source: palm/trunk/SCRIPTS/subjob @ 1547

Last change on this file since 1547 was 1547, checked in by witha, 10 years ago

adjustments for ForWind? computing cluster (lcflow)

  • Property svn:keywords set to Id Rev
File size: 52.3 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# ------------------
[1547]25# adjustments for ForWind computing cluster (lcflow)
[1351]26#
[1046]27# Former revisions:
28# -----------------
[169]29# $Id: subjob 1547 2015-01-29 15:09:12Z witha $
[1200]30#
[1546]31# 1545 2015-01-29 06:52:23Z heinze
32# local host name for blizzard further specified
33#
[1481]34# 1480 2014-10-17 14:41:49Z raasch
35# adjustments for 2nd stage of HLRNIII
36#
[1469]37# 1468 2014-09-24 14:06:57Z maronga
38# Typo removed (addres->address)
39# Adjustments for lcxe6
40#
[1453]41# 1452 2014-08-22 09:41:06Z heinze
42# local hosts for blizzard added
43#
[1451]44# 1450 2014-08-21 07:31:51Z heinze
45# HLRN-III (lccrayb): testq queue adjusted to mpp1testq
46#
[1443]47# 1442 2014-07-28 07:09:10Z raasch
48# HLRN-III (lccrayb/lccrayh) queues adjusted
49#
[1379]50# 1378 2014-04-28 06:04:58Z raasch
51# -et option added for lctit
52#
[1351]53# 1350 2014-04-04 13:01:30Z maronga
54# location of qsub updated for lcxe6
55#
[1290]56# 1289 2014-03-04 07:12:34Z raasch
57# German comments translated to English
58# fimm-, necriam-, scirocco-, ibmy-, and sgi-specific code removed
59#
[1280]60# 1279 2014-01-28 12:10:14Z raasch
61# node calculation modified due to changes in mrun (tasks_per_node must not be
62# an integral divisor of numprocs any more)
63#
[1275]64# 1274 2014-01-09 13:14:54Z heinze
65# adjustments for lccrayh
66#
[1267]67# 1266 2013-12-11 12:07:34Z heinze
68# further adjustments for lccrayb (use msub instead of qsub)
69#
[1265]70# 1264 2013-12-09 12:46:09Z fricke
71# Bugfix: Using number of nodes instead of number of processors (lccrayb)
72#
[1263]73# 1262 2013-12-09 10:57:20Z fricke
74# further adjustments for lccrayb
75#
[1261]76# 1260 2013-12-04 12:48:04Z raasch
77# jaboticaba admitted
78#
[1256]79# 1255 2013-11-07 14:43:35Z raasch
80# further adjustments for lccrayb
81#
[1225]82# 1224 2013-09-16 07:27:23Z raasch
83# first adjustments for lccrayb
84#
[1203]85# 1202 2013-07-10 16:22:07Z witha
86# adjustments for Forwind cluster (lcflow)
87#
[1200]88# 1199 2013-07-05 14:52:22Z raasch
89# adjustments for CSC Helsinki (lccrayf)
90#
[1185]91# use of cluster/express queue enabled (ibmh)
92# vinessa added (imuk)
[1047]93#
[1104]94# 1103 2013-02-20 02:15:53Z raasch
95# bash compatibility adjustments (usage of OPTIND, output formatting with printf
96# instead typeset -L/R),
97# further adjustments for lckyuh
98#
[1100]99# 2013-02-10 01:47:43Z raasch
100# adjustments for Kyushu-Univeristy computing center (lckyuh - hayaka)
101# and for Forwind cluster (lcflow)
102#
[1096]103# 1094 2013-02-03 01:52:12Z raasch
104# new option -P for explicit setting of ssh/scp port,
105# decalpha parts (yonsei) removed
106#
[1091]107# 2013-02-02 07:06:13Z raasch
[1099]108# adjustments for Kyushu-University computing center (lckyut - tatara)
[1091]109# old changelog messages removed
110#
[1047]111# 1046 2012-11-09 14:38:45Z maronga
112# code put under GPL (PALM 3.9)
113#
[1090]114# 08/07/94 - Siggi - first version finished
115# 29/06/94 - Siggi - script development started
116#--------------------------------------------------------------------------------#
117# subjob - script for automatic generation and submission of batch-job files
118#          for various batch queuing systems
119#--------------------------------------------------------------------------------#
[1]120
121
[1289]122    # VARIABLE-DECLARATIONS AND DEFAULT VALUES
[352]123 delete_dayfile=false
[799]124 email_notification=none
[122]125 group_number=none
[1]126 locat=normal
127 no_default_queue=none
128 no_submit=false
129 job_catalog="~/job_queue"
130 job_name=none
131 local_user=$LOGNAME
132 node_usage=shared
[475]133 numprocs=0
[1]134 punkte="..........................................................."
135 submcom=qsub
136 queue=default
137 remote_host=none
138 remote_user=""
139 verify=true
140
141 typeset  -i   cputime=memory=Memory=0  minuten  resttime  sekunden  stunden
[696]142 typeset  -i   inumprocs  mpi_tasks=nodes=processes_per_node=0 tasks_per_node=threads_per_task=1
[1]143
144
145
[1289]146    # ERROR HANDLING
147    # IN CASE OF EXIT:
[1]148 trap 'if [[ $locat != normal ]]
149       then
150          case  $locat  in
151             (option)  printf "\n  --> available optios can be displayed"
152                       printf " by typing:"
153                       printf "\n      \"subjob ?\" \n";;
154             (ftpcopy|parameter|scp|verify)  printf "\n";;
155             (*)       printf "\n  +++ unknown error"
156                       printf "\n      please inform S. Raasch!\n"
157          esac
158          [[ -f $job_to_send ]]  &&  rm  $job_to_send
159          printf "\n\n+++ SUBJOB killed \n\n"
160       fi' exit
161
162
[1289]163    # IN CASE OF TERMINAL-BREAK:
[1]164 trap '[[ -f $job_to_send ]]  &&  rm  $job_to_send
165       printf "\n\n+++ SUBJOB killed \n\n"
166       exit
167      ' 2
168
169
170
171
[1289]172    # DETERMINE NAME OF LOCAL HOST
[1]173 local_host=$(hostname)
174
175
176
[1289]177    # SET HOST-SPECIFIC VARIABLES VEREINBAREN (CHECK, IF LOCAL HOST
178    # IS ADMITTED AT ALL)
179    # NOTE: ONE OF THE ENTRIES FOR "lck" OR "lckordi" ALWAYS HAS TO BE
180    # COMMENT OUT, BECAUSE THE HOSTNAME (node*) IS SAME FOR BOTH MACHINES
[1]181 case  $local_host  in
[1468]182     (ambiel-lx)             local_address=134.106.74.48;  local_host=lcfor;;
183     (atmos)                 local_address=172.20.25.35;   local_host=lcide;;
184     (austru)                local_address=130.75.105.128; local_host=lcmuk;;
185     (autan)                 local_address=130.75.105.57;  local_host=lcmuk;;
186     (bora)                  local_address=130.75.105.103; local_host=lcmuk;;
187     (b04*)                  local_address=133.5.4.33;     local_host=lckyuh;;
[1545]188     (blizzard1|p0*|p1*|p2*|p3*|p4*|p5*|p6*|p7*|p8*|p9*)   local_address=136.172.40.15;  local_host=ibmh;;
189     (blizzard2|p0*|p1*|p2*|p3*|p4*|p5*|p6*|p7*|p8*|p9*)   local_address=136.172.40.16;  local_host=ibmh;;
[1468]190     (blogin*|bxc*)          local_address=130.73.233.1;   local_host=lccrayb;;
191     (hlogin*|hxc*)          local_address=130.75.4.1;     local_host=lccrayh;;
192     (breva)                 local_address=130.75.105.98;  local_host=lcmuk;;
193     (buran)                 local_address=130.75.105.58;  local_host=lcmuk;;
194     (caurus)                local_address=130.75.105.19;  local_host=lcmuk;;
195     (climate*)              local_address=165.132.26.68;  local_host=lcyon;;
196     (clogin*)               local_address=86.50.166.21;   local_host=lccrayf;;
197     (cs*)                   local_address=136.172.44.131; local_host=nech;;
198     (elephanta)             local_address=130.75.105.6;   local_host=lcmuk;;
199     (flow01)                local_address=10.141.255.71;  local_host=lcflow;;
200     (flow02)                local_address=10.141.255.72;  local_host=lcflow;;
201     (node*)                 local_address=165.132.26.61   local_host=lck;;
202   #  (node*)                 local_address=210.219.61.8    local_host=lckordi;;
203     (gaia*)                 local_address=150.183.146.24; local_host=ibmkisti;;
204     (gallego)               local_address=130.75.105.10;  local_host=lcmuk;;
205     (gregale)               local_address=130.75.105.109; local_host=lcmuk;;
206     (hababai)               local_address=130.75.105.108; local_host=lcmuk;;
207     (hayaka*)               local_address=133.5.4.33;     local_host=lckyuh;;
208     (hexagon.bccs.uib.no)   local_address=129.177.20.113; local_host=lcxe6;;
209     (hx*)                   local_address=133.3.51.11;    local_host=lckyoto;;
210     (inferno)               local_address=130.75.105.5;   local_host=lcmuk;;
211     (irifi)                 local_address=130.75.105.104; local_host=lcmuk;;
212     (jaboticaba)            local_address=150.163.25.181; local_host=lcbr;;
213     (sno)                   local_address=130.75.105.113; local_host=lcmuk;;
214     (levanto)               local_address=130.75.105.45;  local_host=lcmuk;;
215     (login*)                local_address=118.128.66.223; local_host=lckiaps;;
216     (maestro)               local_address=130.75.105.2;   local_host=lcmuk;;
217     (meller)                local_address=134.106.74.155; local_host=lcfor;;
218     (meteo-login*)          local_address=193.166.211.144;local_host=lcxt5m;;
219     (hexagon*)              local_address=129.177.20.113; local_host=lcxe6;;
220     (nobel*)                local_address=150.183.5.101;  local_host=ibms;;
221     (orkan)                 local_address=130.75.105.3;   local_host=lcmuk;;
222     (ostria)                local_address=130.75.105.106; local_host=lcmuk;;
223     (paesano)               local_address=130.75.105.46;  local_host=lcmuk;;
224     (pcj*)                  local_address=172.31.120.1;   local_host=lckyut;;
225     (pingui)                local_address=134.106.74.118; local_host=lcfor;;
226     (quanero)               local_address=130.75.105.107; local_host=lcmuk;;
227     (rte*)                  local_address=133.5.185.60;   local_host=lcrte;;
228     (shiokaze-lx)           local_address=134.106.74.123; local_host=lcfor;;
229     (sisu-login*)           local_address=86.50.166.21;   local_host=lccrayf;;
230     (solano)                local_address=130.75.105.110; local_host=lcmuk;;
231     (sugoka*)               local_address=172.31.120.1;   local_host=lckyut;;
232     (t2a*)                  local_address=10.1.6.165;     local_host=lctit;;
233     (urban*)                local_address=147.46.30.151   local_host=lcsb;;
234     (vinessa)               local_address=130.75.105.112; local_host=lcmuk;;
235     (vorias)                local_address=172.20.25.43;   local_host=lcmuk;;
236     (*.cc.kyushu-u.ac.jp)   local_address=133.5.4.129;    local_host=ibmku;;
[1]237     (*)                     printf "\n  +++ \"$local_host\" unknown";
[1255]238                             printf "\n      please contact the PALM group at IMUK";
[1]239                             locat=parameter; exit;;
240 esac
241
242
243
[1289]244    # BY DEFAULT, THE REMOTE HOST IS THE LOCAL HOST
[1]245 remote_host=$local_host
246
247
248
249
[1289]250    # READ THE SHELLSCRIPT-OPTIONS
[1094]251 while  getopts  :c:dDe:g:h:m:n:N:O:P:q:t:T:u:vX:  option
[1]252 do
253   case  $option  in
254       (c)   job_catalog=$OPTARG;;
255       (d)   delete_dayfile=true;;
256       (D)   no_submit=true;;
[352]257       (e)   email_notification=$OPTARG;;
[125]258       (g)   group_number=$OPTARG;;
[1]259       (h)   remote_host=$OPTARG;;
260       (m)   memory=$OPTARG;;
261       (n)   job_name=$OPTARG;;
262       (N)   node_usage=$OPTARG;;
263       (O)   threads_per_task=$OPTARG;;
[1094]264       (P)   scp_port=$OPTARG;;
[1]265       (q)   no_default_queue=$OPTARG;;
266       (t)   cputime=$OPTARG;;
267       (T)   tasks_per_node=$OPTARG;;
268       (u)   remote_user=$OPTARG;;
269       (v)   verify=false;;
270       (X)   numprocs=$OPTARG;;
271       (\?)  printf "\n  +++ Option $OPTARG unknown \n";
272             locat=option; exit;;
273   esac
274 done
275
276
[1289]277    # GET THE NAME OF THE JOBFILE AS NEXT ARGUMENT
[1103]278 (( to_shift = $OPTIND - 1 ))
279 shift $to_shift; file_to_send=$1
[1]280
281
[1289]282    # OUTPUT OF SHORT DESCRIPTION OF SCRIPT-OPTIONS
[1]283 if [ "$1" = "?" ]
284 then
285   (printf "\n  *** subjob can be called as follows:\n"
286    printf "\n      subjob -c.. -d -D -h.. -m.. -q.. -t.. -u.. -v  <jobfile>\n"
287    printf "\n      Description of available options:\n"
288    printf "\n      Option  Description                         Default-Value"
289    printf "\n        -c    job-input- and output-catalog       ~/job_queue"
290    printf "\n        -d    no job-protocol will be created     ---"
291    printf "\n        -D    only the job-file will be created   ---"
292    printf "\n        -h    execution host, available hosts:    $remote_host"
[1289]293    printf "\n              ibm, ibmh, ibmkisti, ibmku, ibms, lc...,"
294    printf "\n              lckiaps, lctit, nech"
[1]295    printf "\n        -m    memory demand per process in MByte  ---"
296    printf "\n        -n    jobname                             <jobdatei>"
297    printf "\n        -O    threads per task (for OpenMP usage) 1"
[1094]298    printf "\n        -P    ssh/scp port                        default port"
[1]299    printf "\n        -q    job-queue to be used                default"
300    printf "\n        -t    allowed cpu-time in seconds         ---"
301    printf "\n        -T    tasks per node (on parallel hosts)  ---"
302    printf "\n        -u    username on execution host          from .netrc"
303    printf "\n        -v    no prompt for confirmation          ---"
304    printf "\n        -X    # of processors (on parallel hosts) 1"
305    printf "\n "
306    printf "\n      The only possible positional parameter is <jobfile>:"
307    printf "\n      The complete NQS-job must be provided here."
308    printf "\n      <jobfile>=? creates this outline\n\n") | more
309    exit
310 fi
311
312
313
[1289]314    # CHECK, IF JOB-FILE HAS BEEN GIVEN AS ARGUMENT AND IF THE FILE ITSELF EXISTS
[1]315 if [[ "$file_to_send" = "" ]]
316 then
317    printf "\n  +++ job-file missing"
318    locat=parameter; exit
319 else
320    if [[ -f $file_to_send ]]
321    then
322       true
323    else
324       printf "\n  +++ job-file: "
325       printf "\n           $file_to_send"
326       printf "\n      does not exist"
327       locat=parameter; exit
328    fi
329 fi
330
331
332
[1289]333    # IF NO JOBNAME HAS BEEN GIVEN, JOBNAME IS SET TO THE NAME OF THE JOB-FILE,
334    # PROVIDED THAT THE JOB-FILE NAME DOES NOT CONTAIN ANY PATH
[1]335 if [[ $job_name = none ]]
336 then
337    job_name=$file_to_send
338 fi
339 if [[ $(echo $job_name | grep -c "/") != 0 ]]
340 then
341    printf "\n  +++ job-file name: "
342    printf "\n           $job_name"
343    printf "\n      must not contain \"/\"-characters"
344    locat=parameter; exit
345 fi
346
347
348
349
[1289]350    # SET HOST-SPECIFIC QUANTITIES, OR TERMINATE IN CASE OF UNKNOWN HOST,
351    # OR IF NO HOST HAS BEEN GIVEN
[1]352 if [[ $remote_host = none ]]
353 then
354    printf "\n  +++ host missing"
355    locat=option; exit
356 else
357    case  $remote_host  in
[1468]358        (ibm)     queue=p690_standard; remote_address=134.76.99.81; submcom=/usr/lpp/LoadL/full/bin/llsubmit;;
359        (ibmh)    queue=cluster; remote_address=136.172.40.15; submcom=/usr/lpp/LoadL/full/bin/llsubmit;;
360        (ibmkisti) queue=class.32plus; remote_address=150.183.146.24; submcom=/usr/lpp/LoadL/full/bin/llsubmit;;
361        (ibmku)   queue=s4; remote_address=133.5.4.129; submcom=/usr/local/bin/llsubmit;;
362        (ibms)    queue=p_normal; remote_address=150.183.5.101; submcom=/usr/lpp/LoadL/full/bin/llsubmit;;
[1480]363        (lccrayb) queue=mpp1testq; remote_address=130.73.233.1; submcom=/opt/moab/default/bin/msub;;
364        (lccrayh) queue=mpp1testq; remote_address=130.75.4.1; submcom=/opt/moab/default/bin/msub;;
[1468]365        (lccrayf) queue=small; remote_address=86.50.166.21; submcom=/opt/slurm/default/bin/sbatch;;
[1547]366        (lcflow)  remote_address=10.140.1.72; submcom=qsub;;
[1468]367        (lckyoto) remote_address=133.3.51.11; submcom=/thin/local/bin/qsub;;
368        (lck)     remote_address=165.132.26.61; submcom=/usr/torque/bin/qsub;;
369        (lckiaps) remote_address=118.128.66.223; submcom=/cm/shared/apps/pbspro/11.0.2.110766/bin/qsub;;
370        (lckordi) remote_address=210.219.61.8; submcom=/usr/torque/bin/qsub;;
371        (lckyuh)  remote_address=133.5.4.33; submcom=/usr/bin/pjsub;;
372        (lckyut)  remote_address=133.5.4.37; submcom=/usr/bin/pjsub;;
373        (lcsb)    remote_address=147.46.30.151; submcom=/usr/torque/bin/qsub;;
374        (lctit)   queue=S; remote_address=10.1.6.165; submcom=/opt/pbs/tools/bin/t2sub;;
375        (lcxe6)   remote_address=129.177.20.113; submcom=/opt/torque/default/bin/qsub;;
376        (lcxt5m)  remote_address=193.166.211.144; submcom=/opt/pbs/10.1.0.91350/bin/qsub;;
377        (lcyon)   remote_address=165.132.26.68; submcom=/usr/torque/bin/qsub;;
378        (nech)    qsubmem=memsz_job; qsubtime=cputim_job; remote_address=136.172.44.147; submcom="/usr/local/bin/qsub";;
[251]379        (*)       printf "\n  +++ hostname \"$remote_host\" not allowed";
380                  locat=parameter; exit;;
[1]381    esac
382 fi
383
384
[1289]385    # CHECK, IF A VALID QUEUE HAS BEEN GIVEN
[1]386 if [[ $no_default_queue != none ]]
387 then
388    error=false
389    ndq=$no_default_queue
390    case  $remote_host  in
391        (ibm)    case  $ndq  in
392                     (p690_express|p690_standard|p690_long)  error=false;;
393                     (*)                                     error=true;;
394                 esac;;
395        (ibmh)   case  $ndq  in
[1184]396                     (cluster|express)  error=false;;
[1]397                     (*)                                     error=true;;
398                 esac;;
[693]399        (ibmkisti)   case  $ndq  in
400                     (class.32plus|class.1-2|class.2-32)  error=false;;
401                     (*)                                     error=true;;
402                 esac;;
[622]403        (ibmku)  case  $ndq  in
404                     (sdbg1|sdbg2|sdbg4|s4|s16|s32|s32-s)    error=false;;
405                     (*)                                     error=true;;
406                 esac;;
[1]407        (ibms)   case  $ndq  in
408                     (express|normal|p_express|p_normal|p_normal_1.3|p_normal_1.7|grand)     error=false;;
409                     (*)                                     error=true;;
410                 esac;;
[1224]411        (lccrayb) case  $ndq  in
[1480]412                     (dataq|mpp1q|mpp1testq|mpp2q|mpp2testq|smp1q|smp1testq|specialm1q)   error=false;;
[1224]413                     (*)                                     error=true;;
414                 esac;;
[1274]415        (lccrayh) case  $ndq  in
[1442]416                     (dataq|mpp1q|mpp1testq|smp1q|smp1testq|specialm1q)   error=false;;
[1274]417                     (*)                                     error=true;;
418                 esac;;
[1197]419        (lccrayf) case  $ndq  in
420                     (usup|test*|small|large)                error=false;;
421                     (*)                                     error=true;;
422                 esac;;
[1099]423        (lcflow) case  $ndq  in
[1547]424                     (cfd_lom_long.q|cfd_him_long.q|cfd_lom_serl.q|cfd_lom_shrt.q|cfd_him_shrt.q|cfd_ivy_shrt.q)  error=false;;
[1099]425                     (*)                                     error=true;;
426                 esac;;
[1040]427        (lckiaps) case  $ndq  in
428                     (express|normal)                        error=false;;
429                     (*)                                     error=true;;
430                 esac;;
[440]431        (lckyoto) case  $ndq  in
432                     (eh|ph)                                 error=false;;
433                     (*)                                     error=true;;
434                 esac;;
[1099]435        (lckyuh) case  $ndq  in
436                     (fx-dbg|fx-single|fx-small|fx-middle|fx-large)  error=false;;
437                     (*)                                     error=true;;
438                 esac;;
[1090]439        (lckyut) case  $ndq  in
440                     (cx-dbg|cx-single|cx-small|cx-middle|cx-large)  error=false;;
441                     (*)                                     error=true;;
442                 esac;;
[1]443        (lctit)  case  $ndq  in
[635]444                     (G|L128|L256|L512H|S|S96|V)             error=false;;
[1]445                     (*)                                     error=true;;
446                 esac;;
447        (t3eb)   case  $ndq  in
448                     (berte|p50|p100|p392|forfree|p25himem)  error=false;;
449                     (*)    error=true;;
450                 esac;;
451        (t3eh)   case  $ndq  in
452                     (para_t3e|em|k|l|lm|comp_t3e|c|p|ht)  error=false;;
453                     (*)    error=true;;
454                 esac;;
455        (t3ej2|t3ej5)  case  $ndq  in
456                     (low|normal|high)  error=false;;
457                     (*)    error=true;;
458                 esac;;
459        (t3es)  case  $ndq  in
460                     (batch|serial-4|pe4|p48|pe16|pe32|pe64|pe128)  error=false;;
461                     (*)    error=true;;
462                 esac;;
463    esac
464    if [[ $error = true ]]
465    then
466       printf "\n  +++ queue \"$no_default_queue\" on host \"$remote_host\" not allowed"
467       locat=parameter; exit
468    else
469       queue=$no_default_queue
470    fi
471 fi
472
473
474
[1289]475    # CHECK THE CPU-TIME
476    # SPLIT TIME INTO HOURS, MINUTES, AND SECONDS
[1]477 done=false
478 while [[ $done = false ]]
479 do
480    if (( $cputime <= 0 ))
481    then
482       printf "\n  +++ wrong cpu-time or cpu-time missing"
483       printf "\n  >>> Please type cpu-time in seconds as INTEGER:"
484       printf "\n  >>> "
485       read  cputime  1>/dev/null  2>&1
486    else
487       done=true
488    fi
489 done
490 if [[ $remote_host = nech ]]
491 then
492    if (( tasks_per_node != 0 ))
493    then
494       (( cputime = cputime * tasks_per_node ))
495    elif [[ $numprocs != 0 ]]
496    then
497       (( cputime = cputime * numprocs ))
498    fi
499 fi
500 (( stunden  = cputime / 3600 ))
501 (( resttime = cputime - stunden * 3600 ))
502 (( minuten  = resttime / 60 ))
503 (( sekunden = resttime - minuten * 60 ))
504 timestring=${stunden}:${minuten}:${sekunden}
505
506
507
[1289]508    # CHECK THE MEMORY DEMAND
[1]509 done=false
510 while [[ $done = false ]]
511 do
512    if (( memory <= 0 ))
513    then
514       printf "\n  +++ wrong memory demand or memory demand missing"
515       printf "\n  >>> Please type memory in  MByte per process  as INTEGER:"
516       printf "\n  >>> "
517       read  memory  1>/dev/null  2>&1
518    else
519       done=true
520    fi
521 done
522
[1289]523 if [[ $remote_host = nech ]]
[1]524 then
525    if (( tasks_per_node != 0 ))
526    then
527       (( Memory = memory * tasks_per_node / 1000 ))
528    elif [[ $numprocs != 0 ]]
529    then
530       (( Memory = memory * numprocs / 1000 ))
531    else
532       (( Memory = memory / 1000 ))
533    fi
[635]534 elif [[ $remote_host = lctit ]]
535 then
536    (( Memory = memory * tasks_per_node / 1000 ))
[1]537 fi
538
539
[1289]540    # MEMORY DEMAND IN CASE OF OPENMP-USAGE ON IBM-SYSTEMS
[1]541 if [[ $(echo $remote_host | cut -c1-3) = ibm ]]
542 then
543    (( memory = memory * threads_per_task ))
544 fi
545
546
[1289]547    # CALCULATE NUMBER OF REQUIRED NODES
[1]548 if (( tasks_per_node != 0 ))
549 then
[1279]550    (( nodes = ( numprocs - 1 ) / ( tasks_per_node * threads_per_task ) + 1 ))
[1]551 fi
552
[1094]553
[1289]554    # CALCULATE NUMBER OF PROCESSES PER NODE
[622]555 (( processes_per_node = tasks_per_node * threads_per_task ))
[1]556
[1094]557
[1289]558    # CALCULATE NUMBER OF MPI TASKS
[696]559 (( mpi_tasks = numprocs / threads_per_task ))
[1]560
[696]561
[1289]562    # SET PORT NUMBER OPTION FOR CALLS OF ssh/scp, subjob AND batch_scp SCRIPTS
[1094]563 if [[ "$scp_port" != "" ]]
564 then
565    PORTOPT="-P $scp_port"
566    SSH_PORTOPT="-p $scp_port"
567 fi
568
569
[1289]570    # HEADER-OUTPUT
[1]571 if [[ $verify = true ]]
572 then
573    printf "\n\n"
574    printf "#--------------------------------------------------------------# \n"
575    spalte1=SUBJOB;spalte2=$(date)
[1103]576    printf "| %-20s%40s | \n" "$spalte1" "$spalte2"
[1]577    printf "|                                                              | \n"
578    printf "| values of parameters/options:                                | \n"
[1103]579    spalte1=$(echo local_host$punkte | cut -c-20)
580    spalte2=$punkte$local_host
581    printf "| %-20s%40s | \n" "$spalte1" "${spalte2: -40}"
582    spalte1=$(echo remote_host$punkte | cut -c-20)
583    spalte2=$punkte$remote_host
584    printf "| %-20s%40s | \n" "$spalte1" "${spalte2: -40}"
585    spalte1=$(echo queue$punkte | cut -c-20)
586    spalte2=$punkte$queue
587    printf "| %-20s%40s | \n" "$spalte1" "${spalte2: -40}"
588    spalte1=$(echo memory$punkte | cut -c-20)
589    spalte2="$punkte$memory mb"
590    printf "| %-20s%40s | \n" "$spalte1" "${spalte2: -40}"
591    spalte1=$(echo cputime$punkte | cut -c-20)
592    spalte2="$punkte$cputime sec"
593    printf "| %-20s%40s | \n" "$spalte1" "${spalte2: -40}"
594    spalte1=$(echo job_name$punkte | cut -c-20)
595    spalte2="$punkte$job_name"
596    printf "| %-20s%40s | \n" "$spalte1" "${spalte2: -40}"
[1]597    printf "#--------------------------------------------------------------# \n\n"
598
599
[1289]600       # QUERY CHECK
[1]601    antwort="dummy"
602    while [[ $antwort != y  &&  $antwort != Y  &&  $antwort != n  &&  $antwort != N ]]
603    do
604       read antwort?" >>> continue (y/n) ? "
605    done
606    if [[ $antwort = n  ||  $antwort = N ]]
607    then
608       locat=verify; exit
609    fi
610    printf "\n"
611 fi
612
613
614
[1289]615    # GENERATE RANDOM IDENTIFIER, AND DETERMINE THE JOBNAME ON THE TARGET HOST
616 identifier=$RANDOM
617 job_on_remhost=${job_name}_${identifier}_$local_host
618 job_to_send=job_to_send_$identifier
[1]619 if [[ $delete_dayfile = false ]]
620 then
[1289]621    remote_dayfile=${local_host}_${job_name}_result_$identifier
[1]622    local_dayfile=${remote_host}_${job_name}
623 else
624    remote_dayfile=/dev/null
625 fi
626
627
628
[1289]629    # GENERATE THE BATCH-JOB SCRIPTS (FOR QUEUEING-SYSTEMS qsub/msub/LoadLeveler)
[1]630 if [[ $(echo $remote_host | cut -c1-3) = ibm  &&  $numprocs != 0 ]]
631 then
632
[1289]633       # GENERAL LOADLEVELER SETTINGS
[622]634    execute_in_shell="#!/bin/ksh"
635    use_shell="# @ shell = /bin/ksh"
636    consumable_memory="ConsumableMemory($memory mb)"
637    class="# @ class = $queue"
638    environment="# @ environment = OMP_NUM_THREADS=$threads_per_task; MP_SHARED_MEMORY=yes"
639    network_to_use="# @ network.mpi = sn_all,shared,us"
640    data_limit="# @ data_limit = 1.76gb"
641    image_size="# @ image_size = 50"
[693]642    wall_clock_limit="# @ wall_clock_limit = ${timestring},$timestring"
[312]643
[693]644    if [[ $email_notification = none ]]
645    then
646       notify_user=""
647    else
648       notify_user="# @ notify_user = $email_notification"
649       if [[ $delete_dayfile = true ]]
650       then
651          notification='# @ notification = never'
652       fi
653    fi
[622]654
[312]655    if [[ $remote_host = ibmh ]]
[1]656    then
[312]657       data_limit=""
658       network_to_use=""
[1184]659       class="# @ class = $queue"
[312]660       environment=""
[814]661       rset="# @ rset = RSET_MCM_AFFINITY"
662       task_affinity="# @ task_affinity = core(1)"
[693]663    elif [[ $remote_host = ibmkisti ]]
664    then
665       network_to_use="# @ network.MPI = sn_all,shared,US"
666       wall_clock_limit="# @ wall_clock_limit = $timestring"
[696]667       if [[ $threads_per_task = 1 ]]
668       then
669          rset="# @ rset = RSET_MCM_AFFINITY"
670          mcm_affinity_options="# @ mcm_affinity_options = mcm_mem_pref mcm_sni_none mcm_distribute"
671       fi
[693]672       environment=""
673       use_shell=""
674       data_limit=""
675       image_size=""
[622]676    elif [[ $remote_host = ibmku ]]
677    then
678       execute_in_shell="#!/usr/bin/ksh"
679       use_shell="# @ shell = /usr/bin/ksh"
680       consumable_memory=""
681       environment=""
682       network_to_use="# @ network.mpi = sn_all,shared,us"
683       data_limit=""
684       image_size=""
685    elif [[ $remote_host = ibms ]]
686    then
687       network_to_use="# @ network.mpi = csss,shared,us"
[1]688    fi
689
690    cat > $job_to_send << %%END%%
[622]691$execute_in_shell
692$use_shell
[1]693
694# @ job_type = parallel
[693]695# @ job_name = $job_name
[1]696# @ resources = ConsumableCpus($threads_per_task) $consumable_memory
697# @ output = $remote_dayfile
698# @ error = $remote_dayfile
[693]699$wall_clock_limit
[622]700$image_size
[312]701$class
702$environment
[1]703$network_to_use
704$data_limit
[693]705$rset
706$mcm_affinity_options
[814]707$task_affinity
[1]708$notification
[693]709$notify_user
[1]710
711%%END%%
712
713    if (( nodes > 0 ))
714    then
715
[693]716       if [[ $remote_host != ibmkisti ]]
717       then
718
719          cat >> $job_to_send << %%END%%
[1]720# @ node = $nodes
[622]721# @ tasks_per_node = $processes_per_node
[1]722# @ node_usage = $node_usage
723# @ queue
724
725%%END%%
726
[693]727       else
728
729          cat >> $job_to_send << %%END%%
[696]730# @ total_tasks = $mpi_tasks
[693]731# @ blocking = unlimited
732# @ queue
733
734%%END%%
735
736       fi
737
[1]738    else
739
[1289]740       cat >> $job_to_send << %%END%%
[1]741# @ blocking = unlimited
742# @ total_tasks = $numprocs
743# @ node_usage = $node_usage
744# @ queue
745
746%%END%%
747
748    fi
749
[1289]750       # WORKAROUND BECAUSE OF SILLY JOB FILTER ON ibmkisti
[696]751    if [[ $remote_host = ibmkisti  &&  $threads_per_task != 1 ]]
752    then
753       echo  "export OMP_NUM_THREADS=$threads_per_task"  >>  $job_to_send
754    fi
755
[1]756 elif [[ $(echo $remote_host | cut -c1-3) = ibm  &&  $numprocs = 0 ]]
757 then
758
759    cat > $job_to_send << %%END%%
760#!/bin/ksh
761
762# @ job_type = serial
763# @ node_usage = $node_usage
764# @ job_name = palm
765# @ wall_clock_limit = ${timestring},$timestring
766# @ resources = ConsumableCpus(1) ConsumableMemory(1 gb)
767# @ output = $remote_dayfile
768# @ error = $remote_dayfile
[312]769$class
[1]770$notification
771
772# @ queue
773
774%%END%%
775
[1274]776 elif [[ $remote_host = lccrayb || $remote_host = lccrayh ]]
[1224]777 then
778
[1480]779    if [[ "$feature" != "" ]]
780    then
781       featuredir="#PBS -l feature=$feature"
782    fi
783
[1224]784    if [[ $numprocs != 0 ]]
785    then
786       cat > $job_to_send << %%END%%
[1255]787#!/bin/bash
[1224]788#PBS -N $job_name
789#PBS -l walltime=$timestring
[1264]790#PBS -l nodes=$nodes:ppn=$processes_per_node
[1224]791#PBS -o $remote_dayfile
792#PBS -j oe
793#PBS -q $queue
[1480]794$featuredir
[1224]795
796$module_calls
797
798%%END%%
799
800    else
801
802       continue
803
804    fi
805
[1197]806 elif [[ $remote_host = lccrayf ]]
807 then
808
809    if [[ $numprocs != 0 ]]
810    then
811       cat > $job_to_send << %%END%%
812#!/bin/bash -l
813#SBATCH -J $job_name
814#SBATCH -t $timestring
815#SBATCH -N $nodes
816#SBATCH --ntasks-per-node=$processes_per_node
817#SBATCH -p $queue
818#SBATCH -o $remote_dayfile
819#SBATCH -e $remote_dayfile
820
821$init_cmds
822$module_calls
823
824%%END%%
825
826    else
827       cat > $job_to_send << %%END%%
828#!/bin/bash -l
829#SBATCH -J $job_name
830#SBATCH -t $timestring
831#SBATCH -l ncpus=1
832#SBATCH -l pmem=${memory}mb
833#SBATCH -m abe
834#SBATCH -o $remote_dayfile
835#SBATCH -e $remote_dayfile
836
837$init_cmds
838$module_calls
839
840%%END%%
841
842    fi
843
[1099]844 elif [[ $remote_host = lcflow ]]
845 then
846    if [[ $numprocs != 0 ]]
847    then
[1547]848      pe_set="#$ -pe impi $numprocs"
[1099]849    else
[1547]850      pe_set="#$ -pe impi 1"
[1099]851    fi
852    if [[ $queue = default ]]
853    then
854      queue_set=""
855    else
856      queue_set="#$ -q $queue"
857    fi
858    [[ "$disc_space" = "" ]]  &&  disc_space=50
859
860       cat > $job_to_send << %%END%%
861#!/bin/bash
862#$ -S /bin/bash
863#$ -N $job_name
864#$ -cwd
865#$ -l h_rt=$timestring
866#$ -l h_vmem=${memory}M
867#$ -o $remote_dayfile
868#$ -j y
869$pe_set
870#$ -R y
871#$ -l h_fsize=${disc_space}G
872$queue_set
873
874%%END%%
875
[1021]876 elif [[ $remote_host = lck || $remote_host = lckordi || $remote_host = lcsb ]]
[368]877 then
878
879    if [[ $numprocs != 0 ]]
880    then
881       cat > $job_to_send << %%END%%
882#!/bin/ksh
883#PBS -N $job_name
884#PBS -l walltime=$timestring
885#PBS -l ncpus=$numprocs
886#PBS -l pmem=${memory}mb
887#PBS -o $remote_dayfile
[1021]888#PBS -l nodes=$nodes:ppn=${processes_per_node}
[368]889#PBS -j oe
890
891mpd &
892
893%%END%%
894
895    else
896       cat > $job_to_send << %%END%%
897#!/bin/ksh
898#PBS -N $job_name
899#PBS -l walltime=$timestring
900#PBS -l ncpus=1
901#PBS -l pmem=${memory}mb
902#PBS -o $remote_dayfile
903#PBS -j oe
904
905%%END%%
906
907    fi
908
[1040]909 elif [[ $remote_host = lckiaps ]]
910 then
911
912    if [[ $numprocs != 0 ]]
913    then
914       cat > $job_to_send << %%END%%
915#!/bin/ksh
916#PBS -N $job_name
917#PBS -l walltime=$timestring
918#PBS -l select=1:ncpus=$numprocs
919#PBS -l pmem=${memory}mb
920#PBS -q $queue
921#PBS -o $remote_dayfile
922#PBS -j oe
923#PBS -V
924
925%%END%%
926
927    else
928       cat > $job_to_send << %%END%%
929#!/bin/ksh
930#PBS -N $job_name
931#PBS -l walltime=$timestring
932#PBS -l ncpus=1
933#PBS -l pmem=${memory}mb
934#PBS -o $remote_dayfile
935#PBS -j oe
936
937%%END%%
938
939    fi
940
[693]941 elif [[ $remote_host = lcyon ]]
942 then
943
944    if [[ $numprocs != 0 ]]
945    then
946       cat > $job_to_send << %%END%%
947#!/bin/ksh
948#PBS -N $job_name
949#PBS -l walltime=$timestring
950#PBS -l ncpus=$numprocs
951#PBS -l pmem=${memory}mb
952#PBS -o $remote_dayfile
953#PBS -j oe
954
955%%END%%
956
957    else
958       cat > $job_to_send << %%END%%
959#!/bin/ksh
960#PBS -N $job_name
961#PBS -l walltime=$timestring
962#PBS -l ncpus=1
963#PBS -l pmem=${memory}mb
964#PBS -o $remote_dayfile
965#PBS -j oe
966
967%%END%%
968
969    fi
970
[892]971 elif [[ $remote_host = lcxe6 ]]
[164]972 then
973
974    if [[ $numprocs != 0 ]]
975    then
976       cat > $job_to_send << %%END%%
[799]977#!/bin/ksh
[164]978#PBS -S /bin/ksh
979#PBS -N $job_name
[552]980#PBS -A $project_account
[206]981#PBS -j oe
[164]982#PBS -l walltime=$timestring
983#PBS -l mppwidth=${numprocs}
[622]984#PBS -l mppnppn=${processes_per_node}
[164]985#PBS -m abe
986#PBS -o $remote_dayfile
[492]987$email_directive
[164]988
[892]989$init_cmds
[493]990$module_calls
[343]991
[164]992%%END%%
993
[108]994    else
995       cat > $job_to_send << %%END%%
996#!/bin/ksh
[168]997#PBS -S /bin/ksh
[108]998#PBS -N $job_name
[552]999#PBS -A $project_account
[206]1000#PBS -j oe
[108]1001#PBS -l walltime=$timestring
1002#PBS -l ncpus=1
1003#PBS -l pmem=${memory}mb
1004#PBS -m abe
[492]1005$email_directive
[108]1006#PBS -o $remote_dayfile
1007
[892]1008$init_cmds
[493]1009$module_calls
[343]1010
[108]1011%%END%%
1012
1013    fi
1014
[440]1015 elif [[ $remote_host = lckyoto ]]
[437]1016 then
1017
[440]1018       cat > $job_to_send << %%END%%
[799]1019#!/bin/ksh
[440]1020# @\$-o $remote_dayfile
1021# @\$-eo -oi
1022# @\$-lP 16
[799]1023# @\$-lp 1
[440]1024# @\$-lm 28gb  -llm unlimited -ls unlimited
1025# @\$-q $queue
1026# @\$-Pvn abs_pack
1027##for intel? @\$-Pvn abs_unpack -Pvs unpack -Pvc unpack
1028#. /thin/local/etc/setprofile/intel-11.0.sh
1029#. /thin/local/etc/setprofile/mvapich2-1.4+intel-11.0.sh
1030. ~/.myprofile
1031#. /home2/t/t51254/palm/current_version/myprofile
1032#. /thin/apps/pgi/mpi.sh
1033#
1034env
1035#
1036set -x
1037
1038%%END%%
1039
1040 elif [[ $remote_host = lcxt5m ]]
1041 then
1042
[437]1043    if [[ $numprocs != 0 ]]
1044    then
1045       cat > $job_to_send << %%END%%
[799]1046#!/bin/ksh
[437]1047#PBS -S /bin/ksh
1048#PBS -N $job_name
1049#PBS -j oe
1050#PBS -l walltime=$timestring
1051#PBS -l mppwidth=${numprocs}
[622]1052#PBS -l mppnppn=${processes_per_node}
[437]1053#PBS -m abe
1054#PBS -o $remote_dayfile
1055
[892]1056$init_cmds
[493]1057$module_calls
[437]1058
1059%%END%%
1060
1061    else
1062       cat > $job_to_send << %%END%%
1063#!/bin/ksh
1064#PBS -S /bin/ksh
1065#PBS -N $job_name
1066#PBS -j oe
1067#PBS -l walltime=$timestring
1068#PBS -l ncpus=1
1069#PBS -l pmem=${memory}mb
1070#PBS -m abe
1071#PBS -o $remote_dayfile
1072
[892]1073$init_cmds
[493]1074$module_calls
[437]1075
1076%%END%%
1077
1078    fi
1079
[1099]1080 elif [[ $remote_host = lckyuh ]]
1081 then
1082    cat > $job_to_send << %%END%%
1083#!/bin/bash
1084#PJM -L "rscgrp=$queue"
1085#PJM -L "node=$nodes"
1086#PJM --mpi "proc=$numprocs"
1087#PJM -L "elapse=$timestring"
1088#PJM -o $remote_dayfile
1089#PJM -j
1090#PJM -X
1091#PJM --no-stging
1092
1093export LANG=en_US.UTF-8
1094%%END%%
1095
[1090]1096 elif [[ $remote_host = lckyut ]]
1097 then
1098    cat > $job_to_send << %%END%%
1099#!/bin/bash
[1099]1100#PJM -L "rscgrp=$queue"
[1090]1101#PJM -L "vnode=$numprocs"
1102#PJM -L "vnode-core=1"
1103#PJM -L "elapse=$timestring"
1104#PJM --mpi proc=$numprocs
1105#PJM -o $remote_dayfile
1106#PJM -j
[1099]1107#PJM -X
1108#PJM --no-stging
[1090]1109
1110export LANG=en_US.UTF-8
1111%%END%%
1112
[1]1113 elif [[ $remote_host = nech ]]
1114 then
1115
1116    if (( nodes > 1 ))
1117    then
1118       cat > $job_to_send << %%END%%
1119#!/bin/ksh
[622]1120#PBS -l cpunum_prc=$processes_per_node,cputim_job=$cputime
[1]1121#PBS -l ${qsubmem}=${Memory}gb
1122#PBS -b $nodes
1123#PBS -o $remote_dayfile
1124#PBS -N palm
1125#PBS -j o
1126#PBS -T mpisx
1127
1128%%END%%
1129
1130    elif [[ $numprocs != 0 ]]
1131    then
1132       cat > $job_to_send << %%END%%
1133#!/bin/ksh
[622]1134#PBS -l cpunum_prc=$processes_per_node,cputim_job=$cputime
[1]1135#PBS -l ${qsubmem}=${Memory}gb
1136#PBS -o $remote_dayfile
1137#PBS -N palm
1138#PBS -j o
1139
1140%%END%%
1141
1142    else
1143       cat > $job_to_send << %%END%%
1144#!/bin/ksh
1145#PBS -l ${qsubmem}=${Memory}gb,${qsubtime}=$cputime
1146#PBS -o $remote_dayfile
1147#PBS -j o
1148
1149%%END%%
1150
1151    fi
1152
1153 elif [[ $remote_host = lctit ]]
1154 then
1155    cat > $job_to_send << %%END%%
[635]1156#!/bin/ksh
[892]1157$init_cmds
[678]1158$module_calls
1159
[1]1160%%END%%
1161
[1289]1162       # SET OPTIONS FOR SUBMIT-COMMAND
[678]1163    if [[ $tasks_per_node != $processes_per_node ]]
1164    then
[1378]1165       submcom="$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 $queue "
[678]1166    else
[1378]1167       submcom="$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]1168    fi
[1]1169
1170 else
1171
1172    cat > $job_to_send << %%END%%
1173# @\$-q ${queue}
1174# @\$-l${qsubtime} $timestring
1175# @\$-l${qsubmem} ${memory}mb
1176# @\$-o $remote_dayfile
1177# @\$-eo
1178
1179%%END%%
1180
1181 fi
1182
1183
[1289]1184    # IN CASE OF JOBS EXECUTING ON REMOTE-HOSTS, THE TRANSFER OF THE DAYFILES
1185    # TO THE LOCAL HOSTS WILL BE INITIATED BY TRAP ON EXIT
1186    # NO TRANSFER POSSIBLE ON IBM IN SEOUL
[1]1187 if [[ $delete_dayfile = false  &&  $remote_host != $local_host ]]
1188 then
1189    echo "set +vx"                              >>  $job_to_send
1190    echo "trap '"                               >>  $job_to_send
1191    echo "set +vx"                              >>  $job_to_send
[1468]1192    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]1193    then
[622]1194       if [[ $remote_host = ibmh ]]
[1]1195       then
1196          return_queue=c1
[693]1197       elif [[ $remote_host = ibmkisti ]]
1198       then
1199          return_queue=class.1-2
[622]1200       elif [[ $remote_host = ibmku ]]
1201       then
1202          return_queue=sdbg2
[1]1203       elif [[ $remote_host = ibms ]]
1204       then
1205          return_queue=p_normal
[1274]1206       elif [[ $remote_host = lccrayb || $remote_host = lccrayh ]]
[1255]1207       then
1208          return_queue=dataq
[1468]1209       elif [[ $remote_host = lcxe6 ]]
1210       then
1211          return_queue=debug
[1040]1212       elif [[ $remote_host = lckiaps ]]
1213       then
1214          return_queue=express
[1099]1215       elif [[ $remote_host = lckyuh ]]
1216       then
1217          return_queue=cx-single
[1090]1218       elif [[ $remote_host = lckyut ]]
1219       then
1220          return_queue=cx-single
[1]1221       else
1222          return_queue=unknown
1223       fi
1224
1225       if [[ $(echo $remote_host | cut -c1-3) = ibm ]]
1226       then
1227
[622]1228          if [[ $remote_host = ibmku ]]
1229          then
[1289]1230             echo "echo \"#!/usr/bin/ksh\" >> scpjob.$identifier"            >>  $job_to_send
1231             echo "echo \"# @ shell = /usr/bin/ksh\" >> scpjob.$identifier"  >>  $job_to_send
[622]1232          else
[1289]1233             echo "echo \"#!/bin/ksh\" >> scpjob.$identifier"                >>  $job_to_send
[622]1234          fi
[1289]1235          echo "echo \"# @ job_type = serial\" >> scpjob.$identifier"    >>  $job_to_send
1236          echo "echo \"# @ job_name = transfer\" >> scpjob.$identifier"  >>  $job_to_send
1237          echo "echo \"# @ resources = ConsumableCpus(1) ConsumableMemory(1 gb)\" >> scpjob.$identifier"  >>  $job_to_send
1238          echo "echo \"# @ wall_clock_limit = 00:10:00,00:10:00\" >> scpjob.$identifier "  >>  $job_to_send
1239          echo "echo \"# @ output = job_queue/last_job_transfer_protocol\" >> scpjob.$identifier"  >>  $job_to_send
1240          echo "echo \"# @ error = job_queue/last_job_transfer_protocol\" >> scpjob.$identifier"  >>  $job_to_send
[312]1241          if [[ $host != "ibmh" ]]
1242          then
[1289]1243             echo "echo \"# @ class = $return_queue\" >> scpjob.$identifier"  >>  $job_to_send
[312]1244          fi
[1289]1245          echo "echo \"# @ image_size = 10\" >> scpjob.$identifier"      >>  $job_to_send
1246          echo "echo \"# @ notification = never\" >> scpjob.$identifier" >>  $job_to_send
[1]1247
[1289]1248          echo "echo \"# @ queue\" >> scpjob.$identifier"                >>  $job_to_send
1249          echo "echo \" \" >> scpjob.$identifier"                        >>  $job_to_send
[1]1250
[1289]1251          echo "echo \"set -x\" >> scpjob.$identifier"                   >>  $job_to_send
[1468]1252          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]1253          if [[ $remote_host = ibmku ]]
1254          then
[1289]1255             echo "echo \"rm  scpjob.$identifier\" >> scpjob.$identifier"   >>  $job_to_send
[622]1256          fi
[1289]1257          echo "echo \"exit\" >> scpjob.$identifier"                     >>  $job_to_send
[1]1258
1259       elif [[ $remote_host = nech ]]
1260       then
1261          echo "cd /pf/b/${remote_user}/job_queue" >>  $job_to_send
[1289]1262          echo "cat > scpjob.$identifier << %%END%%"  >>  $job_to_send
[1]1263          echo "#PBS -l ${qsubmem}=1GB,${qsubtime}=100"  >>  $job_to_send
1264          echo "#PBS -o last_job_transfer_protocol"      >>  $job_to_send
1265          echo "#PBS -j o"                         >>  $job_to_send
1266          echo " "                                 >>  $job_to_send
1267          echo "set -x"                            >>  $job_to_send
1268          echo "cd /pf/b/${remote_user}/job_queue" >>  $job_to_send
[1468]1269          echo "batch_scp  $PORTOPT  -d  -w 10  -u $local_user $local_address  $remote_dayfile  \"$job_catalog\"  $local_dayfile"  >>  $job_to_send
[1]1270          echo "[[ \"\$for_subjob_to_do\" != \"\" ]]  &&  eval \$for_subjob_to_do"  >>  $job_to_send
1271          echo "%%END%%"                           >>  $job_to_send
1272
[1099]1273       elif [[ $remote_host = lckyuh ]]
1274       then
[1289]1275          echo "cat > scpjob.$identifier << %%END%%"  >>  $job_to_send
[1099]1276          echo "#!/bin/bash"                       >>  $job_to_send
1277          echo "#PJM -L \"node=1\""                >>  $job_to_send
1278          echo "#PJM -L \"rscgrp=$return_queue\""  >>  $job_to_send
1279          echo "#PJM --no-stging"                  >>  $job_to_send
1280          echo "#PJM -L \"elapse=30:00\""          >>  $job_to_send
1281          echo "#PJM -o \$HOME/job_queue/last_job_transfer_protocol"  >>  $job_to_send
1282          echo "#PJM -j"                           >>  $job_to_send
1283          echo " "                                 >>  $job_to_send
1284          echo "export LANG=en_US.UTF-8"           >>  $job_to_send
1285          echo "set -x"                            >>  $job_to_send
[1468]1286          echo "batch_scp  $PORTOPT  -d  -w 10  -u $local_user $local_address  $remote_dayfile  \"$job_catalog\"  $local_dayfile"  >>  $job_to_send
[1099]1287          echo "[[ \"\$for_subjob_to_do\" != \"\" ]]  &&  eval \$for_subjob_to_do"  >>  $job_to_send
1288          echo "%%END%%"                           >>  $job_to_send
1289
[1090]1290       elif [[ $remote_host = lckyut ]]
1291       then
[1289]1292          echo "cat > scpjob.$identifier << %%END%%"  >>  $job_to_send
[1090]1293          echo "#!/bin/bash"                       >>  $job_to_send
1294          echo "#PJM -L \"vnode=1\""               >>  $job_to_send
1295          echo "#PJM -L \"rscgrp=$return_queue\""  >>  $job_to_send
1296          echo "#PJM --no-stging"                  >>  $job_to_send
1297          echo "#PJM -L \"elapse=30:00\""          >>  $job_to_send
1298          echo "#PJM -o \$HOME/job_queue/last_job_transfer_protocol"  >>  $job_to_send
1299          echo "#PJM -j"                           >>  $job_to_send
1300          echo " "                                 >>  $job_to_send
1301          echo "export LANG=en_US.UTF-8"           >>  $job_to_send
1302          echo "set -x"                            >>  $job_to_send
[1468]1303          echo "batch_scp  $PORTOPT  -d  -w 10  -u $local_user $local_address  $remote_dayfile  \"$job_catalog\"  $local_dayfile"  >>  $job_to_send
[1090]1304          echo "[[ \"\$for_subjob_to_do\" != \"\" ]]  &&  eval \$for_subjob_to_do"  >>  $job_to_send
1305          echo "%%END%%"                           >>  $job_to_send
1306
[1274]1307       elif [[ $remote_host = lccrayb || $remote_host = lccrayh ]]
[1255]1308       then
[1289]1309          echo "cat > scpjob.$identifier << %%END%%"        >>  $job_to_send
[1255]1310          echo "#!/bin/bash"                             >>  $job_to_send
1311          echo "#PBS -N job_protocol_transfer"           >>  $job_to_send
1312          echo "#PBS -l walltime=00:30:00"               >>  $job_to_send
[1262]1313          echo "#PBS -l nodes=1:ppn=1"                   >>  $job_to_send
[1255]1314          echo "#PBS -o \$HOME/job_queue/last_job_transfer_protocol"      >>  $job_to_send
1315          echo "#PBS -j oe"                              >>  $job_to_send
1316          echo " "                                       >>  $job_to_send
1317          echo "set -x"                                  >>  $job_to_send
[1468]1318          echo "batch_scp  $PORTOPT  -d  -w 10  -u $local_user $local_address  ${job_catalog}/$remote_dayfile  \"$job_catalog\"  $local_dayfile"  >>  $job_to_send
[1255]1319          echo "[[ \"\$for_subjob_to_do\" != \"\" ]]  &&  eval \$for_subjob_to_do"  >>  $job_to_send
1320          echo "%%END%%"                                 >>  $job_to_send
1321
[1099]1322       elif [[ $remote_host = lcflow ]]
1323       then
[1289]1324          echo "cat > scpjob.${identifier}.tmp << %%END%%"                  >>  $job_to_send
[1099]1325          echo "#!/bin/bash"                                             >>  $job_to_send
1326          echo "SGEPREFIX -S /bin/bash"                                  >>  $job_to_send
1327          echo "SGEPREFIX -N transfer_$job_name"                         >>  $job_to_send
1328          echo "SGEPREFIX -cwd"                                          >>  $job_to_send
1329          echo "SGEPREFIX -l h_rt=01:00:00"                              >>  $job_to_send
[1202]1330          echo "SGEPREFIX -l h_vmem=500M"                                >>  $job_to_send
[1099]1331          echo "SGEPREFIX -j y"                                          >>  $job_to_send
[1289]1332          echo "SGEPREFIX -o ${local_host}_${job_name}_scpjob_$identifier"  >>  $job_to_send 
[1099]1333          echo " "                                                       >>  $job_to_send 
1334          echo "set -x"                                                  >>  $job_to_send 
1335          echo "export PALM_BIN=$PALM_BIN" | sed -e 's:'$HOME':$HOME:'   >>  $job_to_send
1336          echo "export PATH=\$PATH:\$PALM_BIN"                           >>  $job_to_send
1337          echo ""                                 >>  $job_to_send         
[1468]1338          echo "batch_scp  $PORTOPT  -d  -w 10  -u $local_user $local_address  ${job_catalog}/$remote_dayfile  \"$job_catalog\"  $local_dayfile"  >>  $job_to_send
[1099]1339          echo "[[ \"\$for_subjob_to_do\" != \"\" ]]  &&  eval \$for_subjob_to_do"  >>  $job_to_send
[1289]1340          echo "rm -f scpjob.${identifier}"                                 >>  $job_to_send         
[1099]1341          echo "%%END%%"                                                 >>  $job_to_send
[1289]1342          echo "sed -e 's/SGEPREFIX/#$/g' scpjob.${identifier}.tmp > scpjob.${identifier}" >>  $job_to_send         
1343          echo "rm -f scpjob.${identifier}.tmp"                             >>  $job_to_send         
[1468]1344       elif [[ $remote_host = lcxe6 ]]
1345       then
1346          echo "cat > scpjob.${identifier}  << %%END%%"  >>  $job_to_send
1347          echo "#!/bin/ksh"                              >>  $job_to_send
1348          echo "#PBS -N job_protocol_transfer"           >>  $job_to_send
1349          echo "#PBS -l walltime=00:30:00"               >>  $job_to_send
1350          echo "#PBS -A $project_account"                >>  $job_to_send
1351          echo "#PBS -l mppwidth=1"                      >>  $job_to_send
1352          echo "#PBS -l mppnppn=1"                       >>  $job_to_send
1353          echo "#PBS -o \$HOME/job_queue/last_job_transfer_protocol"  >>  $job_to_send
1354          echo "#PBS -j oe"                              >>  $job_to_send
1355          echo " "                                       >>  $job_to_send
1356          echo "set -x"                                  >>  $job_to_send
1357          echo "batch_scp  $PORTOPT  -d  -w 10  -u $local_user $local_address  ${job_catalog}/$remote_dayfile  \"$job_catalog\"  $local_dayfile"  >>  $job_to_send
1358          echo "[[ \"\$for_subjob_to_do\" != \"\" ]]  &&  eval \$for_subjob_to_do"  >>  $job_to_send
1359          echo "%%END%%"                                 >>  $job_to_send
[1]1360       else
1361
[1289]1362          echo "cat > scpjob.$identifier << %%END%%"  >>  $job_to_send
[1]1363          echo "# @\\\$-q $return_queue"           >>  $job_to_send
1364          echo "# @\\\$-l${qsubtime} 10"           >>  $job_to_send
1365          echo "# @\\\$-l${qsubmem} 10mb"          >>  $job_to_send
1366          if [[ $remote_host = t3ej2  ||  $remote_host = t3ej5  ||  $remote_host = t3es ]]
1367          then
1368             echo "# @\$-l mpp_p=0"                >>  $job_to_send
1369          fi
1370          echo '# @\$-lF 10mb'                     >>  $job_to_send
1371          echo '# @\$-o job_queue/last_job_transfer_protocol'    >>  $job_to_send
1372          echo '# @\\\$-eo'                          >>  $job_to_send
1373          echo " "                                 >>  $job_to_send
1374          if [[ $remote_host = t3ej2  ||  $remote_host = t3ej5 ]]
1375          then
1376             echo "set +vx"                        >>  $job_to_send
1377             echo ". .profile"                     >>  $job_to_send
1378          fi
1379          echo "set -x"                            >>  $job_to_send
[1468]1380          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]1381          echo "[[ \"\$for_subjob_to_do\" != \"\" ]]  &&  eval \$for_subjob_to_do"  >>  $job_to_send
1382          echo "%%END%%"                           >>  $job_to_send
1383       fi
1384
1385       if [[ $(echo $remote_host | cut -c1-3) = ibm ]]
1386       then
[1289]1387          echo "llsubmit  scpjob.$identifier"      >>  $job_to_send
[1274]1388       elif [[ $remote_host = lccrayb || $remote_host = lccrayh ]]
[1255]1389       then
[1289]1390          echo "msub -q $return_queue  scpjob.$identifier"               >>  $job_to_send
[1]1391       elif [[ $remote_host = t3eb  ||  $remote_host = t3eh  ||  $remote_host = t3ej2  ||  $remote_host = t3ej5 ]]
1392       then
[1289]1393          echo "qsub -J n  scpjob.$identifier"     >>  $job_to_send
[1]1394       elif [[ $remote_host = t3es ]]
1395       then
[1289]1396          echo "qsub -J n  -s /bin/ksh  scpjob.$identifier"     >>  $job_to_send
[1043]1397       elif [[ $remote_host = lckiaps ]]
1398       then
[1289]1399          echo "mv  scpjob.$identifier  $job_catalog"           >>  $job_to_send
[1468]1400          echo "ssh $SSH_PORTOPT ${remote_username}@${remote_address}  \"$submcom ${job_catalog}/scpjob.$identifier\" "  >>  $job_to_send
[1289]1401          echo "rm  ${job_catalog}/scpjob.$identifier"          >>  $job_to_send
[1099]1402       elif [[ $remote_host = lckyu* ]]
[1090]1403       then
[1468]1404          echo "scp $PORTOPT scpjob.$identifier  ${remote_username}@${remote_address}:job_queue"           >>  $job_to_send
1405          echo "ssh $SSH_PORTOPT ${remote_username}@${remote_address}  \"cd job_queue; $submcom scpjob.$identifier; rm scpjob.$identifier\" "  >>  $job_to_send
[1099]1406       elif [[ $remote_host = lcflow ]]
1407       then
[1289]1408          echo "mv  scpjob.$identifier  $job_catalog"           >>  $job_to_send
[1468]1409          echo "/usr/bin/ssh ${remote_username}@${remote_address}  \"$init_cmds $module_calls cd $job_catalog; $submcom scpjob.$identifier\" "  >>  $job_to_send
[1]1410       else
[1289]1411          echo "$submcom  scpjob.$identifier"      >>  $job_to_send
[1]1412       fi
[1043]1413       if [[ $remote_host != ibmku  &&  $remote_host != lckiaps ]]
[622]1414       then
[1289]1415          echo "rm  scpjob.$identifier"            >>  $job_to_send
[622]1416       fi
[1]1417       if [[ $remote_host = nech ]]
1418       then
1419          echo "cd -"                           >>  $job_to_send
1420       fi
1421    else
[1468]1422#       echo "ftpcopy  -d  $local_address  ${job_catalog}/$remote_dayfile  \"$job_catalog\"  $local_dayfile"  >>  $job_to_send
[1289]1423       # ??? funktioniert das ÃŒberhaupt noch ???
[1468]1424       echo "nohup  ftpcopy  -d  -w 15  $local_address  ${job_catalog}/$remote_dayfile  \"$job_catalog\"  $local_dayfile  >  /dev/null  &"  >>  $job_to_send
[1]1425    fi
1426    echo "set -x"                               >>  $job_to_send
1427    echo "     ' exit"                          >>  $job_to_send
1428    echo "set -x"                               >>  $job_to_send
1429 fi
1430
1431
1432
[1289]1433    # APPEND THE JOB-FILE (CREATE BY mrun) TO THE JOB-DIRECTIVES GENERATED ABOVE
[1]1434 cat  $file_to_send  >>  $job_to_send
[69]1435
[1]1436 if [[ $remote_host = ibm ]]
1437 then
1438    echo " "         >>  $job_to_send
1439    echo "exit"      >>  $job_to_send
1440 fi
[635]1441
[1289]1442    # REMOVE JOB-FILE
[1099]1443 if [[ $remote_host = lctit  ||  $remote_host = ibmku  ||  $remote_host = lcflow ]]
[69]1444 then
1445    echo " "                               >>  $job_to_send
1446    echo "rm ~/job_queue/$job_on_remhost"  >>  $job_to_send
1447 fi
[1]1448
1449
[1289]1450    # TRANSFER JOB TO THE TARGET HOST (JOB-DIRECTORY)
[1]1451 if [[ $no_submit = false ]]
1452 then
1453    if [[ $remote_host != $local_host ]]
1454    then
1455       [[ $verify = true ]]  &&  printf "\n >>> transfering job to \"$remote_host\"..."
[1289]1456       if [[ $remote_host = ibms ]]    # ssh on ibms cannot handle "~/"
[1]1457       then
[82]1458          job_catalog_save=$job_catalog
1459          job_catalog=job_queue
1460       elif [[ $remote_host = nech ]]
1461       then
1462          job_catalog_save=$job_catalog
1463          job_catalog=/hpf/b/${remote_user}/job_queue
1464       fi
[1096]1465       if [[ $remote_host = nech ]]
[82]1466       then
[1289]1467             # FILES CAN ONLY BE TRANSFERED VIA DKRZ'S ARCHIVE-SERVER
[1094]1468          scp  $PORTOPT  $job_to_send  ${remote_user}@136.172.44.205:${job_catalog}/$job_on_remhost
[1]1469       else
[1468]1470          scp  $PORTOPT  $job_to_send  ${remote_user}@${remote_address}:${job_catalog}/$job_on_remhost
[1]1471       fi
[82]1472       if [[ $? = 1 ]]
1473       then
1474          locat=scp; exit
1475       fi
1476       if [[ $remote_host = ibms ]]
1477       then
1478          job_catalog=$job_catalog_save
1479       fi
[1]1480       [[ $verify = true ]]  &&  printf "\n >>> finished\n"
1481    else
1482       eval  job_catalog=$job_catalog
1483       cp  $job_to_send  ${job_catalog}/$job_on_remhost
1484    fi
1485
1486
1487
[1289]1488       # START NQS- / LOADLEVELER-JOB
[1]1489    if [[ $remote_host != $local_host ]]
1490    then
1491       [[ $verify = true ]]  &&  printf "\n >>> submitting job using \"qsub\"...\n"
[635]1492
[1289]1493       if [[ $remote_host = ibmku ]]
[1]1494       then
[1468]1495          ssh  $SSH_PORTOPT $remote_address  -l $remote_user  "cd $job_catalog; $submcom $job_on_remhost"
[1099]1496       elif [[ $remote_host = lcflow ]]
1497       then
[1468]1498          /usr/bin/ssh  $SSH_PORTOPT $remote_address  -l $remote_user  "$init_cmds $module_calls cd $job_catalog; $submcom $job_on_remhost"
[1]1499       else
[1468]1500          ssh  $SSH_PORTOPT $remote_address  -l $remote_user  "cd $job_catalog; $submcom $job_on_remhost; rm $job_on_remhost"
[82]1501       fi
[1]1502
1503       [[ $verify = true ]]  &&  printf " >>> o.k.\n"
1504    else
1505       cd  $job_catalog
[1289]1506       if [[ $(echo $local_host | cut -c1-3) = ibm  ||  $(echo $local_host | cut -c1-6) = lccray ]]
[1]1507       then
1508          eval  $submcom  $job_on_remhost
[1289]1509       elif [[  $local_host = lctit  ||  $localhost = lcxe6  ||  $localhost = lck  || $localhost = lckordi ||  $localhost = lcyon || $localhost = lcsb  ||  $localhost = lckyu* ]]
[108]1510       then
[635]1511          chmod  u+x  $job_on_remhost
[108]1512          eval  $submcom  $job_on_remhost
[1]1513       elif [[ $local_host = nech ]]
1514       then
1515          if [[ $queue = default ]]
1516          then
[799]1517             eval  $submcom  $job_on_remhost
[1]1518          else
[799]1519             eval  $submcom  -q $queue  $job_on_remhost
[1]1520          fi
1521       else
1522          qsub  $job_on_remhost
1523       fi
[622]1524
[1289]1525          # JOBFILE MUST NOT BE DELETED ON lctit/ibmku/lcflow. THIS WILL BE DONE
1526          # AT THE END OF THE JOB
[1099]1527       if [[ $local_host != lctit  &&  $local_host != ibmku  &&  $local_host != lcflow ]]
[622]1528       then
1529          rm  $job_on_remhost
1530       fi
[1]1531       cd  -  > /dev/null
1532    fi
1533 fi
1534
1535
1536
[1289]1537    # FINAL ACTIONS
[1]1538 if [[ $no_submit = false ]]
1539 then
[1099]1540    rm  -f $job_to_send
[1]1541 fi
[1266]1542 [[ $verify = true ]]  &&  printf "\n\n *** SUBJOB finished \n\n"
Note: See TracBrowser for help on using the repository browser.