source: palm/trunk/SCRIPTS/subjob @ 1469

Last change on this file since 1469 was 1469, checked in by maronga, 10 years ago

last commit documented

  • Property svn:keywords set to Id Rev
File size: 52.2 KB
RevLine 
[1]1#!/bin/ksh
[1090]2
3# subjob - script for automatic generation and submission of batch-job files
4#          for various batch queuing systems
5
[1046]6#--------------------------------------------------------------------------------#
7# This file is part of PALM.
8#
9# PALM is free software: you can redistribute it and/or modify it under the terms
10# of the GNU General Public License as published by the Free Software Foundation,
11# either version 3 of the License, or (at your option) any later version.
12#
13# PALM is distributed in the hope that it will be useful, but WITHOUT ANY
14# WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
15# A PARTICULAR PURPOSE.  See the GNU General Public License for more details.
16#
17# You should have received a copy of the GNU General Public License along with
18# PALM. If not, see <http://www.gnu.org/licenses/>.
19#
[1310]20# Copyright 1997-2014  Leibniz Universitaet Hannover
[1046]21#--------------------------------------------------------------------------------#
22#
23# Current revisions:
[1090]24# ------------------
[1351]25#
[1469]26#
[1046]27# Former revisions:
28# -----------------
[169]29# $Id: subjob 1469 2014-09-24 14:09:56Z maronga $
[1200]30#
[1469]31# 1468 2014-09-24 14:06:57Z maronga
32# Typo removed (addres->address)
33# Adjustments for lcxe6
34#
[1453]35# 1452 2014-08-22 09:41:06Z heinze
36# local hosts for blizzard added
37#
[1451]38# 1450 2014-08-21 07:31:51Z heinze
39# HLRN-III (lccrayb): testq queue adjusted to mpp1testq
40#
[1443]41# 1442 2014-07-28 07:09:10Z raasch
42# HLRN-III (lccrayb/lccrayh) queues adjusted
43#
[1379]44# 1378 2014-04-28 06:04:58Z raasch
45# -et option added for lctit
46#
[1351]47# 1350 2014-04-04 13:01:30Z maronga
48# location of qsub updated for lcxe6
49#
[1290]50# 1289 2014-03-04 07:12:34Z raasch
51# German comments translated to English
52# fimm-, necriam-, scirocco-, ibmy-, and sgi-specific code removed
53#
[1280]54# 1279 2014-01-28 12:10:14Z raasch
55# node calculation modified due to changes in mrun (tasks_per_node must not be
56# an integral divisor of numprocs any more)
57#
[1275]58# 1274 2014-01-09 13:14:54Z heinze
59# adjustments for lccrayh
60#
[1267]61# 1266 2013-12-11 12:07:34Z heinze
62# further adjustments for lccrayb (use msub instead of qsub)
63#
[1265]64# 1264 2013-12-09 12:46:09Z fricke
65# Bugfix: Using number of nodes instead of number of processors (lccrayb)
66#
[1263]67# 1262 2013-12-09 10:57:20Z fricke
68# further adjustments for lccrayb
69#
[1261]70# 1260 2013-12-04 12:48:04Z raasch
71# jaboticaba admitted
72#
[1256]73# 1255 2013-11-07 14:43:35Z raasch
74# further adjustments for lccrayb
75#
[1225]76# 1224 2013-09-16 07:27:23Z raasch
77# first adjustments for lccrayb
78#
[1203]79# 1202 2013-07-10 16:22:07Z witha
80# adjustments for Forwind cluster (lcflow)
81#
[1200]82# 1199 2013-07-05 14:52:22Z raasch
83# adjustments for CSC Helsinki (lccrayf)
84#
[1185]85# use of cluster/express queue enabled (ibmh)
86# vinessa added (imuk)
[1047]87#
[1104]88# 1103 2013-02-20 02:15:53Z raasch
89# bash compatibility adjustments (usage of OPTIND, output formatting with printf
90# instead typeset -L/R),
91# further adjustments for lckyuh
92#
[1100]93# 2013-02-10 01:47:43Z raasch
94# adjustments for Kyushu-Univeristy computing center (lckyuh - hayaka)
95# and for Forwind cluster (lcflow)
96#
[1096]97# 1094 2013-02-03 01:52:12Z raasch
98# new option -P for explicit setting of ssh/scp port,
99# decalpha parts (yonsei) removed
100#
[1091]101# 2013-02-02 07:06:13Z raasch
[1099]102# adjustments for Kyushu-University computing center (lckyut - tatara)
[1091]103# old changelog messages removed
104#
[1047]105# 1046 2012-11-09 14:38:45Z maronga
106# code put under GPL (PALM 3.9)
107#
[1090]108# 08/07/94 - Siggi - first version finished
109# 29/06/94 - Siggi - script development started
110#--------------------------------------------------------------------------------#
111# subjob - script for automatic generation and submission of batch-job files
112#          for various batch queuing systems
113#--------------------------------------------------------------------------------#
[1]114
115
[1289]116    # VARIABLE-DECLARATIONS AND DEFAULT VALUES
[352]117 delete_dayfile=false
[799]118 email_notification=none
[122]119 group_number=none
[1]120 locat=normal
121 no_default_queue=none
122 no_submit=false
123 job_catalog="~/job_queue"
124 job_name=none
125 local_user=$LOGNAME
126 node_usage=shared
[475]127 numprocs=0
[1]128 punkte="..........................................................."
129 submcom=qsub
130 queue=default
131 remote_host=none
132 remote_user=""
133 verify=true
134
135 typeset  -i   cputime=memory=Memory=0  minuten  resttime  sekunden  stunden
[696]136 typeset  -i   inumprocs  mpi_tasks=nodes=processes_per_node=0 tasks_per_node=threads_per_task=1
[1]137
138
139
[1289]140    # ERROR HANDLING
141    # IN CASE OF EXIT:
[1]142 trap 'if [[ $locat != normal ]]
143       then
144          case  $locat  in
145             (option)  printf "\n  --> available optios can be displayed"
146                       printf " by typing:"
147                       printf "\n      \"subjob ?\" \n";;
148             (ftpcopy|parameter|scp|verify)  printf "\n";;
149             (*)       printf "\n  +++ unknown error"
150                       printf "\n      please inform S. Raasch!\n"
151          esac
152          [[ -f $job_to_send ]]  &&  rm  $job_to_send
153          printf "\n\n+++ SUBJOB killed \n\n"
154       fi' exit
155
156
[1289]157    # IN CASE OF TERMINAL-BREAK:
[1]158 trap '[[ -f $job_to_send ]]  &&  rm  $job_to_send
159       printf "\n\n+++ SUBJOB killed \n\n"
160       exit
161      ' 2
162
163
164
165
[1289]166    # DETERMINE NAME OF LOCAL HOST
[1]167 local_host=$(hostname)
168
169
170
[1289]171    # SET HOST-SPECIFIC VARIABLES VEREINBAREN (CHECK, IF LOCAL HOST
172    # IS ADMITTED AT ALL)
173    # NOTE: ONE OF THE ENTRIES FOR "lck" OR "lckordi" ALWAYS HAS TO BE
174    # COMMENT OUT, BECAUSE THE HOSTNAME (node*) IS SAME FOR BOTH MACHINES
[1]175 case  $local_host  in
[1468]176     (ambiel-lx)             local_address=134.106.74.48;  local_host=lcfor;;
177     (atmos)                 local_address=172.20.25.35;   local_host=lcide;;
178     (austru)                local_address=130.75.105.128; local_host=lcmuk;;
179     (autan)                 local_address=130.75.105.57;  local_host=lcmuk;;
180     (bora)                  local_address=130.75.105.103; local_host=lcmuk;;
181     (b04*)                  local_address=133.5.4.33;     local_host=lckyuh;;
182     (blizzard1|p*)          local_address=136.172.40.15;  local_host=ibmh;;
183     (blizzard2|p*)          local_address=136.172.40.16;  local_host=ibmh;;
184     (blogin*|bxc*)          local_address=130.73.233.1;   local_host=lccrayb;;
185     (hlogin*|hxc*)          local_address=130.75.4.1;     local_host=lccrayh;;
186     (breva)                 local_address=130.75.105.98;  local_host=lcmuk;;
187     (buran)                 local_address=130.75.105.58;  local_host=lcmuk;;
188     (caurus)                local_address=130.75.105.19;  local_host=lcmuk;;
189     (climate*)              local_address=165.132.26.68;  local_host=lcyon;;
190     (clogin*)               local_address=86.50.166.21;   local_host=lccrayf;;
191     (cs*)                   local_address=136.172.44.131; local_host=nech;;
192     (elephanta)             local_address=130.75.105.6;   local_host=lcmuk;;
193     (flow01)                local_address=10.141.255.71;  local_host=lcflow;;
194     (flow02)                local_address=10.141.255.72;  local_host=lcflow;;
195     (node*)                 local_address=165.132.26.61   local_host=lck;;
196   #  (node*)                 local_address=210.219.61.8    local_host=lckordi;;
197     (gaia*)                 local_address=150.183.146.24; local_host=ibmkisti;;
198     (gallego)               local_address=130.75.105.10;  local_host=lcmuk;;
199     (gregale)               local_address=130.75.105.109; local_host=lcmuk;;
200     (hababai)               local_address=130.75.105.108; local_host=lcmuk;;
201     (hayaka*)               local_address=133.5.4.33;     local_host=lckyuh;;
202     (hexagon.bccs.uib.no)   local_address=129.177.20.113; local_host=lcxe6;;
203     (hx*)                   local_address=133.3.51.11;    local_host=lckyoto;;
204     (inferno)               local_address=130.75.105.5;   local_host=lcmuk;;
205     (irifi)                 local_address=130.75.105.104; local_host=lcmuk;;
206     (jaboticaba)            local_address=150.163.25.181; local_host=lcbr;;
207     (sno)                   local_address=130.75.105.113; local_host=lcmuk;;
208     (levanto)               local_address=130.75.105.45;  local_host=lcmuk;;
209     (login*)                local_address=118.128.66.223; local_host=lckiaps;;
210     (maestro)               local_address=130.75.105.2;   local_host=lcmuk;;
211     (meller)                local_address=134.106.74.155; local_host=lcfor;;
212     (meteo-login*)          local_address=193.166.211.144;local_host=lcxt5m;;
213     (hexagon*)              local_address=129.177.20.113; local_host=lcxe6;;
214     (nobel*)                local_address=150.183.5.101;  local_host=ibms;;
215     (orkan)                 local_address=130.75.105.3;   local_host=lcmuk;;
216     (ostria)                local_address=130.75.105.106; local_host=lcmuk;;
217     (paesano)               local_address=130.75.105.46;  local_host=lcmuk;;
218     (pcj*)                  local_address=172.31.120.1;   local_host=lckyut;;
219     (pingui)                local_address=134.106.74.118; local_host=lcfor;;
220     (quanero)               local_address=130.75.105.107; local_host=lcmuk;;
221     (rte*)                  local_address=133.5.185.60;   local_host=lcrte;;
222     (shiokaze-lx)           local_address=134.106.74.123; local_host=lcfor;;
223     (sisu-login*)           local_address=86.50.166.21;   local_host=lccrayf;;
224     (solano)                local_address=130.75.105.110; local_host=lcmuk;;
225     (sugoka*)               local_address=172.31.120.1;   local_host=lckyut;;
226     (t2a*)                  local_address=10.1.6.165;     local_host=lctit;;
227     (urban*)                local_address=147.46.30.151   local_host=lcsb;;
228     (vinessa)               local_address=130.75.105.112; local_host=lcmuk;;
229     (vorias)                local_address=172.20.25.43;   local_host=lcmuk;;
230     (*.cc.kyushu-u.ac.jp)   local_address=133.5.4.129;    local_host=ibmku;;
[1]231     (*)                     printf "\n  +++ \"$local_host\" unknown";
[1255]232                             printf "\n      please contact the PALM group at IMUK";
[1]233                             locat=parameter; exit;;
234 esac
235
236
237
[1289]238    # BY DEFAULT, THE REMOTE HOST IS THE LOCAL HOST
[1]239 remote_host=$local_host
240
241
242
243
[1289]244    # READ THE SHELLSCRIPT-OPTIONS
[1094]245 while  getopts  :c:dDe:g:h:m:n:N:O:P:q:t:T:u:vX:  option
[1]246 do
247   case  $option  in
248       (c)   job_catalog=$OPTARG;;
249       (d)   delete_dayfile=true;;
250       (D)   no_submit=true;;
[352]251       (e)   email_notification=$OPTARG;;
[125]252       (g)   group_number=$OPTARG;;
[1]253       (h)   remote_host=$OPTARG;;
254       (m)   memory=$OPTARG;;
255       (n)   job_name=$OPTARG;;
256       (N)   node_usage=$OPTARG;;
257       (O)   threads_per_task=$OPTARG;;
[1094]258       (P)   scp_port=$OPTARG;;
[1]259       (q)   no_default_queue=$OPTARG;;
260       (t)   cputime=$OPTARG;;
261       (T)   tasks_per_node=$OPTARG;;
262       (u)   remote_user=$OPTARG;;
263       (v)   verify=false;;
264       (X)   numprocs=$OPTARG;;
265       (\?)  printf "\n  +++ Option $OPTARG unknown \n";
266             locat=option; exit;;
267   esac
268 done
269
270
[1289]271    # GET THE NAME OF THE JOBFILE AS NEXT ARGUMENT
[1103]272 (( to_shift = $OPTIND - 1 ))
273 shift $to_shift; file_to_send=$1
[1]274
275
[1289]276    # OUTPUT OF SHORT DESCRIPTION OF SCRIPT-OPTIONS
[1]277 if [ "$1" = "?" ]
278 then
279   (printf "\n  *** subjob can be called as follows:\n"
280    printf "\n      subjob -c.. -d -D -h.. -m.. -q.. -t.. -u.. -v  <jobfile>\n"
281    printf "\n      Description of available options:\n"
282    printf "\n      Option  Description                         Default-Value"
283    printf "\n        -c    job-input- and output-catalog       ~/job_queue"
284    printf "\n        -d    no job-protocol will be created     ---"
285    printf "\n        -D    only the job-file will be created   ---"
286    printf "\n        -h    execution host, available hosts:    $remote_host"
[1289]287    printf "\n              ibm, ibmh, ibmkisti, ibmku, ibms, lc...,"
288    printf "\n              lckiaps, lctit, nech"
[1]289    printf "\n        -m    memory demand per process in MByte  ---"
290    printf "\n        -n    jobname                             <jobdatei>"
291    printf "\n        -O    threads per task (for OpenMP usage) 1"
[1094]292    printf "\n        -P    ssh/scp port                        default port"
[1]293    printf "\n        -q    job-queue to be used                default"
294    printf "\n        -t    allowed cpu-time in seconds         ---"
295    printf "\n        -T    tasks per node (on parallel hosts)  ---"
296    printf "\n        -u    username on execution host          from .netrc"
297    printf "\n        -v    no prompt for confirmation          ---"
298    printf "\n        -X    # of processors (on parallel hosts) 1"
299    printf "\n "
300    printf "\n      The only possible positional parameter is <jobfile>:"
301    printf "\n      The complete NQS-job must be provided here."
302    printf "\n      <jobfile>=? creates this outline\n\n") | more
303    exit
304 fi
305
306
307
[1289]308    # CHECK, IF JOB-FILE HAS BEEN GIVEN AS ARGUMENT AND IF THE FILE ITSELF EXISTS
[1]309 if [[ "$file_to_send" = "" ]]
310 then
311    printf "\n  +++ job-file missing"
312    locat=parameter; exit
313 else
314    if [[ -f $file_to_send ]]
315    then
316       true
317    else
318       printf "\n  +++ job-file: "
319       printf "\n           $file_to_send"
320       printf "\n      does not exist"
321       locat=parameter; exit
322    fi
323 fi
324
325
326
[1289]327    # IF NO JOBNAME HAS BEEN GIVEN, JOBNAME IS SET TO THE NAME OF THE JOB-FILE,
328    # PROVIDED THAT THE JOB-FILE NAME DOES NOT CONTAIN ANY PATH
[1]329 if [[ $job_name = none ]]
330 then
331    job_name=$file_to_send
332 fi
333 if [[ $(echo $job_name | grep -c "/") != 0 ]]
334 then
335    printf "\n  +++ job-file name: "
336    printf "\n           $job_name"
337    printf "\n      must not contain \"/\"-characters"
338    locat=parameter; exit
339 fi
340
341
342
343
[1289]344    # SET HOST-SPECIFIC QUANTITIES, OR TERMINATE IN CASE OF UNKNOWN HOST,
345    # OR IF NO HOST HAS BEEN GIVEN
[1]346 if [[ $remote_host = none ]]
347 then
348    printf "\n  +++ host missing"
349    locat=option; exit
350 else
351    case  $remote_host  in
[1468]352        (ibm)     queue=p690_standard; remote_address=134.76.99.81; submcom=/usr/lpp/LoadL/full/bin/llsubmit;;
353        (ibmh)    queue=cluster; remote_address=136.172.40.15; submcom=/usr/lpp/LoadL/full/bin/llsubmit;;
354        (ibmkisti) queue=class.32plus; remote_address=150.183.146.24; submcom=/usr/lpp/LoadL/full/bin/llsubmit;;
355        (ibmku)   queue=s4; remote_address=133.5.4.129; submcom=/usr/local/bin/llsubmit;;
356        (ibms)    queue=p_normal; remote_address=150.183.5.101; submcom=/usr/lpp/LoadL/full/bin/llsubmit;;
357        (lccrayb) queue=mpp1testq; remote_address=130.73.233.1; submcom=/opt/moab/7.2.6/bin/msub;;
358        (lccrayh) queue=mpp1testq; remote_address=130.75.4.1; submcom=/opt/moab/7.2.6/bin/msub;;
359        (lccrayf) queue=small; remote_address=86.50.166.21; submcom=/opt/slurm/default/bin/sbatch;;
360        (lcflow)  remote_address=10.140.1.71; submcom=/cm/shared/apps/sge/6.2u5p2/bin/lx26-amd64/qsub;;
361        (lckyoto) remote_address=133.3.51.11; submcom=/thin/local/bin/qsub;;
362        (lck)     remote_address=165.132.26.61; submcom=/usr/torque/bin/qsub;;
363        (lckiaps) remote_address=118.128.66.223; submcom=/cm/shared/apps/pbspro/11.0.2.110766/bin/qsub;;
364        (lckordi) remote_address=210.219.61.8; submcom=/usr/torque/bin/qsub;;
365        (lckyuh)  remote_address=133.5.4.33; submcom=/usr/bin/pjsub;;
366        (lckyut)  remote_address=133.5.4.37; submcom=/usr/bin/pjsub;;
367        (lcsb)    remote_address=147.46.30.151; submcom=/usr/torque/bin/qsub;;
368        (lctit)   queue=S; remote_address=10.1.6.165; submcom=/opt/pbs/tools/bin/t2sub;;
369        (lcxe6)   remote_address=129.177.20.113; submcom=/opt/torque/default/bin/qsub;;
370        (lcxt5m)  remote_address=193.166.211.144; submcom=/opt/pbs/10.1.0.91350/bin/qsub;;
371        (lcyon)   remote_address=165.132.26.68; submcom=/usr/torque/bin/qsub;;
372        (nech)    qsubmem=memsz_job; qsubtime=cputim_job; remote_address=136.172.44.147; submcom="/usr/local/bin/qsub";;
[251]373        (*)       printf "\n  +++ hostname \"$remote_host\" not allowed";
374                  locat=parameter; exit;;
[1]375    esac
376 fi
377
378
[1289]379    # CHECK, IF A VALID QUEUE HAS BEEN GIVEN
[1]380 if [[ $no_default_queue != none ]]
381 then
382    error=false
383    ndq=$no_default_queue
384    case  $remote_host  in
385        (ibm)    case  $ndq  in
386                     (p690_express|p690_standard|p690_long)  error=false;;
387                     (*)                                     error=true;;
388                 esac;;
389        (ibmh)   case  $ndq  in
[1184]390                     (cluster|express)  error=false;;
[1]391                     (*)                                     error=true;;
392                 esac;;
[693]393        (ibmkisti)   case  $ndq  in
394                     (class.32plus|class.1-2|class.2-32)  error=false;;
395                     (*)                                     error=true;;
396                 esac;;
[622]397        (ibmku)  case  $ndq  in
398                     (sdbg1|sdbg2|sdbg4|s4|s16|s32|s32-s)    error=false;;
399                     (*)                                     error=true;;
400                 esac;;
[1]401        (ibms)   case  $ndq  in
402                     (express|normal|p_express|p_normal|p_normal_1.3|p_normal_1.7|grand)     error=false;;
403                     (*)                                     error=true;;
404                 esac;;
[1224]405        (lccrayb) case  $ndq  in
[1442]406                     (dataq|mpp1q|mpp1testq|smp1q|smp1testq|specialm1q)   error=false;;
[1224]407                     (*)                                     error=true;;
408                 esac;;
[1274]409        (lccrayh) case  $ndq  in
[1442]410                     (dataq|mpp1q|mpp1testq|smp1q|smp1testq|specialm1q)   error=false;;
[1274]411                     (*)                                     error=true;;
412                 esac;;
[1197]413        (lccrayf) case  $ndq  in
414                     (usup|test*|small|large)                error=false;;
415                     (*)                                     error=true;;
416                 esac;;
[1099]417        (lcflow) case  $ndq  in
418                     (cfd_lom_long.q|cfd_him_long.q|cfd_lom_serl.q|cfd_lom_shrt.q|cfd_him_shrt.q)  error=false;;
419                     (*)                                     error=true;;
420                 esac;;
[1040]421        (lckiaps) case  $ndq  in
422                     (express|normal)                        error=false;;
423                     (*)                                     error=true;;
424                 esac;;
[440]425        (lckyoto) case  $ndq  in
426                     (eh|ph)                                 error=false;;
427                     (*)                                     error=true;;
428                 esac;;
[1099]429        (lckyuh) case  $ndq  in
430                     (fx-dbg|fx-single|fx-small|fx-middle|fx-large)  error=false;;
431                     (*)                                     error=true;;
432                 esac;;
[1090]433        (lckyut) case  $ndq  in
434                     (cx-dbg|cx-single|cx-small|cx-middle|cx-large)  error=false;;
435                     (*)                                     error=true;;
436                 esac;;
[1]437        (lctit)  case  $ndq  in
[635]438                     (G|L128|L256|L512H|S|S96|V)             error=false;;
[1]439                     (*)                                     error=true;;
440                 esac;;
441        (t3eb)   case  $ndq  in
442                     (berte|p50|p100|p392|forfree|p25himem)  error=false;;
443                     (*)    error=true;;
444                 esac;;
445        (t3eh)   case  $ndq  in
446                     (para_t3e|em|k|l|lm|comp_t3e|c|p|ht)  error=false;;
447                     (*)    error=true;;
448                 esac;;
449        (t3ej2|t3ej5)  case  $ndq  in
450                     (low|normal|high)  error=false;;
451                     (*)    error=true;;
452                 esac;;
453        (t3es)  case  $ndq  in
454                     (batch|serial-4|pe4|p48|pe16|pe32|pe64|pe128)  error=false;;
455                     (*)    error=true;;
456                 esac;;
457    esac
458    if [[ $error = true ]]
459    then
460       printf "\n  +++ queue \"$no_default_queue\" on host \"$remote_host\" not allowed"
461       locat=parameter; exit
462    else
463       queue=$no_default_queue
464    fi
465 fi
466
467
468
[1289]469    # CHECK THE CPU-TIME
470    # SPLIT TIME INTO HOURS, MINUTES, AND SECONDS
[1]471 done=false
472 while [[ $done = false ]]
473 do
474    if (( $cputime <= 0 ))
475    then
476       printf "\n  +++ wrong cpu-time or cpu-time missing"
477       printf "\n  >>> Please type cpu-time in seconds as INTEGER:"
478       printf "\n  >>> "
479       read  cputime  1>/dev/null  2>&1
480    else
481       done=true
482    fi
483 done
484 if [[ $remote_host = nech ]]
485 then
486    if (( tasks_per_node != 0 ))
487    then
488       (( cputime = cputime * tasks_per_node ))
489    elif [[ $numprocs != 0 ]]
490    then
491       (( cputime = cputime * numprocs ))
492    fi
493 fi
494 (( stunden  = cputime / 3600 ))
495 (( resttime = cputime - stunden * 3600 ))
496 (( minuten  = resttime / 60 ))
497 (( sekunden = resttime - minuten * 60 ))
498 timestring=${stunden}:${minuten}:${sekunden}
499
500
501
[1289]502    # CHECK THE MEMORY DEMAND
[1]503 done=false
504 while [[ $done = false ]]
505 do
506    if (( memory <= 0 ))
507    then
508       printf "\n  +++ wrong memory demand or memory demand missing"
509       printf "\n  >>> Please type memory in  MByte per process  as INTEGER:"
510       printf "\n  >>> "
511       read  memory  1>/dev/null  2>&1
512    else
513       done=true
514    fi
515 done
516
[1289]517 if [[ $remote_host = nech ]]
[1]518 then
519    if (( tasks_per_node != 0 ))
520    then
521       (( Memory = memory * tasks_per_node / 1000 ))
522    elif [[ $numprocs != 0 ]]
523    then
524       (( Memory = memory * numprocs / 1000 ))
525    else
526       (( Memory = memory / 1000 ))
527    fi
[635]528 elif [[ $remote_host = lctit ]]
529 then
530    (( Memory = memory * tasks_per_node / 1000 ))
[1]531 fi
532
533
[1289]534    # MEMORY DEMAND IN CASE OF OPENMP-USAGE ON IBM-SYSTEMS
[1]535 if [[ $(echo $remote_host | cut -c1-3) = ibm ]]
536 then
537    (( memory = memory * threads_per_task ))
538 fi
539
540
[1289]541    # CALCULATE NUMBER OF REQUIRED NODES
[1]542 if (( tasks_per_node != 0 ))
543 then
[1279]544    (( nodes = ( numprocs - 1 ) / ( tasks_per_node * threads_per_task ) + 1 ))
[1]545 fi
546
[1094]547
[1289]548    # CALCULATE NUMBER OF PROCESSES PER NODE
[622]549 (( processes_per_node = tasks_per_node * threads_per_task ))
[1]550
[1094]551
[1289]552    # CALCULATE NUMBER OF MPI TASKS
[696]553 (( mpi_tasks = numprocs / threads_per_task ))
[1]554
[696]555
[1289]556    # SET PORT NUMBER OPTION FOR CALLS OF ssh/scp, subjob AND batch_scp SCRIPTS
[1094]557 if [[ "$scp_port" != "" ]]
558 then
559    PORTOPT="-P $scp_port"
560    SSH_PORTOPT="-p $scp_port"
561 fi
562
563
[1289]564    # HEADER-OUTPUT
[1]565 if [[ $verify = true ]]
566 then
567    printf "\n\n"
568    printf "#--------------------------------------------------------------# \n"
569    spalte1=SUBJOB;spalte2=$(date)
[1103]570    printf "| %-20s%40s | \n" "$spalte1" "$spalte2"
[1]571    printf "|                                                              | \n"
572    printf "| values of parameters/options:                                | \n"
[1103]573    spalte1=$(echo local_host$punkte | cut -c-20)
574    spalte2=$punkte$local_host
575    printf "| %-20s%40s | \n" "$spalte1" "${spalte2: -40}"
576    spalte1=$(echo remote_host$punkte | cut -c-20)
577    spalte2=$punkte$remote_host
578    printf "| %-20s%40s | \n" "$spalte1" "${spalte2: -40}"
579    spalte1=$(echo queue$punkte | cut -c-20)
580    spalte2=$punkte$queue
581    printf "| %-20s%40s | \n" "$spalte1" "${spalte2: -40}"
582    spalte1=$(echo memory$punkte | cut -c-20)
583    spalte2="$punkte$memory mb"
584    printf "| %-20s%40s | \n" "$spalte1" "${spalte2: -40}"
585    spalte1=$(echo cputime$punkte | cut -c-20)
586    spalte2="$punkte$cputime sec"
587    printf "| %-20s%40s | \n" "$spalte1" "${spalte2: -40}"
588    spalte1=$(echo job_name$punkte | cut -c-20)
589    spalte2="$punkte$job_name"
590    printf "| %-20s%40s | \n" "$spalte1" "${spalte2: -40}"
[1]591    printf "#--------------------------------------------------------------# \n\n"
592
593
[1289]594       # QUERY CHECK
[1]595    antwort="dummy"
596    while [[ $antwort != y  &&  $antwort != Y  &&  $antwort != n  &&  $antwort != N ]]
597    do
598       read antwort?" >>> continue (y/n) ? "
599    done
600    if [[ $antwort = n  ||  $antwort = N ]]
601    then
602       locat=verify; exit
603    fi
604    printf "\n"
605 fi
606
607
608
[1289]609    # GENERATE RANDOM IDENTIFIER, AND DETERMINE THE JOBNAME ON THE TARGET HOST
610 identifier=$RANDOM
611 job_on_remhost=${job_name}_${identifier}_$local_host
612 job_to_send=job_to_send_$identifier
[1]613 if [[ $delete_dayfile = false ]]
614 then
[1289]615    remote_dayfile=${local_host}_${job_name}_result_$identifier
[1]616    local_dayfile=${remote_host}_${job_name}
617 else
618    remote_dayfile=/dev/null
619 fi
620
621
622
[1289]623    # GENERATE THE BATCH-JOB SCRIPTS (FOR QUEUEING-SYSTEMS qsub/msub/LoadLeveler)
[1]624 if [[ $(echo $remote_host | cut -c1-3) = ibm  &&  $numprocs != 0 ]]
625 then
626
[1289]627       # GENERAL LOADLEVELER SETTINGS
[622]628    execute_in_shell="#!/bin/ksh"
629    use_shell="# @ shell = /bin/ksh"
630    consumable_memory="ConsumableMemory($memory mb)"
631    class="# @ class = $queue"
632    environment="# @ environment = OMP_NUM_THREADS=$threads_per_task; MP_SHARED_MEMORY=yes"
633    network_to_use="# @ network.mpi = sn_all,shared,us"
634    data_limit="# @ data_limit = 1.76gb"
635    image_size="# @ image_size = 50"
[693]636    wall_clock_limit="# @ wall_clock_limit = ${timestring},$timestring"
[312]637
[693]638    if [[ $email_notification = none ]]
639    then
640       notify_user=""
641    else
642       notify_user="# @ notify_user = $email_notification"
643       if [[ $delete_dayfile = true ]]
644       then
645          notification='# @ notification = never'
646       fi
647    fi
[622]648
[312]649    if [[ $remote_host = ibmh ]]
[1]650    then
[312]651       data_limit=""
652       network_to_use=""
[1184]653       class="# @ class = $queue"
[312]654       environment=""
[814]655       rset="# @ rset = RSET_MCM_AFFINITY"
656       task_affinity="# @ task_affinity = core(1)"
[693]657    elif [[ $remote_host = ibmkisti ]]
658    then
659       network_to_use="# @ network.MPI = sn_all,shared,US"
660       wall_clock_limit="# @ wall_clock_limit = $timestring"
[696]661       if [[ $threads_per_task = 1 ]]
662       then
663          rset="# @ rset = RSET_MCM_AFFINITY"
664          mcm_affinity_options="# @ mcm_affinity_options = mcm_mem_pref mcm_sni_none mcm_distribute"
665       fi
[693]666       environment=""
667       use_shell=""
668       data_limit=""
669       image_size=""
[622]670    elif [[ $remote_host = ibmku ]]
671    then
672       execute_in_shell="#!/usr/bin/ksh"
673       use_shell="# @ shell = /usr/bin/ksh"
674       consumable_memory=""
675       environment=""
676       network_to_use="# @ network.mpi = sn_all,shared,us"
677       data_limit=""
678       image_size=""
679    elif [[ $remote_host = ibms ]]
680    then
681       network_to_use="# @ network.mpi = csss,shared,us"
[1]682    fi
683
684    cat > $job_to_send << %%END%%
[622]685$execute_in_shell
686$use_shell
[1]687
688# @ job_type = parallel
[693]689# @ job_name = $job_name
[1]690# @ resources = ConsumableCpus($threads_per_task) $consumable_memory
691# @ output = $remote_dayfile
692# @ error = $remote_dayfile
[693]693$wall_clock_limit
[622]694$image_size
[312]695$class
696$environment
[1]697$network_to_use
698$data_limit
[693]699$rset
700$mcm_affinity_options
[814]701$task_affinity
[1]702$notification
[693]703$notify_user
[1]704
705%%END%%
706
707    if (( nodes > 0 ))
708    then
709
[693]710       if [[ $remote_host != ibmkisti ]]
711       then
712
713          cat >> $job_to_send << %%END%%
[1]714# @ node = $nodes
[622]715# @ tasks_per_node = $processes_per_node
[1]716# @ node_usage = $node_usage
717# @ queue
718
719%%END%%
720
[693]721       else
722
723          cat >> $job_to_send << %%END%%
[696]724# @ total_tasks = $mpi_tasks
[693]725# @ blocking = unlimited
726# @ queue
727
728%%END%%
729
730       fi
731
[1]732    else
733
[1289]734       cat >> $job_to_send << %%END%%
[1]735# @ blocking = unlimited
736# @ total_tasks = $numprocs
737# @ node_usage = $node_usage
738# @ queue
739
740%%END%%
741
742    fi
743
[1289]744       # WORKAROUND BECAUSE OF SILLY JOB FILTER ON ibmkisti
[696]745    if [[ $remote_host = ibmkisti  &&  $threads_per_task != 1 ]]
746    then
747       echo  "export OMP_NUM_THREADS=$threads_per_task"  >>  $job_to_send
748    fi
749
[1]750 elif [[ $(echo $remote_host | cut -c1-3) = ibm  &&  $numprocs = 0 ]]
751 then
752
753    cat > $job_to_send << %%END%%
754#!/bin/ksh
755
756# @ job_type = serial
757# @ node_usage = $node_usage
758# @ job_name = palm
759# @ wall_clock_limit = ${timestring},$timestring
760# @ resources = ConsumableCpus(1) ConsumableMemory(1 gb)
761# @ output = $remote_dayfile
762# @ error = $remote_dayfile
[312]763$class
[1]764$notification
765
766# @ queue
767
768%%END%%
769
[1274]770 elif [[ $remote_host = lccrayb || $remote_host = lccrayh ]]
[1224]771 then
772
773    if [[ $numprocs != 0 ]]
774    then
775       cat > $job_to_send << %%END%%
[1255]776#!/bin/bash
[1224]777#PBS -N $job_name
778#PBS -l walltime=$timestring
[1264]779#PBS -l nodes=$nodes:ppn=$processes_per_node
[1224]780#PBS -o $remote_dayfile
781#PBS -j oe
782#PBS -q $queue
783
784$module_calls
785
786%%END%%
787
788    else
789
790       continue
791
792    fi
793
[1197]794 elif [[ $remote_host = lccrayf ]]
795 then
796
797    if [[ $numprocs != 0 ]]
798    then
799       cat > $job_to_send << %%END%%
800#!/bin/bash -l
801#SBATCH -J $job_name
802#SBATCH -t $timestring
803#SBATCH -N $nodes
804#SBATCH --ntasks-per-node=$processes_per_node
805#SBATCH -p $queue
806#SBATCH -o $remote_dayfile
807#SBATCH -e $remote_dayfile
808
809$init_cmds
810$module_calls
811
812%%END%%
813
814    else
815       cat > $job_to_send << %%END%%
816#!/bin/bash -l
817#SBATCH -J $job_name
818#SBATCH -t $timestring
819#SBATCH -l ncpus=1
820#SBATCH -l pmem=${memory}mb
821#SBATCH -m abe
822#SBATCH -o $remote_dayfile
823#SBATCH -e $remote_dayfile
824
825$init_cmds
826$module_calls
827
828%%END%%
829
830    fi
831
[1099]832 elif [[ $remote_host = lcflow ]]
833 then
[1202]834    if [ $memory -gt 1800 ]; then
[1099]835      use_himem=""
836    else
837      use_himem="#"
838    fi
839    if [[ $numprocs != 0 ]]
840    then
[1202]841      pe_set="#$ -pe impi41 $numprocs"
[1099]842    else
[1202]843      pe_set="#$ -pe impi41 1"
[1099]844    fi
845    if [[ $queue = default ]]
846    then
847      queue_set=""
848    else
849      queue_set="#$ -q $queue"
850    fi
851    [[ "$disc_space" = "" ]]  &&  disc_space=50
852
853       cat > $job_to_send << %%END%%
854#!/bin/bash
855#$ -S /bin/bash
856#$ -N $job_name
857#$ -cwd
858#$ -l h_rt=$timestring
859#$ -l h_vmem=${memory}M
860#$ -o $remote_dayfile
861#$ -j y
862$pe_set
863#$ -R y
864${use_himem}#$ -l highmem=true
865#$ -l h_fsize=${disc_space}G
866$queue_set
867
868%%END%%
869
[1021]870 elif [[ $remote_host = lck || $remote_host = lckordi || $remote_host = lcsb ]]
[368]871 then
872
873    if [[ $numprocs != 0 ]]
874    then
875       cat > $job_to_send << %%END%%
876#!/bin/ksh
877#PBS -N $job_name
878#PBS -l walltime=$timestring
879#PBS -l ncpus=$numprocs
880#PBS -l pmem=${memory}mb
881#PBS -o $remote_dayfile
[1021]882#PBS -l nodes=$nodes:ppn=${processes_per_node}
[368]883#PBS -j oe
884
885mpd &
886
887%%END%%
888
889    else
890       cat > $job_to_send << %%END%%
891#!/bin/ksh
892#PBS -N $job_name
893#PBS -l walltime=$timestring
894#PBS -l ncpus=1
895#PBS -l pmem=${memory}mb
896#PBS -o $remote_dayfile
897#PBS -j oe
898
899%%END%%
900
901    fi
902
[1040]903 elif [[ $remote_host = lckiaps ]]
904 then
905
906    if [[ $numprocs != 0 ]]
907    then
908       cat > $job_to_send << %%END%%
909#!/bin/ksh
910#PBS -N $job_name
911#PBS -l walltime=$timestring
912#PBS -l select=1:ncpus=$numprocs
913#PBS -l pmem=${memory}mb
914#PBS -q $queue
915#PBS -o $remote_dayfile
916#PBS -j oe
917#PBS -V
918
919%%END%%
920
921    else
922       cat > $job_to_send << %%END%%
923#!/bin/ksh
924#PBS -N $job_name
925#PBS -l walltime=$timestring
926#PBS -l ncpus=1
927#PBS -l pmem=${memory}mb
928#PBS -o $remote_dayfile
929#PBS -j oe
930
931%%END%%
932
933    fi
934
[693]935 elif [[ $remote_host = lcyon ]]
936 then
937
938    if [[ $numprocs != 0 ]]
939    then
940       cat > $job_to_send << %%END%%
941#!/bin/ksh
942#PBS -N $job_name
943#PBS -l walltime=$timestring
944#PBS -l ncpus=$numprocs
945#PBS -l pmem=${memory}mb
946#PBS -o $remote_dayfile
947#PBS -j oe
948
949%%END%%
950
951    else
952       cat > $job_to_send << %%END%%
953#!/bin/ksh
954#PBS -N $job_name
955#PBS -l walltime=$timestring
956#PBS -l ncpus=1
957#PBS -l pmem=${memory}mb
958#PBS -o $remote_dayfile
959#PBS -j oe
960
961%%END%%
962
963    fi
964
[892]965 elif [[ $remote_host = lcxe6 ]]
[164]966 then
967
968    if [[ $numprocs != 0 ]]
969    then
970       cat > $job_to_send << %%END%%
[799]971#!/bin/ksh
[164]972#PBS -S /bin/ksh
973#PBS -N $job_name
[552]974#PBS -A $project_account
[206]975#PBS -j oe
[164]976#PBS -l walltime=$timestring
977#PBS -l mppwidth=${numprocs}
[622]978#PBS -l mppnppn=${processes_per_node}
[164]979#PBS -m abe
980#PBS -o $remote_dayfile
[492]981$email_directive
[164]982
[892]983$init_cmds
[493]984$module_calls
[343]985
[164]986%%END%%
987
[108]988    else
989       cat > $job_to_send << %%END%%
990#!/bin/ksh
[168]991#PBS -S /bin/ksh
[108]992#PBS -N $job_name
[552]993#PBS -A $project_account
[206]994#PBS -j oe
[108]995#PBS -l walltime=$timestring
996#PBS -l ncpus=1
997#PBS -l pmem=${memory}mb
998#PBS -m abe
[492]999$email_directive
[108]1000#PBS -o $remote_dayfile
1001
[892]1002$init_cmds
[493]1003$module_calls
[343]1004
[108]1005%%END%%
1006
1007    fi
1008
[440]1009 elif [[ $remote_host = lckyoto ]]
[437]1010 then
1011
[440]1012       cat > $job_to_send << %%END%%
[799]1013#!/bin/ksh
[440]1014# @\$-o $remote_dayfile
1015# @\$-eo -oi
1016# @\$-lP 16
[799]1017# @\$-lp 1
[440]1018# @\$-lm 28gb  -llm unlimited -ls unlimited
1019# @\$-q $queue
1020# @\$-Pvn abs_pack
1021##for intel? @\$-Pvn abs_unpack -Pvs unpack -Pvc unpack
1022#. /thin/local/etc/setprofile/intel-11.0.sh
1023#. /thin/local/etc/setprofile/mvapich2-1.4+intel-11.0.sh
1024. ~/.myprofile
1025#. /home2/t/t51254/palm/current_version/myprofile
1026#. /thin/apps/pgi/mpi.sh
1027#
1028env
1029#
1030set -x
1031
1032%%END%%
1033
1034 elif [[ $remote_host = lcxt5m ]]
1035 then
1036
[437]1037    if [[ $numprocs != 0 ]]
1038    then
1039       cat > $job_to_send << %%END%%
[799]1040#!/bin/ksh
[437]1041#PBS -S /bin/ksh
1042#PBS -N $job_name
1043#PBS -j oe
1044#PBS -l walltime=$timestring
1045#PBS -l mppwidth=${numprocs}
[622]1046#PBS -l mppnppn=${processes_per_node}
[437]1047#PBS -m abe
1048#PBS -o $remote_dayfile
1049
[892]1050$init_cmds
[493]1051$module_calls
[437]1052
1053%%END%%
1054
1055    else
1056       cat > $job_to_send << %%END%%
1057#!/bin/ksh
1058#PBS -S /bin/ksh
1059#PBS -N $job_name
1060#PBS -j oe
1061#PBS -l walltime=$timestring
1062#PBS -l ncpus=1
1063#PBS -l pmem=${memory}mb
1064#PBS -m abe
1065#PBS -o $remote_dayfile
1066
[892]1067$init_cmds
[493]1068$module_calls
[437]1069
1070%%END%%
1071
1072    fi
1073
[1099]1074 elif [[ $remote_host = lckyuh ]]
1075 then
1076    cat > $job_to_send << %%END%%
1077#!/bin/bash
1078#PJM -L "rscgrp=$queue"
1079#PJM -L "node=$nodes"
1080#PJM --mpi "proc=$numprocs"
1081#PJM -L "elapse=$timestring"
1082#PJM -o $remote_dayfile
1083#PJM -j
1084#PJM -X
1085#PJM --no-stging
1086
1087export LANG=en_US.UTF-8
1088%%END%%
1089
[1090]1090 elif [[ $remote_host = lckyut ]]
1091 then
1092    cat > $job_to_send << %%END%%
1093#!/bin/bash
[1099]1094#PJM -L "rscgrp=$queue"
[1090]1095#PJM -L "vnode=$numprocs"
1096#PJM -L "vnode-core=1"
1097#PJM -L "elapse=$timestring"
1098#PJM --mpi proc=$numprocs
1099#PJM -o $remote_dayfile
1100#PJM -j
[1099]1101#PJM -X
1102#PJM --no-stging
[1090]1103
1104export LANG=en_US.UTF-8
1105%%END%%
1106
[1]1107 elif [[ $remote_host = nech ]]
1108 then
1109
1110    if (( nodes > 1 ))
1111    then
1112       cat > $job_to_send << %%END%%
1113#!/bin/ksh
[622]1114#PBS -l cpunum_prc=$processes_per_node,cputim_job=$cputime
[1]1115#PBS -l ${qsubmem}=${Memory}gb
1116#PBS -b $nodes
1117#PBS -o $remote_dayfile
1118#PBS -N palm
1119#PBS -j o
1120#PBS -T mpisx
1121
1122%%END%%
1123
1124    elif [[ $numprocs != 0 ]]
1125    then
1126       cat > $job_to_send << %%END%%
1127#!/bin/ksh
[622]1128#PBS -l cpunum_prc=$processes_per_node,cputim_job=$cputime
[1]1129#PBS -l ${qsubmem}=${Memory}gb
1130#PBS -o $remote_dayfile
1131#PBS -N palm
1132#PBS -j o
1133
1134%%END%%
1135
1136    else
1137       cat > $job_to_send << %%END%%
1138#!/bin/ksh
1139#PBS -l ${qsubmem}=${Memory}gb,${qsubtime}=$cputime
1140#PBS -o $remote_dayfile
1141#PBS -j o
1142
1143%%END%%
1144
1145    fi
1146
1147 elif [[ $remote_host = lctit ]]
1148 then
1149    cat > $job_to_send << %%END%%
[635]1150#!/bin/ksh
[892]1151$init_cmds
[678]1152$module_calls
1153
[1]1154%%END%%
1155
[1289]1156       # SET OPTIONS FOR SUBMIT-COMMAND
[678]1157    if [[ $tasks_per_node != $processes_per_node ]]
1158    then
[1378]1159       submcom="$submcom -W group_list=$group_number -N $job_name -l walltime=$timestring -l select=$nodes:ncpus=$processes_per_node:mpiprocs=$tasks_per_node:mem=${Memory}gb -o $remote_dayfile -j oe -et 1 -q $queue "
[678]1160    else
[1378]1161       submcom="$submcom -W group_list=$group_number -N $job_name -l walltime=$timestring -l select=$nodes:ncpus=$processes_per_node:mpiprocs=$tasks_per_node:mem=${Memory}gb -l place=scatter -o $remote_dayfile -j oe -et 1 -q $queue "
[678]1162    fi
[1]1163
1164 else
1165
1166    cat > $job_to_send << %%END%%
1167# @\$-q ${queue}
1168# @\$-l${qsubtime} $timestring
1169# @\$-l${qsubmem} ${memory}mb
1170# @\$-o $remote_dayfile
1171# @\$-eo
1172
1173%%END%%
1174
1175 fi
1176
1177
[1289]1178    # IN CASE OF JOBS EXECUTING ON REMOTE-HOSTS, THE TRANSFER OF THE DAYFILES
1179    # TO THE LOCAL HOSTS WILL BE INITIATED BY TRAP ON EXIT
1180    # NO TRANSFER POSSIBLE ON IBM IN SEOUL
[1]1181 if [[ $delete_dayfile = false  &&  $remote_host != $local_host ]]
1182 then
1183    echo "set +vx"                              >>  $job_to_send
1184    echo "trap '"                               >>  $job_to_send
1185    echo "set +vx"                              >>  $job_to_send
[1468]1186    if [[ $(echo $remote_host | cut -c1-3) = ibm  ||  $remote_host = lccrayb  ||  $remote_host = lccrayh  ||  $(echo $remote_host | cut -c1-3) = nec  ||  $remote_host = lcflow  ||  $remote_host = lckiaps  ||  $remote_host = lckyu* || $remote_host = lcxe6 ]]
[1]1187    then
[622]1188       if [[ $remote_host = ibmh ]]
[1]1189       then
1190          return_queue=c1
[693]1191       elif [[ $remote_host = ibmkisti ]]
1192       then
1193          return_queue=class.1-2
[622]1194       elif [[ $remote_host = ibmku ]]
1195       then
1196          return_queue=sdbg2
[1]1197       elif [[ $remote_host = ibms ]]
1198       then
1199          return_queue=p_normal
[1274]1200       elif [[ $remote_host = lccrayb || $remote_host = lccrayh ]]
[1255]1201       then
1202          return_queue=dataq
[1468]1203       elif [[ $remote_host = lcxe6 ]]
1204       then
1205          return_queue=debug
[1040]1206       elif [[ $remote_host = lckiaps ]]
1207       then
1208          return_queue=express
[1099]1209       elif [[ $remote_host = lckyuh ]]
1210       then
1211          return_queue=cx-single
[1090]1212       elif [[ $remote_host = lckyut ]]
1213       then
1214          return_queue=cx-single
[1]1215       else
1216          return_queue=unknown
1217       fi
1218
1219       if [[ $(echo $remote_host | cut -c1-3) = ibm ]]
1220       then
1221
[622]1222          if [[ $remote_host = ibmku ]]
1223          then
[1289]1224             echo "echo \"#!/usr/bin/ksh\" >> scpjob.$identifier"            >>  $job_to_send
1225             echo "echo \"# @ shell = /usr/bin/ksh\" >> scpjob.$identifier"  >>  $job_to_send
[622]1226          else
[1289]1227             echo "echo \"#!/bin/ksh\" >> scpjob.$identifier"                >>  $job_to_send
[622]1228          fi
[1289]1229          echo "echo \"# @ job_type = serial\" >> scpjob.$identifier"    >>  $job_to_send
1230          echo "echo \"# @ job_name = transfer\" >> scpjob.$identifier"  >>  $job_to_send
1231          echo "echo \"# @ resources = ConsumableCpus(1) ConsumableMemory(1 gb)\" >> scpjob.$identifier"  >>  $job_to_send
1232          echo "echo \"# @ wall_clock_limit = 00:10:00,00:10:00\" >> scpjob.$identifier "  >>  $job_to_send
1233          echo "echo \"# @ output = job_queue/last_job_transfer_protocol\" >> scpjob.$identifier"  >>  $job_to_send
1234          echo "echo \"# @ error = job_queue/last_job_transfer_protocol\" >> scpjob.$identifier"  >>  $job_to_send
[312]1235          if [[ $host != "ibmh" ]]
1236          then
[1289]1237             echo "echo \"# @ class = $return_queue\" >> scpjob.$identifier"  >>  $job_to_send
[312]1238          fi
[1289]1239          echo "echo \"# @ image_size = 10\" >> scpjob.$identifier"      >>  $job_to_send
1240          echo "echo \"# @ notification = never\" >> scpjob.$identifier" >>  $job_to_send
[1]1241
[1289]1242          echo "echo \"# @ queue\" >> scpjob.$identifier"                >>  $job_to_send
1243          echo "echo \" \" >> scpjob.$identifier"                        >>  $job_to_send
[1]1244
[1289]1245          echo "echo \"set -x\" >> scpjob.$identifier"                   >>  $job_to_send
[1468]1246          echo "echo \"batch_scp  $PORTOPT  -d  -w 10  -u $local_user  $local_address  ${job_catalog}/$remote_dayfile  \\\"$job_catalog\\\"  $local_dayfile\" >> scpjob.$identifier"  >>  $job_to_send
[622]1247          if [[ $remote_host = ibmku ]]
1248          then
[1289]1249             echo "echo \"rm  scpjob.$identifier\" >> scpjob.$identifier"   >>  $job_to_send
[622]1250          fi
[1289]1251          echo "echo \"exit\" >> scpjob.$identifier"                     >>  $job_to_send
[1]1252
1253       elif [[ $remote_host = nech ]]
1254       then
1255          echo "cd /pf/b/${remote_user}/job_queue" >>  $job_to_send
[1289]1256          echo "cat > scpjob.$identifier << %%END%%"  >>  $job_to_send
[1]1257          echo "#PBS -l ${qsubmem}=1GB,${qsubtime}=100"  >>  $job_to_send
1258          echo "#PBS -o last_job_transfer_protocol"      >>  $job_to_send
1259          echo "#PBS -j o"                         >>  $job_to_send
1260          echo " "                                 >>  $job_to_send
1261          echo "set -x"                            >>  $job_to_send
1262          echo "cd /pf/b/${remote_user}/job_queue" >>  $job_to_send
[1468]1263          echo "batch_scp  $PORTOPT  -d  -w 10  -u $local_user $local_address  $remote_dayfile  \"$job_catalog\"  $local_dayfile"  >>  $job_to_send
[1]1264          echo "[[ \"\$for_subjob_to_do\" != \"\" ]]  &&  eval \$for_subjob_to_do"  >>  $job_to_send
1265          echo "%%END%%"                           >>  $job_to_send
1266
[1099]1267       elif [[ $remote_host = lckyuh ]]
1268       then
[1289]1269          echo "cat > scpjob.$identifier << %%END%%"  >>  $job_to_send
[1099]1270          echo "#!/bin/bash"                       >>  $job_to_send
1271          echo "#PJM -L \"node=1\""                >>  $job_to_send
1272          echo "#PJM -L \"rscgrp=$return_queue\""  >>  $job_to_send
1273          echo "#PJM --no-stging"                  >>  $job_to_send
1274          echo "#PJM -L \"elapse=30:00\""          >>  $job_to_send
1275          echo "#PJM -o \$HOME/job_queue/last_job_transfer_protocol"  >>  $job_to_send
1276          echo "#PJM -j"                           >>  $job_to_send
1277          echo " "                                 >>  $job_to_send
1278          echo "export LANG=en_US.UTF-8"           >>  $job_to_send
1279          echo "set -x"                            >>  $job_to_send
[1468]1280          echo "batch_scp  $PORTOPT  -d  -w 10  -u $local_user $local_address  $remote_dayfile  \"$job_catalog\"  $local_dayfile"  >>  $job_to_send
[1099]1281          echo "[[ \"\$for_subjob_to_do\" != \"\" ]]  &&  eval \$for_subjob_to_do"  >>  $job_to_send
1282          echo "%%END%%"                           >>  $job_to_send
1283
[1090]1284       elif [[ $remote_host = lckyut ]]
1285       then
[1289]1286          echo "cat > scpjob.$identifier << %%END%%"  >>  $job_to_send
[1090]1287          echo "#!/bin/bash"                       >>  $job_to_send
1288          echo "#PJM -L \"vnode=1\""               >>  $job_to_send
1289          echo "#PJM -L \"rscgrp=$return_queue\""  >>  $job_to_send
1290          echo "#PJM --no-stging"                  >>  $job_to_send
1291          echo "#PJM -L \"elapse=30:00\""          >>  $job_to_send
1292          echo "#PJM -o \$HOME/job_queue/last_job_transfer_protocol"  >>  $job_to_send
1293          echo "#PJM -j"                           >>  $job_to_send
1294          echo " "                                 >>  $job_to_send
1295          echo "export LANG=en_US.UTF-8"           >>  $job_to_send
1296          echo "set -x"                            >>  $job_to_send
[1468]1297          echo "batch_scp  $PORTOPT  -d  -w 10  -u $local_user $local_address  $remote_dayfile  \"$job_catalog\"  $local_dayfile"  >>  $job_to_send
[1090]1298          echo "[[ \"\$for_subjob_to_do\" != \"\" ]]  &&  eval \$for_subjob_to_do"  >>  $job_to_send
1299          echo "%%END%%"                           >>  $job_to_send
1300
[1274]1301       elif [[ $remote_host = lccrayb || $remote_host = lccrayh ]]
[1255]1302       then
[1289]1303          echo "cat > scpjob.$identifier << %%END%%"        >>  $job_to_send
[1255]1304          echo "#!/bin/bash"                             >>  $job_to_send
1305          echo "#PBS -N job_protocol_transfer"           >>  $job_to_send
1306          echo "#PBS -l walltime=00:30:00"               >>  $job_to_send
[1262]1307          echo "#PBS -l nodes=1:ppn=1"                   >>  $job_to_send
[1255]1308          echo "#PBS -o \$HOME/job_queue/last_job_transfer_protocol"      >>  $job_to_send
1309          echo "#PBS -j oe"                              >>  $job_to_send
1310          echo " "                                       >>  $job_to_send
1311          echo "set -x"                                  >>  $job_to_send
[1468]1312          echo "batch_scp  $PORTOPT  -d  -w 10  -u $local_user $local_address  ${job_catalog}/$remote_dayfile  \"$job_catalog\"  $local_dayfile"  >>  $job_to_send
[1255]1313          echo "[[ \"\$for_subjob_to_do\" != \"\" ]]  &&  eval \$for_subjob_to_do"  >>  $job_to_send
1314          echo "%%END%%"                                 >>  $job_to_send
1315
[1099]1316       elif [[ $remote_host = lcflow ]]
1317       then
[1289]1318          echo "cat > scpjob.${identifier}.tmp << %%END%%"                  >>  $job_to_send
[1099]1319          echo "#!/bin/bash"                                             >>  $job_to_send
1320          echo "SGEPREFIX -S /bin/bash"                                  >>  $job_to_send
1321          echo "SGEPREFIX -N transfer_$job_name"                         >>  $job_to_send
1322          echo "SGEPREFIX -cwd"                                          >>  $job_to_send
1323          echo "SGEPREFIX -l h_rt=01:00:00"                              >>  $job_to_send
[1202]1324          echo "SGEPREFIX -l h_vmem=500M"                                >>  $job_to_send
1325          echo "SGEPREFIX -l excl_flow=false"                            >>  $job_to_send
[1099]1326          echo "SGEPREFIX -j y"                                          >>  $job_to_send
[1289]1327          echo "SGEPREFIX -o ${local_host}_${job_name}_scpjob_$identifier"  >>  $job_to_send 
[1099]1328          echo " "                                                       >>  $job_to_send 
1329          echo "set -x"                                                  >>  $job_to_send 
1330          echo "export PALM_BIN=$PALM_BIN" | sed -e 's:'$HOME':$HOME:'   >>  $job_to_send
1331          echo "export PATH=\$PATH:\$PALM_BIN"                           >>  $job_to_send
1332          echo ""                                 >>  $job_to_send         
[1468]1333          echo "batch_scp  $PORTOPT  -d  -w 10  -u $local_user $local_address  ${job_catalog}/$remote_dayfile  \"$job_catalog\"  $local_dayfile"  >>  $job_to_send
[1099]1334          echo "[[ \"\$for_subjob_to_do\" != \"\" ]]  &&  eval \$for_subjob_to_do"  >>  $job_to_send
[1289]1335          echo "rm -f scpjob.${identifier}"                                 >>  $job_to_send         
[1099]1336          echo "%%END%%"                                                 >>  $job_to_send
[1289]1337          echo "sed -e 's/SGEPREFIX/#$/g' scpjob.${identifier}.tmp > scpjob.${identifier}" >>  $job_to_send         
1338          echo "rm -f scpjob.${identifier}.tmp"                             >>  $job_to_send         
[1468]1339       elif [[ $remote_host = lcxe6 ]]
1340       then
1341          echo "cat > scpjob.${identifier}  << %%END%%"  >>  $job_to_send
1342          echo "#!/bin/ksh"                              >>  $job_to_send
1343          echo "#PBS -N job_protocol_transfer"           >>  $job_to_send
1344          echo "#PBS -l walltime=00:30:00"               >>  $job_to_send
1345          echo "#PBS -A $project_account"                >>  $job_to_send
1346          echo "#PBS -l mppwidth=1"                      >>  $job_to_send
1347          echo "#PBS -l mppnppn=1"                       >>  $job_to_send
1348          echo "#PBS -o \$HOME/job_queue/last_job_transfer_protocol"  >>  $job_to_send
1349          echo "#PBS -j oe"                              >>  $job_to_send
1350          echo " "                                       >>  $job_to_send
1351          echo "set -x"                                  >>  $job_to_send
1352          echo "batch_scp  $PORTOPT  -d  -w 10  -u $local_user $local_address  ${job_catalog}/$remote_dayfile  \"$job_catalog\"  $local_dayfile"  >>  $job_to_send
1353          echo "[[ \"\$for_subjob_to_do\" != \"\" ]]  &&  eval \$for_subjob_to_do"  >>  $job_to_send
1354          echo "%%END%%"                                 >>  $job_to_send
[1]1355       else
1356
[1289]1357          echo "cat > scpjob.$identifier << %%END%%"  >>  $job_to_send
[1]1358          echo "# @\\\$-q $return_queue"           >>  $job_to_send
1359          echo "# @\\\$-l${qsubtime} 10"           >>  $job_to_send
1360          echo "# @\\\$-l${qsubmem} 10mb"          >>  $job_to_send
1361          if [[ $remote_host = t3ej2  ||  $remote_host = t3ej5  ||  $remote_host = t3es ]]
1362          then
1363             echo "# @\$-l mpp_p=0"                >>  $job_to_send
1364          fi
1365          echo '# @\$-lF 10mb'                     >>  $job_to_send
1366          echo '# @\$-o job_queue/last_job_transfer_protocol'    >>  $job_to_send
1367          echo '# @\\\$-eo'                          >>  $job_to_send
1368          echo " "                                 >>  $job_to_send
1369          if [[ $remote_host = t3ej2  ||  $remote_host = t3ej5 ]]
1370          then
1371             echo "set +vx"                        >>  $job_to_send
1372             echo ". .profile"                     >>  $job_to_send
1373          fi
1374          echo "set -x"                            >>  $job_to_send
[1468]1375          echo "batch_scp  $PORTOPT  -d  -w 10  -u $local_user $local_address  ${job_catalog}/$remote_dayfile  \"$job_catalog\"  $local_dayfile  >  /dev/null"  >>  $job_to_send
[1]1376          echo "[[ \"\$for_subjob_to_do\" != \"\" ]]  &&  eval \$for_subjob_to_do"  >>  $job_to_send
1377          echo "%%END%%"                           >>  $job_to_send
1378       fi
1379
1380       if [[ $(echo $remote_host | cut -c1-3) = ibm ]]
1381       then
[1289]1382          echo "llsubmit  scpjob.$identifier"      >>  $job_to_send
[1274]1383       elif [[ $remote_host = lccrayb || $remote_host = lccrayh ]]
[1255]1384       then
[1289]1385          echo "msub -q $return_queue  scpjob.$identifier"               >>  $job_to_send
[1]1386       elif [[ $remote_host = t3eb  ||  $remote_host = t3eh  ||  $remote_host = t3ej2  ||  $remote_host = t3ej5 ]]
1387       then
[1289]1388          echo "qsub -J n  scpjob.$identifier"     >>  $job_to_send
[1]1389       elif [[ $remote_host = t3es ]]
1390       then
[1289]1391          echo "qsub -J n  -s /bin/ksh  scpjob.$identifier"     >>  $job_to_send
[1043]1392       elif [[ $remote_host = lckiaps ]]
1393       then
[1289]1394          echo "mv  scpjob.$identifier  $job_catalog"           >>  $job_to_send
[1468]1395          echo "ssh $SSH_PORTOPT ${remote_username}@${remote_address}  \"$submcom ${job_catalog}/scpjob.$identifier\" "  >>  $job_to_send
[1289]1396          echo "rm  ${job_catalog}/scpjob.$identifier"          >>  $job_to_send
[1099]1397       elif [[ $remote_host = lckyu* ]]
[1090]1398       then
[1468]1399          echo "scp $PORTOPT scpjob.$identifier  ${remote_username}@${remote_address}:job_queue"           >>  $job_to_send
1400          echo "ssh $SSH_PORTOPT ${remote_username}@${remote_address}  \"cd job_queue; $submcom scpjob.$identifier; rm scpjob.$identifier\" "  >>  $job_to_send
[1099]1401       elif [[ $remote_host = lcflow ]]
1402       then
[1289]1403          echo "mv  scpjob.$identifier  $job_catalog"           >>  $job_to_send
[1468]1404          echo "/usr/bin/ssh ${remote_username}@${remote_address}  \"$init_cmds $module_calls cd $job_catalog; $submcom scpjob.$identifier\" "  >>  $job_to_send
[1]1405       else
[1289]1406          echo "$submcom  scpjob.$identifier"      >>  $job_to_send
[1]1407       fi
[1043]1408       if [[ $remote_host != ibmku  &&  $remote_host != lckiaps ]]
[622]1409       then
[1289]1410          echo "rm  scpjob.$identifier"            >>  $job_to_send
[622]1411       fi
[1]1412       if [[ $remote_host = nech ]]
1413       then
1414          echo "cd -"                           >>  $job_to_send
1415       fi
1416    else
[1468]1417#       echo "ftpcopy  -d  $local_address  ${job_catalog}/$remote_dayfile  \"$job_catalog\"  $local_dayfile"  >>  $job_to_send
[1289]1418       # ??? funktioniert das ÃŒberhaupt noch ???
[1468]1419       echo "nohup  ftpcopy  -d  -w 15  $local_address  ${job_catalog}/$remote_dayfile  \"$job_catalog\"  $local_dayfile  >  /dev/null  &"  >>  $job_to_send
[1]1420    fi
1421    echo "set -x"                               >>  $job_to_send
1422    echo "     ' exit"                          >>  $job_to_send
1423    echo "set -x"                               >>  $job_to_send
1424 fi
1425
1426
1427
[1289]1428    # APPEND THE JOB-FILE (CREATE BY mrun) TO THE JOB-DIRECTIVES GENERATED ABOVE
[1]1429 cat  $file_to_send  >>  $job_to_send
[69]1430
[1]1431 if [[ $remote_host = ibm ]]
1432 then
1433    echo " "         >>  $job_to_send
1434    echo "exit"      >>  $job_to_send
1435 fi
[635]1436
[1289]1437    # REMOVE JOB-FILE
[1099]1438 if [[ $remote_host = lctit  ||  $remote_host = ibmku  ||  $remote_host = lcflow ]]
[69]1439 then
1440    echo " "                               >>  $job_to_send
1441    echo "rm ~/job_queue/$job_on_remhost"  >>  $job_to_send
1442 fi
[1]1443
1444
[1289]1445    # TRANSFER JOB TO THE TARGET HOST (JOB-DIRECTORY)
[1]1446 if [[ $no_submit = false ]]
1447 then
1448    if [[ $remote_host != $local_host ]]
1449    then
1450       [[ $verify = true ]]  &&  printf "\n >>> transfering job to \"$remote_host\"..."
[1289]1451       if [[ $remote_host = ibms ]]    # ssh on ibms cannot handle "~/"
[1]1452       then
[82]1453          job_catalog_save=$job_catalog
1454          job_catalog=job_queue
1455       elif [[ $remote_host = nech ]]
1456       then
1457          job_catalog_save=$job_catalog
1458          job_catalog=/hpf/b/${remote_user}/job_queue
1459       fi
[1096]1460       if [[ $remote_host = nech ]]
[82]1461       then
[1289]1462             # FILES CAN ONLY BE TRANSFERED VIA DKRZ'S ARCHIVE-SERVER
[1094]1463          scp  $PORTOPT  $job_to_send  ${remote_user}@136.172.44.205:${job_catalog}/$job_on_remhost
[1]1464       else
[1468]1465          scp  $PORTOPT  $job_to_send  ${remote_user}@${remote_address}:${job_catalog}/$job_on_remhost
[1]1466       fi
[82]1467       if [[ $? = 1 ]]
1468       then
1469          locat=scp; exit
1470       fi
1471       if [[ $remote_host = ibms ]]
1472       then
1473          job_catalog=$job_catalog_save
1474       fi
[1]1475       [[ $verify = true ]]  &&  printf "\n >>> finished\n"
1476    else
1477       eval  job_catalog=$job_catalog
1478       cp  $job_to_send  ${job_catalog}/$job_on_remhost
1479    fi
1480
1481
1482
[1289]1483       # START NQS- / LOADLEVELER-JOB
[1]1484    if [[ $remote_host != $local_host ]]
1485    then
1486       [[ $verify = true ]]  &&  printf "\n >>> submitting job using \"qsub\"...\n"
[635]1487
[1289]1488       if [[ $remote_host = ibmku ]]
[1]1489       then
[1468]1490          ssh  $SSH_PORTOPT $remote_address  -l $remote_user  "cd $job_catalog; $submcom $job_on_remhost"
[1099]1491       elif [[ $remote_host = lcflow ]]
1492       then
[1468]1493          /usr/bin/ssh  $SSH_PORTOPT $remote_address  -l $remote_user  "$init_cmds $module_calls cd $job_catalog; $submcom $job_on_remhost"
[1]1494       else
[1468]1495          ssh  $SSH_PORTOPT $remote_address  -l $remote_user  "cd $job_catalog; $submcom $job_on_remhost; rm $job_on_remhost"
[82]1496       fi
[1]1497
1498       [[ $verify = true ]]  &&  printf " >>> o.k.\n"
1499    else
1500       cd  $job_catalog
[1289]1501       if [[ $(echo $local_host | cut -c1-3) = ibm  ||  $(echo $local_host | cut -c1-6) = lccray ]]
[1]1502       then
1503          eval  $submcom  $job_on_remhost
[1289]1504       elif [[  $local_host = lctit  ||  $localhost = lcxe6  ||  $localhost = lck  || $localhost = lckordi ||  $localhost = lcyon || $localhost = lcsb  ||  $localhost = lckyu* ]]
[108]1505       then
[635]1506          chmod  u+x  $job_on_remhost
[108]1507          eval  $submcom  $job_on_remhost
[1]1508       elif [[ $local_host = nech ]]
1509       then
1510          if [[ $queue = default ]]
1511          then
[799]1512             eval  $submcom  $job_on_remhost
[1]1513          else
[799]1514             eval  $submcom  -q $queue  $job_on_remhost
[1]1515          fi
1516       else
1517          qsub  $job_on_remhost
1518       fi
[622]1519
[1289]1520          # JOBFILE MUST NOT BE DELETED ON lctit/ibmku/lcflow. THIS WILL BE DONE
1521          # AT THE END OF THE JOB
[1099]1522       if [[ $local_host != lctit  &&  $local_host != ibmku  &&  $local_host != lcflow ]]
[622]1523       then
1524          rm  $job_on_remhost
1525       fi
[1]1526       cd  -  > /dev/null
1527    fi
1528 fi
1529
1530
1531
[1289]1532    # FINAL ACTIONS
[1]1533 if [[ $no_submit = false ]]
1534 then
[1099]1535    rm  -f $job_to_send
[1]1536 fi
[1266]1537 [[ $verify = true ]]  &&  printf "\n\n *** SUBJOB finished \n\n"
Note: See TracBrowser for help on using the repository browser.