1 | ###################### |
---|
2 | ## IRENE TGCC/CEA ## |
---|
3 | ###################### |
---|
4 | #MSUB -r PACKOUTPUT # Job name |
---|
5 | #MSUB -o PACKOUTPUT.out_%I |
---|
6 | #MSUB -e PACKOUTPUT.out_%I |
---|
7 | #MSUB -n 1 # Number of cores |
---|
8 | #MSUB -T 36000 # Maximum elapsed time |
---|
9 | #MSUB -q xlarge |
---|
10 | #MSUB -c 8 |
---|
11 | #MSUB -Q normal |
---|
12 | #MSUB -A scecmip6 |
---|
13 | #MSUB -m store,work,scratch |
---|
14 | set +x |
---|
15 | |
---|
16 | #************************************************************** |
---|
17 | # Author: Sebastien Denvil |
---|
18 | # Contact: Sebastien.Denvil__at__ipsl.jussieu.fr |
---|
19 | # $Revision:: 1468 $ Revision of last commit |
---|
20 | # $Author:: mafoipsl $ Author of last commit |
---|
21 | # $Date:: 2018-08-14 15:55:41 +0200 (Tue, 14 Aug 2018) $ Date of last commit |
---|
22 | # IPSL (2006) |
---|
23 | # This software is governed by the CeCILL licence see libIGCM/libIGCM_CeCILL.LIC |
---|
24 | # |
---|
25 | #************************************************************** |
---|
26 | |
---|
27 | #set -eu |
---|
28 | #set -vx |
---|
29 | |
---|
30 | date |
---|
31 | |
---|
32 | #D- Task type DO NOT CHANGE (computing, post-processing or checking) |
---|
33 | TaskType=post-processing |
---|
34 | |
---|
35 | ######################################################################## |
---|
36 | |
---|
37 | #D- Flag to determine if this job in a standalone mode |
---|
38 | #D- Default : value from AA_job if any |
---|
39 | StandAlone=${StandAlone:=true} |
---|
40 | |
---|
41 | #D- Path to libIGCM |
---|
42 | #D- Default : value from AA_job if any |
---|
43 | # WARNING For StandAlone use : To run this script on some machine (ulam and cesium) |
---|
44 | # WARNING you must check MirrorlibIGCM variable in sys library. |
---|
45 | # WARNING If this variable is true, you must use libIGCM_POST path instead |
---|
46 | # WARNING of your running libIGCM directory. |
---|
47 | libIGCM=${libIGCM:=/ccc/cont003/dsku/perle1/home/app/gencmip6/devillim/modipsl_v20fev/libIGCM} |
---|
48 | |
---|
49 | #-D- $hostname of the MASTER job when SUBMIT_DIR is not visible on postprocessing computer. |
---|
50 | MASTER=${MASTER:=irene} |
---|
51 | |
---|
52 | #D- Flag to determine begin date for restart pack |
---|
53 | #D- Default : value from AA_job if any |
---|
54 | DateBegin=${DateBegin:=21000101} |
---|
55 | |
---|
56 | #D- Flag to determine end date for restart pack |
---|
57 | #D- Default : value from AA_job if any |
---|
58 | DateEnd=${DateEnd:=21001231} |
---|
59 | |
---|
60 | #D- Flag to determine pack period |
---|
61 | #D- Default : value from AA_job if any |
---|
62 | PeriodPack=${PeriodPack:=1Y} |
---|
63 | |
---|
64 | #D- Uncomment to run interactively |
---|
65 | #D- For testing purpose, will be remove |
---|
66 | #SUBMIT_DIR=${PWD} |
---|
67 | #RUN_DIR_PATH=${SCRATCHDIR}/Pack_Test |
---|
68 | |
---|
69 | #D- Increased verbosity (1, 2, 3) |
---|
70 | #D- Default : value from AA_job if any |
---|
71 | Verbosity=${Verbosity:=3} |
---|
72 | |
---|
73 | #D- Low level debug : to bypass lib test checks and stack construction |
---|
74 | #D- Default : value from AA_job if any |
---|
75 | DEBUG_debug=${DEBUG_debug:=false} |
---|
76 | |
---|
77 | ######################################################################## |
---|
78 | |
---|
79 | . ${libIGCM}/libIGCM_debug/libIGCM_debug.ksh |
---|
80 | . ${libIGCM}/libIGCM_card/libIGCM_card.ksh |
---|
81 | . ${libIGCM}/libIGCM_date/libIGCM_date.ksh |
---|
82 | #------- |
---|
83 | . ${libIGCM}/libIGCM_sys/libIGCM_sys.ksh |
---|
84 | . ${libIGCM}/libIGCM_config/libIGCM_config.ksh |
---|
85 | . ${libIGCM}/libIGCM_post/libIGCM_post.ksh |
---|
86 | #------- |
---|
87 | RUN_DIR=${RUN_DIR_PATH} |
---|
88 | IGCM_sys_MkdirWork ${RUN_DIR} |
---|
89 | IGCM_sys_Cd ${RUN_DIR} |
---|
90 | #------- |
---|
91 | ( ${DEBUG_debug} ) && IGCM_debug_Check |
---|
92 | ( ${DEBUG_debug} ) && IGCM_card_Check |
---|
93 | ( ${DEBUG_debug} ) && IGCM_date_Check |
---|
94 | |
---|
95 | ######################################################################## |
---|
96 | |
---|
97 | #set -vx |
---|
98 | |
---|
99 | # ------------------------------------------------------------------ |
---|
100 | # Test if all was right before proceeding further |
---|
101 | # ------------------------------------------------------------------ |
---|
102 | IGCM_debug_Verif_Exit |
---|
103 | |
---|
104 | if [ ${StandAlone} = true ] ; then |
---|
105 | CARD_DIR=${SUBMIT_DIR} |
---|
106 | else |
---|
107 | CARD_DIR=${RUN_DIR_PATH} |
---|
108 | IGCM_sys_Get_Master ${SUBMIT_DIR}/config.card ${RUN_DIR_PATH} |
---|
109 | IGCM_sys_Get_Master ${SUBMIT_DIR}/run.card ${RUN_DIR_PATH} |
---|
110 | IGCM_sys_Get_Master ${SUBMIT_DIR}/COMP ${RUN_DIR_PATH} |
---|
111 | IGCM_sys_Get_Master ${SUBMIT_DIR}/POST ${RUN_DIR_PATH} |
---|
112 | fi |
---|
113 | |
---|
114 | #================================== |
---|
115 | # First of all |
---|
116 | # |
---|
117 | # Read libIGCM compatibility version in config.card |
---|
118 | # Read UserChoices section |
---|
119 | # Read Ensemble section |
---|
120 | # Read Post section |
---|
121 | # Define all netcdf output directories |
---|
122 | #================================== |
---|
123 | IGCM_config_CommonConfiguration ${CARD_DIR}/config.card |
---|
124 | |
---|
125 | # ------------------------------------------------------------------ |
---|
126 | # Activate BigBrother so as to supervise this job |
---|
127 | # ------------------------------------------------------------------ |
---|
128 | IGCM_debug_BigBro_Initialize |
---|
129 | |
---|
130 | #================================== |
---|
131 | # Read ListOfComponents section |
---|
132 | # to drive the loop over find |
---|
133 | IGCM_card_DefineArrayFromSection ${CARD_DIR}/config.card ListOfComponents |
---|
134 | |
---|
135 | #================================== |
---|
136 | # Test and set up directories |
---|
137 | #================================== |
---|
138 | IGCM_sys_TestDirArchive ${R_SAVE} |
---|
139 | [ $? != 0 ] && IGCM_debug_Exit "IGCM_sys_TestDirArchive" |
---|
140 | |
---|
141 | # Where to store used file list /!\ TEMPORARY /!\ |
---|
142 | STORE_DEBUG=${R_SAVE}/DEBUG |
---|
143 | |
---|
144 | # Switch to script variables meaning (try to be compatible with ipsl_pack TGCC moving procedure) |
---|
145 | JobName=${config_UserChoices_JobName} |
---|
146 | echo $JobName $DateBegin $DateEnd |
---|
147 | |
---|
148 | # ------------------------------------------------------------------ |
---|
149 | # Test if all was right before proceeding further |
---|
150 | # ------------------------------------------------------------------ |
---|
151 | IGCM_debug_Verif_Exit |
---|
152 | |
---|
153 | IGCM_debug_Print 1 "Check coherence between PackFrequency and PeriodLength" |
---|
154 | IGCM_post_CheckModuloFrequency PeriodPack config_UserChoices_PeriodLength NbPeriodPerFrequency |
---|
155 | # ------------------------------------------------------------------ |
---|
156 | # Test if all was right before proceeding further |
---|
157 | # ------------------------------------------------------------------ |
---|
158 | IGCM_debug_Verif_Exit |
---|
159 | |
---|
160 | IGCM_debug_Print 1 "We must process ${NbPeriodPerFrequency} files for each pack" |
---|
161 | |
---|
162 | # Init loop |
---|
163 | date_begin_pack=${DateBegin} |
---|
164 | date_end_simulation=${DateEnd} |
---|
165 | number_pack=1 |
---|
166 | |
---|
167 | IGCM_debug_PrintVariables 3 date_begin_pack |
---|
168 | IGCM_debug_PrintVariables 3 date_end_simulation |
---|
169 | |
---|
170 | while [ ${date_begin_pack} -le ${date_end_simulation} ] ; do |
---|
171 | |
---|
172 | IGCM_debug_PrintVariables 3 number_pack |
---|
173 | DaysTemp=$( IGCM_date_DaysInCurrentPeriod ${date_begin_pack} ${PeriodPack} ) |
---|
174 | date_end_pack=$( IGCM_date_AddDaysToGregorianDate ${date_begin_pack} $(( ${DaysTemp} - 1 )) ) |
---|
175 | |
---|
176 | for comp in ${config_ListOfComponents[*]} ; do |
---|
177 | dirList=$( find ${R_BUFR}/${comp}/Output -maxdepth 1 -mindepth 1 -type d ) |
---|
178 | for dir in ${dirList} ; do |
---|
179 | # dirID is like ATM.Output.MO |
---|
180 | dirID=$( echo $dir | sed "s:${R_BUFR}/::" | sed "s:/:.:g" ) |
---|
181 | # Sort what's in the directory |
---|
182 | find ${dir} -type f -name "${JobName}*.nc" -ls | sort -k 11 > liste_files.${dirID}.txt |
---|
183 | # How much file type. Example : 1M_histmthCOSP.nc, 1M_histmth.nc, 1M_histmthNMC.nc, 1M_paramLMDZ_phy.nc |
---|
184 | # /!\ fileType include the .nc extension /!\ |
---|
185 | fileType=$( 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 ) |
---|
186 | # Loop over the file type and pack them when in between date_begin_pack and date_end_pack |
---|
187 | for myType in ${fileType} ; do |
---|
188 | grep ${myType} liste_files.${dirID}.txt > liste_files.${dirID}.${myType}.txt |
---|
189 | nbfile=0 |
---|
190 | for file in $( gawk '{print $11}' liste_files.${dirID}.${myType}.txt ); do |
---|
191 | extract_date_file=$( echo ${file} | sed -e "s/.*${JobName}_[0-9]*_//" ) |
---|
192 | date_file=$( echo ${extract_date_file} | sed 's/\([0-9]\{8\}\)_.*$/\1/g' ) |
---|
193 | # echo pack n°${number_pack} ${date_file} ${date_begin_pack} ${date_end_pack} |
---|
194 | if [ ${date_file} -le ${date_end_pack} ] && [ ${date_file} -ge ${date_begin_pack} ] ; then |
---|
195 | echo ${file} >> liste_pack_${myType}_${date_begin_pack}_${date_end_pack}.txt |
---|
196 | 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 |
---|
197 | (( nbfile = nbfile + 1 )) |
---|
198 | fi |
---|
199 | done |
---|
200 | |
---|
201 | if [ ${nbfile} = 0 ] ; then |
---|
202 | IGCM_debug_Print 1 "We found no file to process" |
---|
203 | IGCM_debug_Print 1 "We should have found ${NbPeriodPerFrequency} files" |
---|
204 | IGCM_debug_Print 1 "As some files can be produced only for some selected period we consider we can move to the next file type" |
---|
205 | continue |
---|
206 | fi |
---|
207 | |
---|
208 | # Select list of variables to work with |
---|
209 | list_var=$( cat liste_variables_${myType}_${date_begin_pack}_${date_end_pack}.txt | sort | uniq -c | awk -v nbfile=$nbfile '{if ($1 != nbfile) {print $2}}' | paste -s -d ',' ) |
---|
210 | liste_file_tmp=$( for i in $( cat liste_pack_${myType}_${date_begin_pack}_${date_end_pack}.txt ) ; do basename $i ; done ) |
---|
211 | # Create packed files |
---|
212 | IGCM_debug_Print 1 "Ncrcat ongoing for ${dir} and ${myType}" |
---|
213 | if [ ! ${nbfile} = ${NbPeriodPerFrequency} ] ; then |
---|
214 | IGCM_debug_Print 1 "Number of files to process is not equal to what it should be" |
---|
215 | IGCM_debug_Print 1 "We found ${nbfile} files and it should have been ${NbPeriodPerFrequency} files" |
---|
216 | IGCM_debug_Exit "ERROR in number of files to process. STOP HERE INCLUDING THE COMPUTING JOB" |
---|
217 | IGCM_debug_Verif_Exit |
---|
218 | fi |
---|
219 | output=${JobName}_${date_begin_pack}_${date_end_pack}_${myType} |
---|
220 | #cat liste_pack_${myType}_${date_begin_pack}_${date_end_pack}.txt | xargs ncrcat -v ${list_var} -o ${output} |
---|
221 | if [ X${list_var} = X ] ; then |
---|
222 | IGCM_sys_ncrcat -p ${dir} ${liste_file_tmp} --output ${output} |
---|
223 | else |
---|
224 | IGCM_sys_ncrcat -x -v ${list_var} -p ${dir} ${liste_file_tmp} --output ${output} |
---|
225 | fi |
---|
226 | # ------------------------------------------------------------------ |
---|
227 | # Test if all was right before proceeding further |
---|
228 | # ------------------------------------------------------------------ |
---|
229 | IGCM_debug_Verif_Exit |
---|
230 | # Save it |
---|
231 | IGCM_sys_Put_Out ${output} ${R_SAVE}/$( echo $dir | sed "s:${R_BUFR}/::" )/${output} |
---|
232 | # Clean file produced by ncrcat |
---|
233 | IGCM_sys_Rm ${output} |
---|
234 | # ------------------------------------------------------------------ |
---|
235 | # Test if all was right before proceeding further |
---|
236 | # ------------------------------------------------------------------ |
---|
237 | IGCM_debug_Verif_Exit |
---|
238 | # Clean files used by ncrcat |
---|
239 | cat liste_pack_${myType}_${date_begin_pack}_${date_end_pack}.txt | xargs rm |
---|
240 | # Save the list of files that has been pack (ncrcat) |
---|
241 | #mv liste_pack_${myType}_${date_begin_pack}_${date_end_pack}.txt ${STORE_DEBUG} |
---|
242 | IGCM_debug_Print 1 "Ncrcat and cleaning done for ${dir} and ${myType}" |
---|
243 | echo |
---|
244 | done |
---|
245 | done |
---|
246 | done |
---|
247 | (( number_pack = number_pack + 1 )) |
---|
248 | # Add 1 day to date_end_pack to have the new date_begin_pack |
---|
249 | date_begin_pack=$( IGCM_date_AddDaysToGregorianDate ${date_end_pack} 1 ) |
---|
250 | done |
---|
251 | |
---|
252 | # Flush post-processing submission |
---|
253 | if [ -f ${R_BUFR}/FlushPost_${DateEnd}.ksh ] ; then |
---|
254 | . ${R_BUFR}/FlushPost_${DateEnd}.ksh |
---|
255 | IGCM_FlushPost |
---|
256 | #IGCM_sys_Rm -f ${R_BUFR}/FlushPost_${DateEnd}.ksh |
---|
257 | fi |
---|
258 | |
---|
259 | # Clean RUN_DIR_PATH (necessary for cesium and titane only) |
---|
260 | IGCM_sys_RmRunDir -Rf ${RUN_DIR_PATH} |
---|
261 | |
---|
262 | # ------------------------------------------------------------------ |
---|
263 | # Finalize BigBrother to inform that the jobs end |
---|
264 | # ------------------------------------------------------------------ |
---|
265 | IGCM_debug_BigBro_Finalize |
---|
266 | |
---|
267 | date |
---|