: Rey, Pablo (CESGA)
        • 5.2.7.4 Implement per-user accounting in a restricted web area (NEW) 15m
          * This is a test deployment to check the use cases for getting report plots and tables in function of the users submitting the jobs. * Only PIC reports per-user accounting. * New document will be created to allow site-managers to send per-user accounting. * Raised security issue in a separate action.
          Speaker: Rey, Pablo (CESGA)
        • 5.2.7.5 Check security issue implementing per-user accounting web (NEW) 15m
          * Related to action 5.2.7.4 * Andreu will contact GDB or EGEE Security Officer
          Speaker: Pacheco, Andreu (PIC)