source: trunk/libIGCM/AA_pack_output @ 712

Last change on this file since 712 was 712, checked in by aclsce, 12 years ago
  • Added TaskType? variable in jobs to specify the type of the job (computing or post-processing).
  • Added the use of TaskType? variable in Curie libsys to source the appropriate .atlas_env* configuration file.
  • Property svn:keywords set to Revision Author Date
File size: 17.9 KB
Line 
1#-Q- cesium #!/bin/ksh
2#-Q- cesium ######################
3#-Q- cesium ## CESIUM   CEA ##
4#-Q- cesium ######################
5#-Q- cesium #MSUB -r PACKOUTPUT    # Nom du job
6#-Q- cesium #MSUB -N 1              # Reservation du noeud
7#-Q- cesium #MSUB -n 1              # Reservation du processus
8#-Q- cesium #MSUB -T 86400          # Limite de temps elapsed du job
9#-Q- cesium #MSUB -E "-j o"
10#-Q- cesium #MSUB -E "-S /bin/ksh"
11#-Q- curie ######################
12#-Q- curie ## CURIE   TGCC/CEA ##
13#-Q- curie ######################
14#-Q- curie #MSUB -r PACKOUTPUT     # Nom du job               
15#-Q- curie #MSUB -eo
16#-Q- curie #MSUB -n 1              # Reservation du processus
17#-Q- curie #MSUB -T 86400          # Limite de temps elapsed du job
18#-Q- curie #MSUB -E '--no-requeue'
19#-Q- curie #MSUB -q large
20#-Q- curie #MSUB -Q normal
21#-Q- curie set +x
22#-Q- platine #!/usr/bin/ksh
23#-Q- platine ###################
24#-Q- platine ## PLATINE   CEA ##
25#-Q- platine ###################
26#-Q- platine #BSUB -J PACKOUTPUT            # Nom du job
27#-Q- platine #BSUB -N                        # message a la fin du job
28#-Q- platine #BSUB -n 1                      # reservation des processeurs pour le job
29#-Q- platine #BSUB -W 4:00                   # Limite temps
30#-Q- platine #BSUB -q post                   # Passage en queue post
31#-Q- platine #BSUB -L /bin/ksh
32#-Q- sx8brodie #!/bin/ksh
33#-Q- sx8brodie #######################
34#-Q- sx8brodie ## SX8BRODIE   IDRIS ##
35#-Q- sx8brodie #######################
36#-Q- sx8brodie # Temps Elapsed max. d'une requete hh:mm:ss
37#-Q- sx8brodie # @ wall_clock_limit = 10:00:00
38#-Q- sx8brodie # Nom du travail LoadLeveler
39#-Q- sx8brodie # @ job_name   = PACKOUTPUT
40#-Q- sx8brodie # Fichier de sortie standard du travail
41#-Q- sx8brodie # @ output     = $(job_name).$(jobid)
42#-Q- sx8brodie # Fichier de sortie d'erreur du travail
43#-Q- sx8brodie # @ error      =  $(job_name).$(jobid)
44#-Q- sx8brodie # pour recevoir un mail en cas de depassement du temps Elapsed (ou autre pb.)
45#-Q- sx8brodie # @ notification = error
46#-Q- sx8brodie # @ environment  = $DEBUG_debug ; $MODIPSL ; $libIGCM ; $libIGCM_SX ; $SUBMIT_DIR ; $DateBegin ; $DateEnd ; $PeriodPack ; $StandAlone ; $MASTER
47#-Q- sx8brodie # @ queue
48#-Q- aix6 #!/bin/ksh
49#-Q- aix6 #######################
50#-Q- aix6 ##   VARGAS   IDRIS  ##
51#-Q- aix6 #######################
52#-Q- aix6 # Temps Elapsed max. d'une requete hh:mm:ss
53#-Q- aix6 # @ wall_clock_limit = 10:00:00
54#-Q- aix6 # Nom du travail LoadLeveler
55#-Q- aix6 # @ job_name   = PACKOUTPUT
56#-Q- aix6 # Fichier de sortie standard du travail
57#-Q- aix6 # @ output     = $(job_name).$(jobid)
58#-Q- aix6 # Fichier de sortie d'erreur du travail
59#-Q- aix6 # @ error      =  $(job_name).$(jobid)
60#-Q- aix6 # pour recevoir un mail en cas de depassement du temps Elapsed (ou autre pb.)
61#-Q- aix6 # @ notification = error
62#-Q- aix6 # @ environment  = $DEBUG_debug ; $MODIPSL ; $libIGCM ; $libIGCM_SX ; $SUBMIT_DIR ; $DateBegin ; $DateEnd ; $PeriodPack ; $StandAlone ; $MASTER
63#-Q- aix6 # @ queue
64#-Q- ulam #!/bin/ksh
65#-Q- ulam #######################
66#-Q- ulam ##   VARGAS   IDRIS  ##
67#-Q- ulam #######################
68#-Q- ulam # Temps Elapsed max. d'une requete hh:mm:ss
69#-Q- ulam # @ wall_clock_limit = 10:00:00
70#-Q- ulam # Nom du travail LoadLeveler
71#-Q- ulam # @ job_name   = PACKOUTPUT
72#-Q- ulam # Fichier de sortie standard du travail
73#-Q- ulam # @ output     = $(job_name).$(jobid)
74#-Q- ulam # Fichier de sortie d'erreur du travail
75#-Q- ulam # @ error      =  $(job_name).$(jobid)
76#-Q- ulam # pour recevoir un mail en cas de depassement du temps Elapsed (ou autre pb.)
77#-Q- ulam # @ notification = error
78#-Q- ulam # @ environment  = $DEBUG_debug ; $MODIPSL ; $libIGCM ; $libIGCM_SX ; $SUBMIT_DIR ; $DateBegin ; $DateEnd ; $PeriodPack ; $StandAlone ; $MASTER
79#-Q- ulam # @ queue
80#-Q- sx8mercure #!/bin/ksh
81#-Q- sx8mercure ######################
82#-Q- sx8mercure ## SX8MERCURE   CEA ##
83#-Q- sx8mercure ######################
84#-Q- sx8mercure #PBS -N PACKOUTPUT          # Nom du job
85#-Q- sx8mercure #PBS -j o                    # regroupement des stdout et stderr
86#-Q- sx8mercure #PBS -S /usr/bin/ksh         # shell de soumission
87#-Q- sx8mercure #PBS -l memsz_job=1gb        # Limite memoire a 1 Go
88#-Q- sx8mercure #PBS -l cputim_job=48:00:00  # Limite temps a 1 heures
89#-Q- sx8mercure #PBS -q scalaire
90#-Q- sx8mercure #PBS -r n
91#-Q- sx9mercure #!/bin/ksh
92#-Q- sx9mercure ######################
93#-Q- sx9mercure ## SX9MERCURE  CCRT ##
94#-Q- sx9mercure ######################
95#-Q- sx9mercure #PBS -N PACKOUTPUT          # Nom du job
96#-Q- sx9mercure #PBS -j o                    # regroupement des stdout et stderr
97#-Q- sx9mercure #PBS -S /usr/bin/ksh         # shell de soumission
98#-Q- sx9mercure #PBS -l memsz_job=1gb        # Limite memoire a 1 Go
99#-Q- sx9mercure #PBS -l elapstim_req=48:00:00  # Limite temps a 1 heures
100#-Q- sx9mercure #PBS -q scalaire
101#-Q- sx9mercure #PBS -r n
102#-Q- titane #!/bin/ksh
103#-Q- titane ######################
104#-Q- titane ## TITANE   CEA     ##
105#-Q- titane ######################
106#-Q- titane #MSUB -r PACKOUTPUT    # Nom du job
107#-Q- titane #MSUB -eo
108#-Q- titane #MSUB -n 1              # Reservation du processus
109#-Q- titane #MSUB -T 86400          # Limite de temps elapsed du job
110#-Q- titane #MSUB -q mono
111#-Q- titane #MSUB -E '-rn'
112#-Q- lxiv8 ######################
113#-Q- lxiv8 ## OBELIX      LSCE ##
114#-Q- lxiv8 ######################
115#-Q- lxiv8 #PBS -N PACKOUTPUT
116#-Q- lxiv8 #PBS -m a
117#-Q- lxiv8 #PBS -j oe
118#-Q- lxiv8 #PBS -q medium
119#-Q- lxiv8 #PBS -o PACKOUTPUT.$$
120#-Q- lxiv8 #PBS -S /bin/ksh
121#-Q- default #!/bin/ksh
122#-Q- default ##################
123#-Q- default ## DEFAULT HOST ##
124#-Q- default ##################
125
126#**************************************************************
127# Author: Sebastien Denvil
128# Contact: Sebastien.Denvil__at__ipsl.jussieu.fr
129# $Revision::                                          $ Revision of last commit
130# $Author::                                            $ Author of last commit
131# $Date::                                              $ Date of last commit
132# IPSL (2006)
133#  This software is governed by the CeCILL licence see libIGCM/libIGCM_CeCILL.LIC
134#
135#**************************************************************
136
137#set -eu
138#set -vx
139
140date
141
142#D- Task type (computing or post-processing)
143TaskType=post-processing
144
145########################################################################
146
147#D- Flag to determine if this job in a standalone mode
148#D- Default : value from AA_job if any
149StandAlone=${StandAlone:=true}
150
151#D- Path to libIGCM
152#D- Default : value from AA_job if any
153# WARNING For StandAlone use : To run this script on some machine (ulam and cesium)
154# WARNING you must check MirrorlibIGCM variable in sys library.
155# WARNING If this variable is true, you must use libIGCM_POST path instead
156# WARNING of your running libIGCM directory.
157libIGCM=${libIGCM:=::modipsl::/libIGCM}
158
159#-D- $hostname of the MASTER job when SUBMIT_DIR is not visible on postprocessing computer.
160MASTER=${MASTER:=mercure|titane|brodie|vargas}
161
162#D- Flag to determine begin date for restart pack
163#D- Default : value from AA_job if any
164DateBegin=${DateBegin:=20000101}
165
166#D- Flag to determine end date for restart pack
167#D- Default : value from AA_job if any
168DateEnd=${DateEnd:=20691231}
169
170#D- Flag to determine pack period
171#D- Default : value from AA_job if any
172PeriodPack=${PeriodPack:=10Y}
173
174#D- Uncomment to run interactively
175#D- For testing purpose, will be remove
176#SUBMIT_DIR=${PWD}
177#RUN_DIR_PATH=${SCRATCHDIR}/Pack_Test
178
179#D- Increased verbosity (1, 2, 3)
180#D- Default : value from AA_job if any
181Verbosity=${Verbosity:=3}
182
183#D- Low level debug : to bypass lib test checks and stack construction
184#D- Default : value from AA_job if any
185DEBUG_debug=${DEBUG_debug:=false}
186
187########################################################################
188
189. ${libIGCM}/libIGCM_debug/libIGCM_debug.ksh
190( ${DEBUG_debug} ) && IGCM_debug_Check
191. ${libIGCM}/libIGCM_card/libIGCM_card.ksh
192( ${DEBUG_debug} ) && IGCM_card_Check
193. ${libIGCM}/libIGCM_date/libIGCM_date.ksh
194( ${DEBUG_debug} ) && IGCM_date_Check
195#-------
196. ${libIGCM}/libIGCM_sys/libIGCM_sys.ksh
197. ${libIGCM}/libIGCM_post/libIGCM_post.ksh
198
199########################################################################
200
201#set -vx
202
203RUN_DIR=${RUN_DIR_PATH}
204IGCM_sys_MkdirWork ${RUN_DIR}
205IGCM_sys_Cd ${RUN_DIR}
206
207# ------------------------------------------------------------------
208# Test if all was right before proceeding further
209# ------------------------------------------------------------------
210IGCM_debug_Verif_Exit_Post
211
212if [ ${StandAlone} = true ] ; then
213    CARD_DIR=${SUBMIT_DIR}
214else
215    CARD_DIR=${RUN_DIR_PATH}
216    IGCM_sys_Get_Master ${SUBMIT_DIR}/config.card ${RUN_DIR_PATH}
217    IGCM_sys_Get_Master ${SUBMIT_DIR}/run.card    ${RUN_DIR_PATH}
218    IGCM_sys_Get_Master ${SUBMIT_DIR}/COMP        ${RUN_DIR_PATH}
219    IGCM_sys_Get_Master ${SUBMIT_DIR}/POST        ${RUN_DIR_PATH}
220fi
221
222#====================================================
223# First of all
224#
225# Read config.card
226
227IGCM_card_DefineArrayFromSection     ${CARD_DIR}/config.card UserChoices
228typeset option
229for option in ${config_UserChoices[*]} ; do
230  IGCM_card_DefineVariableFromOption ${CARD_DIR}/config.card UserChoices ${option}
231done
232#
233IGCM_card_DefineArrayFromSection     ${CARD_DIR}/config.card Ensemble
234typeset option
235for option in ${config_Ensemble[*]} ; do
236  IGCM_card_DefineVariableFromOption ${CARD_DIR}/config.card Ensemble ${option}
237done
238#
239echo
240IGCM_debug_Print 1 "DefineArrayFromOption  : config_UserChoices"
241IGCM_debug_PrintVariables 3 config_UserChoices_JobName
242IGCM_debug_PrintVariables 3 config_UserChoices_LongName
243IGCM_debug_PrintVariables 3 config_UserChoices_TagName
244IGCM_debug_PrintVariables 3 config_UserChoices_SpaceName
245IGCM_debug_PrintVariables 3 config_UserChoices_ExperimentName
246IGCM_debug_PrintVariables 3 config_UserChoices_CalendarType
247IGCM_debug_PrintVariables 3 config_UserChoices_DateBegin
248IGCM_debug_PrintVariables 3 config_UserChoices_DateEnd
249IGCM_debug_PrintVariables 3 config_UserChoices_PeriodLength
250IGCM_debug_PrintVariables 3 config_Ensemble_EnsembleName
251IGCM_debug_PrintVariables 3 config_Ensemble_EnsembleDate
252echo
253
254# Get ListOfComponents to drive the loop over find
255IGCM_card_DefineArrayFromSection ${CARD_DIR}/config.card ListOfComponents
256
257#====================================================
258#R_SAVE : Job output storage directory
259#R_BUFR : Job output buffered directory
260
261if ( [ ! X${config_UserChoices_SpaceName} = X ] && [ ! X${config_UserChoices_ExperimentName} = X ] ) ; then
262  FreeName=$( echo ${config_UserChoices_JobName} | sed 's/.*_//' )
263  if ( [ ! X${config_Ensemble_EnsembleName} = X ] && [ ! X${config_Ensemble_EnsembleDate} = X ] ) ; then
264    R_SAVE=${R_OUT}/${config_UserChoices_TagName}/${config_UserChoices_SpaceName}/${config_UserChoices_ExperimentName}/${config_Ensemble_EnsembleName}/${config_Ensemble_EnsembleDate}/${FreeName}
265    R_BUFR=${R_BUF}/${config_UserChoices_TagName}/${config_UserChoices_SpaceName}/${config_UserChoices_ExperimentName}/${config_Ensemble_EnsembleName}/${config_Ensemble_EnsembleDate}/${FreeName}
266  else
267    R_SAVE=${R_OUT}/${config_UserChoices_TagName}/${config_UserChoices_SpaceName}/${config_UserChoices_ExperimentName}/${FreeName}
268    R_BUFR=${R_BUF}/${config_UserChoices_TagName}/${config_UserChoices_SpaceName}/${config_UserChoices_ExperimentName}/${FreeName}
269  fi
270else
271  if ( [ ! X${config_Ensemble_EnsembleName} = X ] && [ ! X${config_Ensemble_EnsembleDate} = X ] ) ; then
272    R_SAVE=${R_OUT}/${config_UserChoices_TagName}/${config_Ensemble_EnsembleName}/${config_Ensemble_EnsembleDate}/${config_UserChoices_JobName}
273    R_BUFR=${R_BUF}/${config_UserChoices_TagName}/${config_Ensemble_EnsembleName}/${config_Ensemble_EnsembleDate}/${config_UserChoices_JobName}
274  else
275    R_SAVE=${R_OUT}/${config_UserChoices_TagName}/${config_UserChoices_JobName}
276    R_BUFR=${R_BUF}/${config_UserChoices_TagName}/${config_UserChoices_JobName}
277  fi
278fi
279IGCM_sys_TestDirArchive ${R_SAVE}
280[ $? != 0 ] && IGCM_debug_Exit "IGCM_sys_TestDirArchive"
281
282# Where to store used file list /!\ TEMPORARY /!\
283STORE_DEBUG=${R_SAVE}/DEBUG
284
285# Switch to script variables meaning (try to be compatible with ipsl_pack TGCC moving procedure)
286JobName=${config_UserChoices_JobName}
287echo $JobName $DateBegin $DateEnd
288
289# ------------------------------------------------------------------
290# Test if all was right before proceeding further
291# ------------------------------------------------------------------
292IGCM_debug_Verif_Exit_Post
293
294IGCM_debug_Print 1 "Check coherence between PackFrequency and PeriodLength"
295IGCM_post_CheckModuloFrequency PeriodPack config_UserChoices_PeriodLength NbPeriodPerFrequency
296# ------------------------------------------------------------------
297# Test if all was right before proceeding further
298# ------------------------------------------------------------------
299IGCM_debug_Verif_Exit_Post
300
301IGCM_debug_Print 1 "We must process ${NbPeriodPerFrequency} files for each pack"
302
303# Init loop
304date_begin_pack=${DateBegin}
305date_end_simulation=${DateEnd}
306number_pack=1
307
308IGCM_debug_PrintVariables 3 date_begin_pack
309IGCM_debug_PrintVariables 3 date_end_simulation
310
311while [ ${date_begin_pack} -le ${date_end_simulation} ] ; do
312
313  IGCM_debug_PrintVariables 3 number_pack
314  DaysTemp=$( IGCM_date_DaysInCurrentPeriod ${date_begin_pack} ${PeriodPack} )
315  date_end_pack=$( IGCM_date_AddDaysToGregorianDate ${date_begin_pack} $(( ${DaysTemp} - 1 )) )
316
317  for comp in ${config_ListOfComponents[*]} ; do
318    dirList=$( find ${R_BUFR}/${comp}/Output -maxdepth 1 -mindepth 1 -type d )
319    for dir in ${dirList} ; do
320      # dirID is like ATM.Output.MO
321      dirID=$( echo $dir | sed "s:${R_BUFR}/::" | sed "s:/:.:g" )
322      # Sort what's in the directory
323      find ${dir} -type f -name "${JobName}*.nc" -ls | sort -k 11 > liste_files.${dirID}.txt
324      # How much file type. Example : 1M_histmthCOSP.nc, 1M_histmth.nc, 1M_histmthNMC.nc, 1M_paramLMDZ_phy.nc
325      # /!\ fileType include the .nc extension /!\
326      fileType=$( gawk '{print $11}' liste_files.${dirID}.txt | gawk -F$dir/ '{print $2}' | sed "s:${JobName}_[0-9]\{8,9\}_[0-9]\{8,9\}_::g" | sort | uniq )
327      # Loop over the file type and pack them when in between date_begin_pack and date_end_pack
328      for myType in ${fileType} ; do
329        grep ${myType} liste_files.${dirID}.txt > liste_files.${dirID}.${myType}.txt
330        nbfile=0
331        for file in $( gawk '{print $11}' liste_files.${dirID}.${myType}.txt ); do
332          extract_date_file=$( echo ${file}  | sed -e "s/.*${JobName}_[0-9]*_//" )
333          date_file=$( echo ${extract_date_file} | sed 's/\([0-9]\{8\}\)_.*$/\1/g' )
334          # echo pack n°${number_pack}  ${date_file} ${date_begin_pack} ${date_end_pack}
335          if [ ${date_file} -le ${date_end_pack} ] && [ ${date_file} -ge ${date_begin_pack} ] ; then
336            echo ${file} >> liste_pack_${myType}_${date_begin_pack}_${date_end_pack}.txt
337            ncdump -h ${file} | grep -E 'float|double' | cut -f 1 -d '(' | cut -f 2 -d ' ' >> liste_variables_${myType}_${date_begin_pack}_${date_end_pack}.txt
338            (( nbfile = nbfile + 1 ))
339          fi
340        done
341
342        if ( [ X${StandAlone} = Xtrue ] && [ ${nbfile} = 0 ] ) ; then
343          IGCM_debug_Print 1 "We found ${nbfile} files to process"
344          IGCM_debug_Print 1 "We should have found ${NbPeriodPerFrequency} files"
345          IGCM_debug_Print 1 "As we are in StandAlone mode we consider the job has been done and move to the next file type"
346          continue
347        fi
348
349        # Select list of variables to work with
350        list_var=$( cat liste_variables_${myType}_${date_begin_pack}_${date_end_pack}.txt | sort | uniq -c | awk -v nbfile=$nbfile '{if ($1 != nbfile) {print $2}}' | paste -s -d ',' )
351        liste_file_tmp=$( for i in $( cat liste_pack_${myType}_${date_begin_pack}_${date_end_pack}.txt ) ; do basename $i ; done )
352        # Create packed files
353        IGCM_debug_Print 1 "Ncrcat ongoing for ${dir} and ${myType}"
354        if [ ! ${nbfile} = ${NbPeriodPerFrequency} ] ; then
355          IGCM_debug_Print 1 "Number of files to process is not equal to what it should be"
356          IGCM_debug_Print 1 "We found ${nbfile} files and it should have been ${NbPeriodPerFrequency} files"
357          IGCM_debug_Exit "ERROR in number of files to process. STOP HERE INCLUDING THE COMPUTING JOB"
358          IGCM_debug_Verif_Exit
359        fi
360        output=${JobName}_${date_begin_pack}_${date_end_pack}_${myType}
361        #cat liste_pack_${myType}_${date_begin_pack}_${date_end_pack}.txt | xargs ncrcat -v ${list_var} -o ${output}
362        if [ X${list_var} = X ] ; then
363          IGCM_sys_ncrcat -p ${dir} ${liste_file_tmp} --output ${output}
364        else
365          IGCM_sys_ncrcat -x -v ${list_var} -p ${dir} ${liste_file_tmp} --output ${output}
366        fi
367        # ------------------------------------------------------------------
368        # Test if all was right before proceeding further
369        # ------------------------------------------------------------------
370        IGCM_debug_Verif_Exit_Post
371        # Save it
372        IGCM_sys_Put_Out ${output} ${R_SAVE}/$( echo $dir | sed "s:${R_BUFR}/::" )/${output}
373        # Clean file produced by ncrcat
374        IGCM_sys_Rm ${output}
375        # ------------------------------------------------------------------
376        # Test if all was right before proceeding further
377        # ------------------------------------------------------------------
378        IGCM_debug_Verif_Exit_Post
379        # Clean files used by ncrcat
380        cat liste_pack_${myType}_${date_begin_pack}_${date_end_pack}.txt | xargs rm
381        # Save the list of files that has been pack (ncrcat)
382        #mv liste_pack_${myType}_${date_begin_pack}_${date_end_pack}.txt ${STORE_DEBUG}
383        IGCM_debug_Print 1 "Ncrcat and cleaning done for ${dir} and ${myType}"
384        echo
385      done
386    done
387  done
388  (( number_pack = number_pack + 1 ))
389  # Add 1 day to date_end_pack to have the new date_begin_pack
390  date_begin_pack=$( IGCM_date_AddDaysToGregorianDate ${date_end_pack} 1 )
391done
392
393# Flush post-processing submission
394if [ -f ${R_BUFR}/FlushPost_${DateEnd}.ksh ] ; then
395  . ${R_BUFR}/FlushPost_${DateEnd}.ksh
396  IGCM_FlushPost
397  #IGCM_sys_Rm -f ${R_BUFR}/FlushPost_${DateEnd}.ksh
398fi
399
400# Clean RUN_DIR_PATH (necessary for cesium and titane only)
401IGCM_sys_RmRunDir -Rf ${RUN_DIR_PATH}
402
403date
Note: See TracBrowser for help on using the repository browser.