124 | HP XC System 4.x Software guide

124

Image 124

HP XC System 4.x Software manual 124

Contents

HP XC System Software XC User Guide Page Table of Contents Configuring Your Environment with Modulefiles Compiling and Linking Serial Applications Debugging Applications Srun Roles Srun Modes 107 115 125 131 10-1 List of Figures11-1 List of Tables List of Examples About This Document Intended AudienceNew and Changed Information in This Edition Typographic Conventions HP XC and Related HP Products Information Related Information HP XC System Software Users Guide Https//computing.llnl.gov/linux/slurm/documentation.html Http//systemimager.org Manpages Http//www-unix.mcs.anl.gov/mpi $ man discover $ man 8 discover HP Encourages Your Comments$ man -k keyword Page Page Only to the administrator of the HP XC system Node SpecializationDetermining the Node Platform /proc/cpuinfo file is dynamic Storage and I/O File SystemSAN Storage Local Storage File System Layout System Interconnect NetworkHP XC System Interconnects Determining System Configuration Information User EnvironmentNetwork Address Translation NAT 1 LVS Application Development Environment CommandsParallel Applications Run-Time Environment Serial ApplicationsSlurm Load Sharing Facility LSF Requested by the HP-MPI mpirun command How LSF and Slurm InteractHP-MPI Nodes for the job Components, Tools, Compilers, Libraries, and Debuggers Page Using the System LVS Login RoutingUsing the Secure Shell to Log Logging In to the System Introduction Getting Information About QueuesGetting Information About Resources $ bqueues Performing Other Common User Tasks Getting System Help and Information Determining the LSF Cluster Name and the LSF Execution Host$ lsid $ man sinfo Configuring Your Environment with Modulefiles Overview of Modules Supplied Modulefiles Supplied Modulefiles HP-MPI Modulefiles Automatically Loaded on the System Viewing Available ModulefilesViewing Loaded Modulefiles Loading a Modulefile Viewing Modulefile-Specific Help Unloading a ModulefileModulefile Conflicts Creating a Modulefile $ module load modules $ man modulefile Developing Applications Application Development Environment Overview Compilers Compiler CommandsMPI Compiler Setting Debugging Options Examining Nodes and Partitions Before Running JobsInterrupting a Job Developing Serial Applications Developing Parallel Applications Serial Application Build EnvironmentBuilding Serial Applications Parallel Application Build Environment Modulefiles HP-MPIOpenMP Pthreads MPI Library Intel Fortran and C/C++CompilersPGI Fortran and C/C++ Compilers GNU C and C++ Compilers Building Parallel Applications Developing Libraries To compile and link a C application using the mpicc commandDesigning Libraries for the CP4000 Platform Linkcommand 64-bit -L/opt/mypackage/lib/x8664 -lmystuff Linkcommand 32-bit -L/opt/mypackage/lib/i686 -lmystuffTo build a 64-bit application, you might enter Developing Applications Submitting a Serial Job Using LSF Submitting a Serial Job with the LSF bsub CommandSubmitting Jobs Overview of Job Submission Example 5-2 Submitting a Serial Job Using LSF Submitting a Serial Job Through Slurm OnlyExample 5-1 Submitting a Job from the Standard Input $ bsub -I srun hostname Following is the command line used to compile this program Submitting a Parallel JobSubmitting a Non-MPI Parallel Job To submit a parallel job Example 5-5 Submitting a Non-MPI Parallel Job $ bsub -n4 -I srun hostname Example 5-7 Submitting an MPI Job $ bsub -n4 -I mpirun -srun ./helloworld Arguments for the Slurm External Scheduler Nodelist=list of nodes $ bsub -n 10 -ext SLURMnodelist=n1-10 srun hostname Submitting a Batch Job or Job Script Example 5-14 Submitting a Job Script$ bsub -n 10 -ext SLURMconstraint=dualcore -I srun hostname $ bqueues -l dualcore grep Slurm $ bsub -n4 -ext SLURMnodes=4 -I ./myscript.sh $ bsub -n4 -I ./myscript.sh Using a Script to Submit Multiple Jobs Using a Makefile to Submit Multiple JobsUse the squeue command to acquire information on the jobs Submitting Multiple MPI Jobs Across the Same Set of Nodes $ cat mymake Following command line makes the program and executes it$ tail 113.out $ cat 117.out Submitting a Job from a Host Other Than an HP XC Host Running Preexecution ProgramsType=SLINUX64 $ bsub -R type=SLINUX64 -n4 -I srun hostnamePage Debugging Applications Debugging Serial ApplicationsDebugging Parallel Applications TotalView Setting Up TotalView Debugging with TotalViewSSH and TotalView Module load mpi module load totalview Using TotalView with Slurm Using TotalView with LSFSetting TotalView Preferences $ srun -Nx-A $ mpirun -tv -srun application Debugging an Application $ mpirun -tv -srun -n2 ./Psimple $ scancel --user username Debugging Running ApplicationsExiting TotalView Run the applicationPage Monitoring Node Activity Xtools Utilities Running Performance Health Tests Where You can list the available tests with the ovp -lcommand $ ovp -l$ ovp --verify=perfhealth/cpuusage $ ovp --verbose --verify=perfhealth/cpuusage HOMEDIRECTORY/ovpn16mmddyy.log HOMEDIRECTORY/ovpn16mmddyyr1.log Page Using the Intel Trace Collector and Intel Trace Analyzer Tuning ApplicationsBuilding a Program Intel Trace Collector and HP-MPI Running a Program Intel Trace Collector and HP-MPI Example 8-1 The vtjacobic Example ProgramFor more information, see the following Web site Example 8-2 C Example Running the vtjacobic Example Program Intel Trace Collector and Analyzer with HP-MPI on HP XC Installation KitHP-MPI and the Intel Trace Collector Running a Program Following is a Fortran example called vtjacobif Visualizing Data Intel Trace Analyzer and HP-MPI Running a Program Across Nodes Using LSF# bsub -n4 -I mpirun.mpich -np 2 ./vtjacobic Page Using Slurm Launching Jobs with the srun CommandIntroduction to Slurm Slurm Utilities Monitoring Jobs with the squeue Command Using the srun Command with HP-MPIUsing the srun Command with LSF Srun Roles and Modes Getting System Information with the sinfo Command Terminating Jobs with the scancel CommandExample 9-5cancels all pending jobs Fault Tolerance Job AccountingSecurity $ sinfo -R Using LSF Information for LSF Overview of LSF Integrated with Slurm $ squeue --jobs $SLURMJOBID Lsfadmin@n16 ~$ bsub -n4 -I srun hostname Example 10-1 Examples of LSF Job Launch Lshosts and lsload commands display for each of these items Differences Between LSF and LSF Integrated with Slurm Job Terminology Connection back to the terminal from which the job wasSubmitted. This job may run immediately, or it may run $ ssh n15 lshosts Batch jobs are submitted with the srun -bcommand. By Default, the output is written toAre submitted with the bsub -Icommand Batch system scheduling policies Using LSF Integrated with Slurm in the HP XC Environment Useful CommandsSubmitting Jobs Job Startup and Job Control How LSF and Slurm Launch and Manage a Job LSF-SLURM External SchedulerLSF with Slurm Job Launch Exit Codes User logs in to login node n16 How LSF and Slurm Launch and Manage a Job Determining the LSF Execution Host Determining Available System Resources Examining System Core Status Following example shows the output from the lshosts commandGetting Information About the LSF Execution Host Node Getting Information About Jobs Getting Host Load InformationExamining System Queues Getting Information About the lsf Partition Getting Job Allocation Information This allocation string has the following valuesLSF job with Slurm allocated resources Than what the job requests Examining the Status of a Job Example 10-5 Using the bjobs Command Short OutputExample 10-4 Job Allocation Information for a Finished Job Example 10-6 Using the bjobs Command Long Output Example 10-7 Using the bhist Command Short OutputOutput Provided by the bhist Command Viewing the Historical Information for a Job Use the bjobs command to view the Slurm Jobid Example 10-8 Using the bhist Command Long OutputTranslating Slurm and LSF JOBIDs Working Interactively Within an Allocation $ sacct -j$ bsub -I -n4 -ext SLURMnodes=4 /bin/bash $ bjobs -l 124 grep slurm This example assumes 2 cores per node Example 10-9 Launching an Interactive MPI JobAlternatively, you can use the following LSF Equivalents of Slurm srun Options 3describes the srun options and lists their LSF equivalentsLSF Equivalents of Slurm srun Options $ unset Slurmjobid $ unset Slurmnprocs Requests a specific list of nodes. Slurm Suppress informational message Advanced Topics Enabling Remote Execution with OpenSSHRunning an X Terminal Session from a Remote Node Determining IP Address of Your Local Machine Running an X terminal Session Using LSF Running an X terminal Session Using SlurmLogging in to HP XC System Using the GNU Parallel Make Capability Options used in this command are Allocate 4 coresRun the job on 1 core Monitors display server address $ cd subdir srun $MAKE $ cd subdir srun -n1 -N1 $MAKE -j4 Example Procedure $ make PREFIX=’srun -n1 -N1 MAKEJ=-j4 Modified Makefile is invoked as follows Local Disks on Compute Nodes$ make PREFIX=srun -n1 -N1 MAKEJ=-j4 Using Mpich on the HP XC System 11.5 I/O Performance ConsiderationsCommunication Between Nodes Shared File View Using Mpich with LSF Allocation Using Mpich with Slurm AllocationBsub command launches the wrapper script Examples Building and Running a Serial ApplicationLaunching a Serial Interactive Shell Through LSF Examine the LSF execution host information Running LSF Jobs with a Slurm Allocation Request Launching a Parallel Interactive Shell Through LSF Example 2. Four Cores on Two Specific Nodes Examine the running jobs information $ hostname n16 $ srun hostname n5 $ bjobs Submitting a Simple Job Script with LSF Examine the finished jobs informationShow the environment Display the script Submitting an Interactive Job with LSF Run the jobSubmit the job Show the job allocation Run some commands from the pseudo-terminal Show the Slurm job IDExit the pseudo-terminal View the interactive jobs View the node state Submitting an HP-MPI Job with LSFView the running job View the finished job Using a Resource Requirements String in an LSF Command$ bsub -n 8 -R ALPHA5 SLINUX64 \ -ext SLURMnodes=4-4 myjob 124 Glossary Network Availability set Fcfs Ipmi LVS PXE SVA Index Index LVS PGI 135