Changeset 1059


Ignore:
Timestamp:
Nov 21, 2012 7:07:54 AM (12 years ago)
Author:
raasch
Message:

last version documented

File:
1 edited

Legend:

Unmodified
Added
Removed
  • palm/trunk/SCRIPTS/mrun

    r1058 r1059  
    11#!/bin/ksh
     2
     3# mrun - script for running PALM jobs
    24
    35#--------------------------------------------------------------------------------#
     
    2022# Current revisions:
    2123# -----------------
    22 # Intel inspector (inspxe) is given the number of PEs instead of the number of
    23 # nodes
     24#
    2425#
    2526# Former revisions:
     
    2728# $Id$
    2829#
     30# 1058 2012-11-21 07:00:35Z raasch
     31# Intel inspector (inspxe) is given the number of PEs instead of the number of
     32# nodes
     33#
    2934# 1046 2012-11-09 14:38:45Z maronga
    3035# code put under GPL (PALM 3.9)
    3136#
    32 # mrun - script for running PALM jobs
    33 
    34 
    35      # Procedure for interactive/batch PALM runs (initial runs and automatic
    36      # restart runs)
    37 
    38      # Last changes:
    39      # 03/03/94 - Siggi - Entwicklungsbeginn
    40      # 21/03/94 - Siggi - Entwicklungsabschluss (Version 1.0)
    41      # 09/12/94 - Siggi - allexport unterbunden; statt dessen werden jetzt ein-
    42      #                    zelne Variable exportiert. Dies ist noetig, weil an-
    43      #                    sonsten irgendein Speicher ueberlaeuft und von mrun
    44      #                    aufgerufene Programme (z.B. stageout) mit Fehlern
    45      #                    abbrechen (too many arguments)
    46      # 20/06/95 - Siggi - noclobber muss explizit abgeschaltet, da jetzt
    47      #                    defaultmaessig im RRZN-Profile eingeschaltet
    48      # 10/06/97 - Siggi - Zusaetzliche Option -C bei cpp, damit // nicht
    49      #                    wegfallen
    50      # 25/04/00 - Siggi - Version 1.5
    51      #                    remote_addres und remote_user in return_addres bzw.
    52      #                    return_username umbenannt, return_username wird dem
    53      #                    mrun-Aufruf im Job ueber Option -U mitgegeben.
    54      #                    Neue Variable remote_username, die beim interaktiven
    55      #                    mrun-Aufruf zwingend ueber die neue Option -u oder
    56      #                    in der Konfigurationsdatei angegeben werden muss.
    57      # 08/02/01 - Siggi - Alle mrun-Meldungen ins englische uebersetzt,
    58      #                    Version 1.7
    59      # 05/05/02 - Siggi - Uebersetzungen mittels make-Mechanismus moeglich
    60      # 21/11/02 - Siggi - ENV-Variable XLFRTEOPTS wird vor Ausfuehrung auf
    61      #                    IBM gesetzt, um Record-Laenge der NAMELIST-Dateien,
    62      #                    zu bestimmen; Praeprozessoroptionen haben nun alle
    63      #                    die Form -Dtext=text, weil auf IBM sonst text durch
    64      #                    die leere Zeichenkette ersetzt wird
    65      # 01/08/03 - Siggi  - Test of implementing dvrp+1PE on ibm
    66      # 07/01/04 - Siggi  - additional preprocessor directive for ibm included
    67      #                     (-D$OMP=OMP) in order to avoid problems with
    68      #                     OMP_NUM_THREADS
    69      # 04/01/05 - Siggi  - archiving on hanni and berni is done in separate
    70      #                     tar files, one for each node used
    71      # 07/01/04 - Siggi  - old code for t3e, vpp and hpcs eliminated
    72      # 09/01/05 - Siggi  - workaround for getting the IP address on gfdl3
    73      # 17/01/05 - Siggi  - job for data transfer to local machine now within
    74      #                     job class c1 (HLRN)
    75      # 27/01/05 - Siggi  - IP address workaround for gfdl3 removed
    76      # 28/01/05 - Siggi  - tar-filenames on $PERM (HLRN) now include the
    77      #                     base filename and cycle number
    78      # 16/02/05 - Gerald - hababai validated
    79      # 14/03/05 - Siggi  - abort on NEC after first runtime error (F_ERRCNT)
    80      # 29/03/05 - Marcus - berni*-en0 validated
    81      # 21/04/05 - Siggi  - transfer-job-protocol for avs-data is no more stored
    82      # 24/04/05 - Siggi  - netcdf support on lcmuk
    83      # 25/04/05 - Siggi  - netcdf support on gfld3 (decalpha)
    84      # 11/05/05 - Siggi  - files with non-numeric extensions (.xxx after cycle
    85      #                     number) can now be used within interactive runs
    86      # 12/05/05 - Siggi  - netcdf support on ibm
    87      # 13/05/05 - Siggi  - error in tar-filenames on $PERM (HLRN) removed
    88      # 18/05/05 - Siggi  - netcdf support on nec
    89      # 24/05/05 - Siggi  - netcdf support on ibms
    90      # 14/06/05 - Siggi  - sleep interval after submit of restart job increased
    91      #                     to 30 seconds
    92      # 28/06/05 - Siggi  - bora adjusted to new intel compiler (LD_LIBRARY_
    93      #                     PATH is temporarily set)
    94      # 07/09/05 - Siggi  - setting of MP_EAGER_LIMIT switched off because it
    95      #                     caused warnings in the job protocol
    96      # 20/10/05 - Siggi  - update of netcdf-version on decalpha (gfdl3)
    97      # 25/10/05 - Siggi  - error in listing files for getting the cycle number
    98      #                     removed
    99      # 26/10/05 - Siggi  - new paths for dvrp-library, transfer of catalogs
    100      #                     realized by file attribute trpe
    101      # 28/10/05 - Siggi  - if existing, the file extension is output in case
    102      #                     of missing input file - no job abort in case of
    103      #                     missing input files with file extensions
    104      #                     (preliminary solution)
    105      # 31/10/05 - Siggi  - data transfer from nech now within seperate job
    106      # 04/11/05 - Siggi  - netcdf 3.6.0-p1 on ibmh/ibmb
    107      # 07/12/05 - Siggi  - gallego and elephanta admitted
    108      # 30/12/05 - Siggi  - gfdl5 (ibmy) admitted
    109      # 10/01/06 - Siggi  - cpp directive for NetCDF 64bit support
    110      # 20/01/06 - Siggi  - cpp directive for ibmy
    111      # 09/02/06 - Siggi  - ibmy admitted for batch mode
    112      # 13/04/06 - Siggi  - ostria admitted
    113      # 18/04/06 - Siggi  - usage of OpenMP implemented (additional option -O)
    114      # 10/05/06 - Siggi  - environment variable XLSMPOPTS set on IBM for
    115      #                     optimization of OpenMP runs (option stacksize is
    116      #                     necessary for large gridpoint numbers because
    117      #                     otherwise a segmentation fault occurs)
    118      # 23/05/05 - Siggi  - lctit (SUN Fire X4600) admitted
    119      # 21/08/06 - Siggi  - standard mrun path on scirocco is as on bora
    120      # 23/08/06 - Siggi  - netcdf support for scirocco (notebook)
    121      # 20/09/06 - Marcus - add DVRP library on nech
    122      # 24/10/06 - Siggi  - Environment variables for PALM steering are written
    123      #                     on local NAMELIST file ENVPAR
    124      # 24/11/06 - Siggi  - levanto admitted, maestro switched to ifc 9.1
    125      # 28/11/06 - Siggi  - call to interpret_config.x_levanto because of SuSe 10
    126      # 06/02/07 - Siggi  - add DVRP library on lcmuk
    127      # 07/02/07 - Siggi  - revision of source code handling, source code is
    128      #                     now expected in directory given by source_path and
    129      #                     not in the current working directory, source code
    130      #                     files provided in add_source_path are allways
    131      #                     compiled, source code to be compiled is allways
    132      #                     sampled in SOURCES_FOR_RUN_$fname which is a
    133      #                     subdirectory of the current working directory,
    134      #                     -s TEST  changed to  -s WRITE_PERMIT  or -s WP,
    135      #                     adapted for RIAM (neck)
    136      #                     bugfix for cycle number of output file (extout)
    137      #                     all hpmuk-related code removed
    138      # 14/02/07 - Siggi  - dvrp support for neck added
    139      # 28/02/07 - Siggi  - empty lines in configuration file are accepted
    140      #                     mrun_path replaced by PALM_BIN,
    141      #                     all machines are calling interpret_config.x
    142      # 14/03/07 - Siggi  - fimm admitted, revision number added to terminal
    143      #                     output
    144      # 16/03/07 - Siggi  - adjustments for lctit
    145      # 29/03/07 - Siggi  - global revision transfered to batch job by new
    146      #                     option -G
    147      # 30/03/07 - Siggi  - compilation "by hand" removed, cpp-directives/options
    148      #                     + netcdf/dvrp-options are read from configuration
    149      #                     file, host identifier (local_host) is read from
    150      #                     config file, code related to ftp filetransfer
    151      #                     removed (incl. option -f)
    152      # 19/06/07 - Siggi  - time limit for cdata jobs increased
    153      # 25/07/07 - Siggi  - two sets of executables can be started on lcmuk,
    154      #                     if new option -Y (coupling) is given,
    155      #                     output of executables to aout_output removed,
    156      #                     messages are immediately written to stdout instead
    157      # 03/08/07 - Marcus - add XOPT="-X $numprocs" for lcfimm
    158      # 09/08/07 - Marcus - workaround on lcfimm to propagate environment
    159      #                     variables out to the nodes in coupled mode -disabled-
    160      # 13/08/07 - Marcus - start local restart jobs per ssh on lcfimm
    161      # 28/08/07 - Marcus - completely remove workaround on lcfimm to propagate
    162      #                     environment variables out to the nodes in coupled mode
    163      # 15/10/07 - Siggi  - Preliminary adjustments for lctit, based on Jin's
    164      #                     suggestions
    165      # 19/10/07 - Marcus - further adjustments for lctit: add new optional
    166      #                     argument -g group_number, admit all sla* node_usage
    167      #                     queues
    168      # 30/10/07 - Marcus - further adjustments for queues on lctit
    169      # 14/02/08 - Bjoern - link of restart files with long file name possible
    170      # 28/02/08 - Marcus - If environment variable link_local_output is set to
    171      #                     true, mrun tries "ln -f" on local output and resorts
    172      #                     to "cp" or "cp -r" on error
    173      # 15/04/08 - Siggi  - argument -c introduced to most of the subjob calls,
    174      #                     which allows the user to choose his own job catalog
    175      #                     by setting job_catalog in the configuration file
    176      #                     (default is ~/job_queue),
    177      #                     workaround for mpiexec with -env option,
    178      #                     adjustments for lcxt4 (Bergen Center for Computational
    179      #                     Science)
    180      # 22/05/08 - Marcus - If environment variable link_local_input is set to
    181      #                     true, mrun tries "ln -f" on local input and resorts
    182      #                     to "cp" or "cp -r" on error
    183      # 27/05/08 - Siggi  - PATH is set to PALM_BIN everywhere (missing so far)
    184      # 14/07/08 - Siggi  - adjustments for lcsgih
    185      # 08/08/08 - Marcus - typo removed in lcxt4 branch
    186      # 17/09/08 - Siggi  - restart mechanism adjusted for lcsgi
    187      # 02/10/08 - BjornM - argument "-Y" modified, adjustments for coupled runs
    188      # 21/10/08 - Siggi  - bugfix for the case that -K has more than one
    189      #                     argument
    190      # 08/11/08 - Siggi  - bugfix for depository path settings in configuration
    191      #                     file
    192      # 11/11/08 - Siggi  - rules for using user code files changed: user
    193      #                     code splitted into one file per subroutine,
    194      #                     user can provide his/her own makefile, only
    195      #                     default source-code filenames are allowed
    196      # 14/11/08 - Siggi  - cond1 and cond2 are part of the depository name
    197      # 08/01/09 - Siggi  - mpi module on sgi machines to be used can be set
    198      #                     with environment variable mpilib
    199      # 13/01/09 - Siggi  - totalview on hice/bice can be used in debug mode
    200      #                     using environment variable totalview
    201      # 25/02/09 - Siggi  - cputime-option (-t) is allowed to be omitted for
    202      #                     interactive runs
    203      # 05/03/09 - Siggi  - adjustments for new NEC-SX9 at RIAM (necriam)
    204      # 16/03/09 - Siggi  - dvrp_lib, dvrp_inc replaced by dvr_lib, dvr_inc,
    205      #                     dvr streaming server is automatically started using
    206      #                     new variable dvr_server and a configuration file
    207      #                     .dvrserver.config,
    208      #                     processing of dvr output files using new script
    209      #                     process_dvr_output implemented, in order to make
    210      #                     dvr application more user friendly
    211      # 20/03/09 - Marcus - update of n1ge command for lctit
    212      # 16/04/09 - Siggi  - new option -y for precursor (uncoupled) ocean runs
    213      #                     to be followed by a coupled atmosphere-ocean run
    214      # 21/04/09 - Siggi  - adjustments for new IBM at DKRZ, which is now ibmh
    215      # 24/06/09 - BjornM - adjustments for coupled/precursor runs on ibmy
    216      # 08/07/09 - Siggi  - default value for email_notification is none,
    217      #                     option -e added to subjob
    218      # 19/08/09 - Marcus - MPI debug option
    219      # 21/08/09 - Marcus - totalview on hice/bice also available for mvapich2
    220      # 25/08/09 - BjornM - adpated for lck
    221      # 26/08/09 - Siggi  - additional environment variables for mpt on ice2
    222      # 26/08/09 - Marcus - adjustment for ice2 on lcsgib
    223      # 29/09/09 - Siggi  - default value for archiving is false, option "-A"
    224      #                     switches on archiving, archiving on SGI-ICE enabled
    225      # 16/10/09 - Carolin- adjustments for archiving on SGI-ICE of binary files;
    226      #                     adjustment for special1q
    227      # 18/12/09 - Carolin- modification of archiving (permq)
    228      # 01/02/10 - Siggi  - adapted for lcxt5m and lckyoto (Fujitsu HX600)
    229      # 03/02/10 - Siggi  - make options (mopts) to be set by configuration file
    230      #                     implemented
    231      # 08/02/10 - Siggi  - loading of modules now controlled via configuration
    232      #                     file (modules)
    233      # 02/03/10 - Siggi  - for each block in the configuration file, separate
    234      #                     utility programs are used, combine_plot_fields is
    235      #                     directly called and must not be given by an output
    236      #                     command in the configuration file any more
    237      # 15/06/10 - Rieke  - ssh from compute nodes to login nodes on hosts
    238      #                     lcsgih/lcsgib is done using the names
    239      #                     hicegate0/bicegate0 instead of the IP addresses
    240      # 17/08/10 - BjornM - adjustments for interactive runs on lcxt4
    241      # 07/09/10 - Siggi  - bugfix for wrong netcdf/3.6.3 module on lcsgi
    242      # 08/12/10 - Siggi  - new handling of openmp/hybrid runs, option -O
    243      #                     has now argument threads_per_task
    244      #                     adjustments for Kyushu Univ. (lcrte, ibmku)
    245      # 14/12/10 - Siggi  - adjustments for new Tsubame system at Tokyo
    246      #                     institute of technology (lctit)
    247      # 23/12/10 - Micha  - different number of processors in ocean and
    248      #                     atmosphere is now allowed
    249      # 02/02/10 - Siggi  - further adjustments on Tsubame and concerning openMP
    250      #                     usage
    251      # 09/03/10 - Siggi  - adjustments for ibmkisti, mpt bugfix for netCDF4
    252      #                     usage, totalview usage extended
    253      # 17/03/11 - Siggi  - adjustments for openmp usage on ibmkisti
    254      # 03/04/11 - Micha  - added lckordi
    255      # 06/04/11 - BjornM - bugfix for runs with mpt on lcsgi
    256      # 17/08/11 - Siggi  - extensions for impi library
    257      # 18/08/11 - Siggi  - bugfix for local append of output files with suffix
    258      #                     (.nc)
    259      # 18/08/11 - Marcus - support for Linux OS with German locale
    260      #                   - properly report hosts on general Linux clusters
    261      # 29/08/11 - BjornW - adapted for lcflow (ForWind cluster in Oldenburg)
    262      # 29/08/11 - Carolin- initiating restart-run: adjustment of the path at
    263      #                     IMUK
    264      # 15/09/11 - Siggi  - new option -w tp set the maximum number of parallel
    265      #                     io streams, option -T is obligatory from now on
    266      # 25/01/12 - BjornM - a namelist file check prior to the model run has
    267      #                     been implemented (check_namelist_files).
    268      #                     New option -z can be used to skip the check
    269      # 03/02/12 - BjornM - namelist file check temporary disabled
    270      #                     until severel bugs are fixed
    271      # 08/02/12 - BjornM - bugfixes: skipping namelist file check in case of
    272      #                     restart runs. check is now possible with topography
    273      #                     and the -v (silent) option is considered
    274      # 14/02/12 - BjornM/- bugfix: removed slash in \$queue, changed default
    275      #            Farah    queue for lcsgi to testq, since bigq is not
    276      #                     available on ice1
    277      # 16/02/12 - BjornM - bugfix: replaced $localhost by $host in the ENVPAR
    278      #                     file, which is generated for the namelist file check
    279      # 17/02/12 - Siggi  - -z included in description of options (mrun ?)
    280      # 22/02/12 - Siggi  - -s info updated
    281      # 22/02/12 - Siggi  - case "-s WP" removed
    282      # 28/02/12 - BjornM - further adjustments for namelist file check
    283      # 09/03/12 - Siggi  - -T option not obligatory for lcflow
    284      # 12/03/12 - Siggi  - tasks_per_node divisor check not for lcflow
    285      # 21/03/12 - Matthias - default compiler on ibmh changed
    286      # 02/05/12 - Bjoern  - lcxt4 is now lcxe6, adjustments for lcxe6,
    287      #                      bugfixes for the parameter file check, initial
    288      #                      commands on ssh calls sourced out to .mrun.config
    289      # 05/06/12 - Bjoern  - bugfixes for the parameter file check on lcxe6.
    290      #                      combine_plot_fields can be switched off by the
    291      #                      new mrun option -Z
    292      # 08/06/12 - Bjoern  - max_pr_user enabled for the namelist file check,
    293      #                      sleep commands are now disabled in silent mode (-v)
    294      # 06/08/12 - Bjoern  - update: system update at imuk now requires (new)
    295      #                      initial shell commands on ssh calls (automatic
    296      #                      restart runs were not possible)
    297      # 22/08/12 - Bjoern  - bugfix: missing export statement in the previous
    298      #                      update of the ssh calls
    299      # 06/09/12 - Bjoern  - typo removed: "-D_parallel" -> "-D__parallel"
    300      # 02/10/12 - Siggi   - adjusted for lcsb (urban00)
    301      # 31/10/12 - Siggi   - -T option not obligatory for lcxt5m,
    302      #                      tasks_per_node not checked for this machine
    303      # 01/11/12 - Siggi   - adjusted for lckiaps
    304      # 07/11/12 - Bjoern  - parameter file check adapted for batch jobs on
    305      #                      local host
     37# 21/03/94 - Siggi - first version (1.0)
     38# 03/03/94 - Siggi - start with developing the script
     39#--------------------------------------------------------------------------------#
    30640
    30741
Note: See TracChangeset for help on using the changeset viewer.