|
|
@949
|
11 years |
sdipsl |
|
|
obelix case. see #125
|
|
|
@948
|
11 years |
sdipsl |
|
|
catch JobID to be able to connect computing trees and post-processing trees. See #125
Obelix not yet done.
|
|
|
@947
|
11 years |
sdipsl |
|
|
Bugfix affecting ada looking after file on gaya : must use IGCM_sys_TestDirArchive
|
|
|
@946
|
11 years |
mafoipsl |
|
|
Bug fixed in #945.
|
|
|
@945
|
11 years |
mafoipsl |
|
|
On ada, consolidation of IGCM_sys_check_quota. Exemple with : 15.69% used.
|
|
|
@944
|
11 years |
sdipsl |
|
|
we want it quiet : atlas_env_ada_bash
it will disturb debug mode otherwise
|
|
|
@943
|
11 years |
mafoipsl |
|
|
In libIGCM_config, IGCM_config_Finalize, use RmRunDir? to have a dummy call at IDRIS. TMPDIR has to be deleted by the system.
|
|
|
@941
|
11 years |
sdipsl |
|
|
Adapt instrumentation for this kind of syntax:
# IGCM_sys_ncrcat -p ${dir} ${liste_file_tmp} --output ${output}
# IGCM_sys_ncrcat -x -v ${list_var} -p ${dir} ${liste_file_tmp} --output ${output}
|
|
|
@939
|
11 years |
sdipsl |
|
|
cleanup
|
|
|
@938
|
11 years |
sdipsl |
|
|
cleanup.
replace out_rsync by /tmp/out_command.$$
|
|
|
@936
|
11 years |
sdipsl |
|
|
bugfix again with instrumentation. Should be fine now...
syntax error in libIGCM_sys
|
|
|
@934
|
11 years |
sdipsl |
|
|
Bugfix again with IGCM_sys_Get|GetBuffer? instrumentation
|
|
|
@933
|
11 years |
sdipsl |
|
|
properly remove RUN_DIR_PATH used by AA_job on curie. see #141
but add some check to be sure RUN_DIR_PATH do not point to important directory.
Indeed within AA_Job you can overwrite RUN_DIR_PATH and define something like this:
RUN_DIR_PATH=$SCRATCHDIR
You don't want to erase your $SCRATCHDIR don't you
|
|
|
@932
|
11 years |
sdipsl |
|
|
clean-up old fashion instrumentation trace
|
|
|
@931
|
11 years |
sdipsl |
|
|
Bugfix with IGCM_sys_Get|GetBuffer? instrumentation
Avoid collision with testing functions
|
|
|
@929
|
11 years |
sdipsl |
|
|
machine cleanup
whitespace cleanup
|
|
|
@928
|
11 years |
sdipsl |
|
|
clean-up
|
|
|
@927
|
11 years |
sdipsl |
|
|
remove AA_rebuild_fromArchive. AA_rebuild_fromWorkdir can do both
|
|
|
@926
|
11 years |
sdipsl |
|
|
Add instrumentation for IGCM_sys_ncrcat IGCM_sys_ncecat IGCM_sys_ncra IGCM_sys_ncks IGCM_sys_cdo see #140
stack files handling : one stack file per TaskType? : computing.stack, post-processsing, checking.stack
|
|
|
@925
|
11 years |
sdipsl |
|
|
Add IGCM_sys_Put_Out and IGCM_sys_PutBuffer_Out to the instrumented functions
|
|
|
@924
|
11 years |
sdipsl |
|
|
Enhance IGCM_debug_sizeOfTabContent prior to further work on #140
|
|
|
@923
|
11 years |
sdipsl |
|
|
cleanup
|
|
|
@922
|
11 years |
sdipsl |
|
|
Once per simulation send email containing estimated consumed hours for the whole simulation and a recommended PeriodNb?
|
|
|
@921
|
11 years |
sdipsl |
|
|
Not necessary ...
|
|
|
@920
|
11 years |
sdipsl |
|
|
factorization
|
|
|
@919
|
11 years |
sdipsl |
|
|
Bugfix IGCM_date_DaysIn*Period. See #159
|
|
|
@918
|
11 years |
sdipsl |
|
|
Active instrumentation #140 disturbs this kind of call
TarFileLocation?=$( IGCM_sys_PrepareTaredRestart ${TarFileFound?} )
|
|
|
@917
|
11 years |
sdipsl |
|
|
Adapt for diffenrent cases : copy to a file, copy to a directory, and rename during copy.
|
|
|
@916
|
11 years |
sdipsl |
|
|
add IGCM_sys_Cp to the instrumented functions list
fix a bug. In this syntax, length is a length not an indice ${variable:start:length}
|
|
|
@915
|
11 years |
sdipsl |
|
|
call IGCM_debug_PrintInfosActions only for instrumented functions
|
|
|
@914
|
11 years |
sdipsl |
|
|
typo highlighted by check functions
|
|
|
@913
|
11 years |
sdipsl |
|
|
refactoring (supervisor related blocks)
introducing revisited instrumentation from CEA audit see #140
IGCM_sys_Get* IGCM_sys_rebuild* have instrumentation
IGCM_sys_Put* IGCM_sys_nc* IGCM_sys_cdo IGCM_sys_tar not yet instrumented
Not fully tested yet. Use at your own risks
|
|
|
@912
|
11 years |
sdipsl |
|
|
remove machines that are not available anymore
cleanup
|
|
|
@911
|
11 years |
sdipsl |
|
|
- Ada compute node don't see gaya
|
|
|
@910
|
11 years |
sdipsl |
|
|
In some case Script_output was not saved. See #132
|
|
|
@909
|
11 years |
mafoipsl |
|
|
ada : don't use rsh to find gaya HOME. It breakes script and postproccessing jobs are not submitted. Use sed to switch from /linkhome/rech/... (HOME adapp) or /smphome/rech/... (HOME ada) to /u/rech/... (HOME gaya).
|
|
|
@908
|
11 years |
sdipsl |
|
|
- take into account SeasonalFrequencyOffset See #154
|
|
|
@907
|
11 years |
sdipsl |
|
|
svn propset svn:keywords "Revision Date Author" AA_clean_year
|
|
|
@906
|
11 years |
sdipsl |
|
|
- cleanup
- whitespace-cleanup
|
|
|
@905
|
11 years |
sdipsl |
|
|
-cleanup
|
|
|
@904
|
11 years |
sdipsl |
|
|
- handle *Y packed files. See #134
- whitespace-cleanup
|
|
|
@903
|
11 years |
sdipsl |
|
|
- Dods copy is controled by an option in config.card/[Post]. See #150
|
|
|
@902
|
11 years |
sdipsl |
|
|
- further enhance messages when we bring in restart files
- whitespace-cleanup
|
|
|
@901
|
11 years |
sdipsl |
|
|
- can monitor other frequency not only what's in TS_MO. see #157
|
|
|
@900
|
11 years |
labetoulle |
|
|
Remove suffix ${CumulPeriod?} from ${JobName?} when resubmitting (#144).
|
|
|
@899
|
11 years |
mafoipsl |
|
|
!AA_TimeSeries_Checker : Switch to simplified test because it's working on adapp and on curie.
At IDRIS use 'adapp' to run !TimeSeries_Checker instead of ada. HOMEGAYA (ie gaya files) is accessible (through NFS) by adapp and not by ada.
More information here : http://www.idris.fr/adapp/intro_adapp.html
|
|
|
@898
|
11 years |
sdipsl |
|
|
Enhance message when Section or Option do not exist. see #145
|
|
|
@897
|
11 years |
sdipsl |
|
|
try to enhance messages when we bring in restart files
try to avoid meaningless messages as well
see #135
|
|
|
@896
|
11 years |
sdipsl |
|
|
- ask confirmation before erasing something. see #133
|
|
|
@895
|
11 years |
sdipsl |
|
|
|
|
|
@894
|
11 years |
sdipsl |
|
|
- clearer message when error
|
|
|
@893
|
11 years |
sdipsl |
|
|
- to be able to chain DRYRUN execution
|
|
|
@892
|
11 years |
sdipsl |
|
|
|
|
|
@891
|
11 years |
sdipsl |
|
|
|
|
|
@890
|
11 years |
sdipsl |
|
|
- introduce IGCM_config_DateCoherency function see #79
- start cleaning IGCM_config_PeriodStart
- rename IGCM_config_Analyse_PeriodLength by IGCM_config_DaysInPeriodLength
|
|
|
@889
|
11 years |
sdipsl |
|
|
- to avoid future side effects with IGCM_date_DaysInCurrentPeriod, IGCM_date_DaysInPreviousPeriod, IGCM_date_DaysInNextPeriod
useful for #79
|
|
|
@888
|
11 years |
sdipsl |
|
|
- find tar file in dmf_import also for the hindcast/forecast case
|
|
|
@887
|
11 years |
mafoipsl |
|
|
On ada, add a suffix to have a unique jobname for QsubPost.
|
|
|
@886
|
11 years |
labetoulle |
|
|
Fix TimeSeries_Checker in leap calendar (see #148).
|
|
|
@885
|
11 years |
mafoipsl |
|
|
For RESTART files, patterns ${RestartJobName?}_YYYYMMDD_YYYYMMDD_restart.tar and ${RestartJobName?}_restart_YYYYMMDD_YYYYMMDD.tar (used into dmf_import) are recognized.
|
|
|
@884
|
11 years |
mafoipsl |
|
|
add treatment to recognize RESTART file with this pattern : *_20781231_20781231_*.tar. Correction of #152.
|
|
|
@883
|
11 years |
mafoipsl |
|
|
On ada , use STOREHOST variable instead of gaya directly.
|
|
|
@882
|
11 years |
mafoipsl |
|
|
On ada, since june 18th, tha variable HOME changed. It's like /smphome/... on computer nodes and like /linkhome/... on front-end and pre-post nodes. See : http://www.idris.fr/su/Shared/ldap.html.
Use rsh gaya 'echo $HOME' to set ARCHIVE in libIGCM_sys_ada.ksh
|
|
|
@881
|
11 years |
mafoipsl |
|
|
For LSCE, add asterix* with same treatment as obelix.
|
|
|
@880
|
11 years |
mafoipsl |
|
|
On curie, define time limit for post-processing jobs : rebuild and pack_output : 10h
- pack_debug : 1h
- pack_restart, monitoring and atlas : 2 h
- TS : 20 h
- SE : 10 h
On curie, add default project initialisation and usage for TimeSeries_Checker. Necessary for interactive usage of TimeSeries_Checker. Works since ins_job revision 2091.
|
|
|
@874
|
11 years |
sdipsl |
|
|
|
|
|
@873
|
11 years |
sdipsl |
|
|
- Send message related to end of job and end of simulation
|
|
|
@872
|
11 years |
sdipsl |
|
|
- introduce a random error mechanism
- correct a typo
|
|
|
@870
|
11 years |
sdipsl |
|
|
|
|
|
@869
|
11 years |
sdipsl |
|
|
- Enhance error message by providing the stack call tree when error has been detected
- No more fullstack file. Everything is in the stack file
- Log un encoded message to ease debuging
|
|
|
@868
|
11 years |
sdipsl |
|
|
- Initialize Big Brother mode from the main Job
- save message sequence in a specific file to reuse it later
- correct a typo
|
|
|
@864
|
11 years |
sdipsl |
|
|
- hindcast/forecast : can apply perturbation to any component. See #89
|
|
|
@863
|
11 years |
jgipsl |
|
|
Changed to have rebuild in a directory with apropriate name for curie. The old directory will continue to exist containg a link to the new one.
|
|
|
@862
|
11 years |
jgipsl |
|
|
Corrections for run at obelix/LSCE.
|