costantini alessandro infn - igi [email protected] mpi-vt within egi...

14
Costantini Alessandro INFN - IGI [email protected] MPI-VT within EGI 23/03/2012 Costantini A. – MPI-Multicore 1

Upload: tracy-cook

Post on 03-Jan-2016

213 views

Category:

Documents


0 download

TRANSCRIPT

  • Costantini AlessandroINFN - [email protected] within EGI23/03/2012Costantini A. MPI-Multicore*

    23/03/2012Costantini A. MPI-Multicore*

  • Virtual Team23/03/2012Costantini A. MPI-Multicore*Activity established within the EGI-InSPIRE project- user engagement activity- started in 2011Based on two key elementsVirtual Team frameworkNGI International Liaisons

    23/03/2012Costantini A. MPI-Multicore*

  • Virtual Team23/03/2012Costantini A. MPI-Multicore*Virtual Team framework: The Virtual Team framework enables NGI and EGI.eu persons to initiate and participate in short living projects (Virtual Team projects) that focus on well defined, non-operational activities around the production infrastructure.

    23/03/2012Costantini A. MPI-Multicore*

  • Virtual Team23/03/2012Costantini A. MPI-Multicore*NGI International Liaisons: The interaction between the NGI teams and EGI.eu on non-operational activities is undertaken through an "NGI International Liaisons"(NILs in short), who are responsible within the NGIs for the delivery and interaction of non-operational tasks.

    23/03/2012Costantini A. MPI-Multicore*

  • MPI-VT: General informationsLeader: Alvaro Simon (CESGA, Spain) Zdenek Sustr (CESNET, Czech Republic) Start Date: 10/Nov/2011End Date: 31/May/2012

    23/03/2012Costantini A. MPI-Multicore*

    23/03/2012Costantini A. MPI-Multicore*

  • MotivationsDespite a dedicated SA3 activity to support MPI there still seem to be significant issues in uptake and satisfaction amongst the user communities. This VT Works with user communities and projects that use MPI resourcesImprove the communication between MPI users and developers of MPI support within EGI SA3. 23/03/2012Costantini A. MPI-Multicore*

    23/03/2012Costantini A. MPI-Multicore*

  • CZ: Zdenek Sustr (leader) ES/IBERGRID: Alvaro Simon (leader), Enol Fernandez, Ivn Daz, Alvaro Lopez, Pablo Orviz, Isabel Campos, Roberto Rosende Dopazo GR: Dimitris Dellis, Marios Chatziangelou, Paschalis Korosoglou HR: Emir Imamagic, Luko Gjenero IE: John Walsh IT: Daniele Cesini, Alessandro Costantini, Vania Boccia, Marco Bencivenni PT: Gonalo Borges SK: Viera Sipkova, Viet Tran, Jan Astalos UK: John Gordon EGI.eu: Gergely Sipos, Karolis Eigelis, Tiziana Ferrari, Peter Solagna Members23/03/2012Costantini A. MPI-Multicore*

    23/03/2012Costantini A. MPI-Multicore*

  • TasksTask 1: MPI documentation Task 2: Nagios probes Task 3: Information system Task 4: Accounting system Task 5: Batch system status Task 6: Gather information from MPI sites

    23/03/2012Costantini A. MPI-Multicore*

    23/03/2012Costantini A. MPI-Multicore*

  • Task 1: MPI documentation Review EGI documentation- Unser guideAdmin Guide 23/03/2012Costantini A. MPI-Multicore*

    23/03/2012Costantini A. MPI-Multicore*

  • Task 2: Nagios probes Review the current nagios probesMPI nagios new specifications:- org.sam.mpi.EnvSanityCheck -- Test the information published by the (MPI or Parallel) serviceGlueCEPolicyMaxSlotsPerJob - org.sam.mpi.SimpleJob -- Test the MPI functionality with a minimum set of resources (CpuNumber = 2; NodeNumber=2)- org.sam.mpi.ComplexJob -- Test the MPI functionality (CpuNumber = 4; NodeNumber=2; SMPGranularity=2; WholeNodes=True)23/03/2012Costantini A. MPI-Multicore*

    23/03/2012Costantini A. MPI-Multicore*

  • Task 3: Information system - detecting MPI resourcesMaxSlotsPerJobs (GLUE2) value is not filled by the current LRMS Information Providerscput = Maximum amount of CPU time used by all processes in the jobpcput = Maximum amount of CPU time used by any single process in the jobwalltime = Maximum amount of real time during which the job can be in the running state.23/03/2012Costantini A. MPI-Multicore*

    23/03/2012Costantini A. MPI-Multicore*

  • Task 4: Accounting systemImplement MPI accounting systemNo special accounting up to date- Contact with APEL and Accounting Portal23/03/2012Costantini A. MPI-Multicore*

    23/03/2012Costantini A. MPI-Multicore*

  • Task 5: Batch system status

    Two bugs were found during the first UMD verification of WN/Torque + EMI-MPI.1.0. Is a torque/maui problem that affects all MPI jobs. Maui versions prior to 3.3.4 do not allocate correctly all the nodes for the job executionIGI has presented a possible workaround to the problem for Maui versions prior to 3.3.4 Bug fixed in EMI1-IGI (full fixed in EMI2)23/03/2012Costantini A. MPI-Multicore*

    23/03/2012Costantini A. MPI-Multicore*

  • Task 6: Gather information from MPI sites MPI VO -- bring together sites and users interested in MPI This VO is NOT intended for everyday use by all users wishing to use MPI This VO IS intended for users who wish to cooperate with the VT to make MPI support in EGI better The main reason for its establishment is to collect experience that will be later adopted by regular VOs23/03/2012Costantini A. MPI-Multicore*

    23/03/2012Costantini A. MPI-Multicore*

    ****