Par-technologies PARASTATION5 V5 Manual do Utilizador

Administrator's GuideRelease 5.0.5Published April 2010

Software6 ParaStation5 Administrator's GuideSoftwareParaStation requires a RPM-based Linux installation, as the ParaStation software is based on

Installation via RPM packagesParaStation5 Administrator's Guide 7mancontains the manual pages describing the ParaStation daemons, utilities and c

Compiling the ParaStation5 packages from source8 ParaStation5 Administrator's GuidePlease note that the individual version numbers of the distinc

Installing the documentationParaStation5 Administrator's Guide 9 # rpm -Uv psmgmt.5.0.0-0.i586.rpm pscom.5.0.0-0.i586.rpm \ pscom-modules.5.0.0

Installing MPI10 ParaStation5 Administrator's Guide # rpm -Uv psdoc-5.0.0-1.noarch.rpmAll the PDF and HTML files will be installed within the di

Uninstalling ParaStation5ParaStation5 Administrator's Guide 11• testingThese steps will be discussed in Chapter 4, Configuration.3.7. Uninstallin

12 ParaStation5 Administrator's Guide

ParaStation5 Administrator's Guide 13Chapter 4. ConfigurationAfter installing the ParaStation software successfully, only few modificatio

Enable optimized network drivers14 ParaStation5 Administrator's GuideThe values that might be assigned to the HWType parameter have to

Testing the installationParaStation5 Administrator's Guide 15transfer application data across Ethernet, this adapted drivers should be used, too.

ParaStation5 Administrator's GuideParaStation5 Administrator's GuideRelease 5.0.5Copyright © 2002-2010 ParTec Cluster Competence Center GmbH

Testing the installation16 ParaStation5 Administrator's GuideAlternatively, it is possible to use the single command form of the psiadmin command

ParaStation5 Administrator's Guide 17Chapter 5. Insight ParaStation5This chapter provides more technical details and background information about

Directory /proc/sys/ps4/state18 ParaStation5 Administrator's GuideThe p4sock.ko module inserts a number of entries within the /proc filesystem. A

Directory /proc/sys/ps4/localParaStation5 Administrator's Guide 19• MaxAcksPending: maximum number of pending ACK messages until an "urgent&

Using the ParaStation5 queuing facility20 ParaStation5 Administrator's Guidea predefined node list. If not defined, all currently known nodes are

ParaStation5 TCP bypassParaStation5 Administrator's Guide 21In order to run applications linked with one of those MPI libraries, Para

Authentication within ParaStation522 ParaStation5 Administrator's GuidePSP_SHM or PSP_SHAREDMEMDon't use shared memory for communication wit

Homogeneous user ID spaceParaStation5 Administrator's Guide 23etc/passwd. Usage of common authentication schemes like NIS is not required and the

Integration with AFS24 ParaStation5 Administrator's Guide5.14. Integration with AFSTo run parallel tasks spawned by ParaStation on clusters usi

Integration with PBS PROParaStation5 Administrator's Guide 25If an external queuing system is used, the environment variable PSI_NODES_SORTshould

ParaStation5 Administrator's Guide iiiTable of Contents1. Introduction ...

Copying files in parallel26 ParaStation5 Administrator's Guide # UseMCaststatement.If Multicast is enabled, the ParaStation daemons exchange sta

Using ParaStation process pinningParaStation5 Administrator's Guide 27To list, sort and filter all the collected information, the command psaccvi

Changing the default ports for psid(8)28 ParaStation5 Administrator's Guideand change the default port number 888.Modify the entry port

ParaStation5 Administrator's Guide 29Chapter 6. TroubleshootingThis chapter provides some hints to common problems seen while installing or using

Problem: cannot start parallel task30 ParaStation5 Administrator's GuideOr logged on to this node, run psiadmin which also starts up

Warning issued on task startupParaStation5 Administrator's Guide 31This typically happens, if the frontend or head node is included as compute no

Problem: processes cannot access files on remote nodes32 ParaStation5 Administrator's GuideMake sure no other process uses this port. Or use the

ParaStation5 Administrator's Guide 33Reference PagesThis appendix lists all reference pages related to ParaStation5 administration tasks. For ref

34 ParaStation5 Administrator's Guide

ParaStation5 Administrator's Guide 35parastation.confparastation.conf — the ParaStation configuration fileDescriptionUpon execution, the ParaStat

ParaStation5 Administrator's Guideiv ParaStation5 Administrator's Guide6.2. Problem: node shown as "down" ...

36 ParaStation5 Administrator's GuideThe following five types of parameters within the Hardware environment will get a special handling fromthe P

ParaStation5 Administrator's Guide 37p4sockUse optimized communication via (Gigabit) Ethernet.The script handling this hardware type ps_p4sock is

38 ParaStation5 Administrator's GuideaccounterThis is actually a pseudo communication layer. It is only used for configuring nodes running thePar

ParaStation5 Administrator's Guide 39Node[s] hostname id [HWType-entry] [starter-entry] [runJobs-entry] [env name value] [env { nameva

40 ParaStation5 Administrator's GuideSelectTime timeSet the timeout of the central select(2) of the ParaStation daemon psid(8) to time seconds.Th

ParaStation5 Administrator's Guide 41The default port to use is 886.RLimit { Core size | CPUTime time | DataSize size | MemLock size | StackSize

42 ParaStation5 Administrator's GuideThe value part of each line either is a single word or an expression enclosed by single or doublequotes. The

ParaStation5 Administrator's Guide 43This only comes into play, if the user does not define a sorting strategy explicitely viaPSI_

44 ParaStation5 Administrator's GuiderdpMaxRetrans numberSet the maximum number of retransmissions within the RDP facility. If more t

ParaStation5 Administrator's Guide 45ACK is sent piggyback within the next regular packet to this node or as soon as a retransmissionoccurred.If

ParaStation5 Administrator's Guide 1Chapter 1. Introduction1.1. What is ParaStationParaStation is an integrated cluster management and communicat

46 ParaStation5 Administrator's Guide

ParaStation5 Administrator's Guide 47psiadminpsiadmin — the ParaStation administration and management toolSynopsispsiadmin [ -denqrsv? ] [ -c com

48 ParaStation5 Administrator's Guide--usageDisplay a brief usage message.Standard InputThe psiadmin command reads standard input for directives

ParaStation5 Administrator's Guide 49If nodes is empty, the node range preselected via the range command is used. The default preselectednode ran

50 ParaStation5 Administrator's Guidecount [hw hw]List the status of the communication system(s) on the selected node(s). Various coun

ParaStation5 Administrator's Guide 51TaskIDThe ParaStation task ID of the process, both as decimal and hexadecimal number. The taskID of a proces

52 ParaStation5 Administrator's Guiderange {[nodes] | all }Preselect or display the default set of nodesIf nodes or all is given, this directive

ParaStation5 Administrator's Guide 53master [nodes]Show the current master on the selected node(s).The master node's task is the management

54 ParaStation5 Administrator's Guidecpumap [nodes]Show the CPU-slot to core mapping list for the selected nodes.bindmem [nodes]Show flag marking

ParaStation5 Administrator's Guide 55rl_sigpending [nodes]Show RLIMIT_SIGPENDING on this node.rl_stack [nodes]Show RLIMIT_STACK on this node.supp

About this document2 ParaStation5 Administrator's GuideIn the middle of 2004, all rights on ParaStation where transferred from ParTec AG to the P

56 ParaStation5 Administrator's Guidehwstart [hw { hw | all } ] [nodes]Start the declared hardware on the selected nodes.Starting a specific hard

ParaStation5 Administrator's Guide 57adminuser [ + | - ] { name | any } [nodes]Grant authorization to start admin-tasks, i.e. task not blocking a

58 ParaStation5 Administrator's GuidePattern Name Description0x0000001 PSC_LOG_PART Partitioning functions (i.e. PSpart_())0x0000002 PSC_LOG_TASK

ParaStation5 Administrator's Guide 59Pattern Name Description0x0001 RDP_LOG_CONN Uncritical errors on connection loss0x0002 RDP_LOG_INIT Info fro

60 ParaStation5 Administrator's GuidenodesSort { PROC | LOAD_1 | LOAD_5 | LOAD_15 | PROC+LOAD | NONE } [nodes]Define the default sorting strate

ParaStation5 Administrator's Guide 61bindmem [ 0 | 1 ] [nodes]Set flag marking if this nodes will use memory-binding as NUMA policy. Relevant val

62 ParaStation5 Administrator's GuidequietQuiet execution. Only a short message is printed if the test was successful.normalNormal execution with

ParaStation5 Administrator's Guide 63psidpsid — the ParaStation daemon. The organizer of the ParaStation software architecture.Synopsispsid [-v?]

64 ParaStation5 Administrator's GuideOptions-d , --debug=levelActivate the debugging mode and set the debugging level to level. If debugging is e

ParaStation5 Administrator's Guide 65test_configtest_config — verify the ParaStation4 configuration file.Synopsistest_config [-vad? ] [-v ] [-a ]

ParaStation5 Administrator's Guide 3Chapter 2. Technical overviewWithin this section, a brief technical overview of ParaStation5 will be given. T

66 ParaStation5 Administrator's Guide

ParaStation5 Administrator's Guide 67test_nodestest_nodes — test physical connections within a cluster.Synopsistest_nodes [-np num] [-cnt count]

68 ParaStation5 Administrator's Guide

ParaStation5 Administrator's Guide 69test_psetest_pse — test virtual connections within a cluster.Synopsistest_pse [-np num]DescriptionThis comma

70 ParaStation5 Administrator's Guide

ParaStation5 Administrator's Guide 71p4statp4stat — display information about the p4sock protocol.Synopsisp4stat [ -v ] [ -s ] [ -n ] [ -? ] [ --

72 ParaStation5 Administrator's Guide

ParaStation5 Administrator's Guide 73p4tcpp4tcp — configure the ParaStation4 TCP bypass.Synopsisp4tcp [ -v ] [ -a ] [ -d ] [ -? ] [ from [ to ]

74 ParaStation5 Administrator's Guide

ParaStation5 Administrator's Guide 75psaccounterpsaccounter — Write accounting information from the ParaStation psid to the accounting files.Syno

License4 ParaStation5 Administrator's Guide• p4sock.o: this module implements the kernel based ParaStation5 communication protocol.• e1000_glue.o

76 ParaStation5 Administrator's GuideCalling psaccounter with -p gzip would call the command gzip yyyymmdd and therefore compressleast recently u

ParaStation5 Administrator's Guide 77psaccviewpsaccview — Print ParaStation accounting information.Synopsispsaccview [ -? | --help ] [ -h | --hum

78 ParaStation5 Administrator's GuideGrouping jobs-lj, --ljobsPrint detailed jobs list. Lists all jobs, one per line.-lu, --ltotuserPrint user li

ParaStation5 Administrator's Guide 79Upon startup psaccview tries to find the file .psaccviewrc in the user's home directory. Within thisf

80 ParaStation5 Administrator's GuideThese column names may also be used for sorting lists, where applicable.Files/var/account/* , /var/account/

ParaStation5 Administrator's Guide 81mlistenmlisten — display multicast pings from the ParaStation daemon psid(8)Synopsismlisten [-dv?] [-m MCAST

82 ParaStation5 Administrator's Guide

ParaStation5 Administrator's Guide 83Appendix A. Quick Installation GuideThis appendix gives a brief overview how to install ParaStation5 on a cl

84 ParaStation5 Administrator's GuideProvided the ParaStation daemon is started by the xinetd, run the psiadmin(1) command located in /opt/parast

ParaStation5 Administrator's Guide 85Appendix B. ParaStation licenseThe ParaStation software may be used under the following terms and conditions

ParaStation5 Administrator's Guide 5Chapter 3. InstallationThis chapter describes the installation of ParaStation5. At first, the prerequisites t

86 ParaStation5 Administrator's GuideCommercial Usemeans any non-consumer use that is not covered by University Use.Know-howmeans program documen

ParaStation5 Administrator's Guide 87§ 6 Grant-Back1. Licensee grants ParTec for Modifications being severable improvements a nonexclusi

88 ParaStation5 Administrator's Guide2. A breach by Licensee of any one of the obligations under sections §4, §5 and §6, will automaticallytermin

ParaStation5 Administrator's Guide 89Appendix C. Upgrading ParaStation4to ParaStation5This appendix explains how to upgrade an existing ParaStati

Changes to the runtime environment90 ParaStation5 Administrator's GuideUse the mpiexec command instead!Executables linked with ParaStation4 can b

ParaStation5 Administrator's Guide 91GlossaryAddress Resolution Protocol A sending host decides, through a protocols routing mechanism, that itwa

92 ParaStation5 Administrator's Guideto store it to a given address. The rest of the jobs is done by this controllerwithout producing further loa

ParaStation5 Administrator's Guide 93Serial Task A single process running on one of the compute nodes within the cluster.This process does not

94 ParaStation5 Administrator's Guide