source: trunk/libIGCM

Revision Log Mode:


Legend:

Added
Modified
Copied or renamed
Diff Rev Age Author
(edit) @139   15 years sdipsl

Bypass a little bug filling ${FileToBeDeleted?}. Parameter files may be recorded twice. No big deal.

(edit) @138   15 years sdipsl

SD+MAF : prepare .resol to contain more information

(edit) @137   15 years mmaipsl

Add ksh time command for SX.
Add some debug informations.

(edit) @136   15 years sdipsl
  • MPISUSPEND is now OFF by default
  • MPISUSPEND, F_FILEINF, NC_FILEINF can be redefined inside Job or inside comp.driver
(edit) @135   15 years sdipsl
  • Reflect the fact that monitoring should not depend on resolution for atmosphere and surface.
  • Some cleanup
(edit) @134   15 years sdipsl
  • Do not use absolute path with monitoring01 as input_directory when producing files and images
  • More verbosity in monitoring.job
  • Bypass strange rcommand behaviour with a sleep for ulam
(edit) @133   15 years sdipsl

Reflect again the fact that Ulam can see Gaya filesystem

(edit) @132   15 years sdipsl

Minimize rsh from vargas to others machine
Reflect the fact that Ulam can see Gaya filesystem
Manage AIX vargas system time soi that run.card is filled properly.

(edit) @131   15 years mmaipsl

Change default resolution for SRF to ALL and add STOmate component (for MONITORING).
Correct a little mistake in two loops in case of the compname equal to history filename.

(edit) @130   15 years sdipsl

MAF, SD ; Process mfget file by file to avoid trouble. IDRIS works on this issue.

(edit) @129   15 years mmaipsl

Little mistake with JobType? and a test.

(edit) @128   15 years mmaipsl

Add experience type management : JobType? = DEB(ug), DEV(elopment), RUN (default)
(No propoagation in Post-treatment for instance).
Put "Where do we run ?" question in libIGCM_sys.

(edit) @127   15 years mmaipsl

Add section,option informations with error.

(edit) @126   15 years sdipsl

Bugfix : useless sed with config_Post_SeasonalFrequencyOffset

(edit) @125   15 years sdipsl

Bug fix : a bad test leading to issue in post-processing when SeasonalFrequencyOffset? is not set in config.card

(edit) @124   15 years mmaipsl

MM: add another IGCM_config_Analyse_PeriodLength call and suppress arguments in this function.

(edit) @123   15 years brocksce

PB: - Add option --time to monitoring01 to create MONITORING/files

with heritage of the time axis of the TS_MO files.
Assume that those time axis are correct (eventually they have been corrected by patches).

(edit) @122   15 years mmaipsl

Change PeriodNB for real PeriodNb? used in job.

(edit) @121   15 years mmaipsl

Add "uname -a" to know where we have run.
Never use IGCM_post_Initialize now, always IGCM_post_Configure.

(edit) @120   15 years mmaipsl

Change main loop syntaxe to be abble to change PeriodNb? with libIGCM.card

(edit) @119   15 years mmaipsl

Modification of libIGCM behaviour :
To use this function, one must copu libIGCM.card from ${libIGCM} directory
and put it in ${SUBMIT_DIR} directory. Ater modifications of ${SUBMIT_DIR}/libIGCM.card,
variables define inside [UserChanges?] will be modifyd for next Period of libIGCM main loop.

Be carefull : it can give problems with post-treatment !

(edit) @118   15 years mmaipsl

Replace the two big analyses of PeriodLength? by a new function IGCM_config_Analyse_PeriodLength.
Add PID at the end of Script_Output saved files.

(edit) @117   15 years mmaipsl

Update default system file to last libIGCM developments.

(edit) @116   15 years sdipsl

New option that goes together with SeasonalFrequency? : SeasonalFrequencyOffset?.
Let you compute seasonal average with an offset. So you can skip spinup period
and average only when stabilization is good

SeasonalFrequencyOffset? must be a scalar (no letter), same unit as SeasonalFrequency?
Need documentation ...

(edit) @115   15 years sdipsl

Bugfix : In case dmget ends badly do not stop switch. User may want a field time series that does not exist in file ...

DMFDIR now available in SX9 environment

(edit) @114   15 years mmaipsl

Correction of initialization of RunDateBegin? and RunDateEnd? to realistic and protected values.

(edit) @113   15 years sdipsl

Bugfix : try to save stack file only in debug mode ; stack file exists only in debug mode.

(edit) @112   15 years brocksce

PB: - Add initialisation for RUN_DATE_BEGIN and RUN_DATE_END

(useful if AA_job is an old one ie. without RUN_DATE_BEGIN and
RUN_DATE_END sent to the ${Exe_Output})

(edit) @111   15 years sdipsl

gawk specific path on SX9 and SX8-R

(edit) @110   15 years sdipsl

Back to the original method to detect directory that needs rebuild
Keep TMP files for safety and also testing purpose

(edit) @109   15 years mmaipsl

Correct other awk to gawk replacements.

(edit) @108   15 years sdipsl

May be last touch before tag 1.3

  • some cosmetics
  • prepare different rebuild policy
    • from Archive
    • from Workdir
  • for now only "from Archive" is available otherwise inside computing job as before
(edit) @107   15 years sdipsl

Bugfix : Forgot one dot before monitoring01 when using --listcommand 'IGCM_sys_RshArchive ls'

(edit) @106   15 years sdipsl

Double security for sys_Put_Out on ulam.
Sometimes mfput fails telling that :
REBUILD.2846:mfput : getpwuid : No such file or directory

IDRIS issue related to ftuas handling. They know about that and try to fix it.

(edit) @105   15 years sdipsl

MAF : Add daily frequency possibility for asynchronous rebuild.

(edit) @104   15 years sdipsl

PB: - Put a dot before monitoring01 when using --listcommand 'IGCM_sys_RshArchive ls'
because monitoring01 must be run as a user command file (not as a subshell) and thus
inherits of all libIGCM functions

(edit) @103   15 years mmaipsl

Enhancement for calendars.
Before we had 360d, noleap and gregorian (or leap) calendars.
There was no bug.
Now we add "all_leap" (for 366 days in year).

(edit) @102   15 years sdipsl

Add verbosity to this job. It will facilitate debugging and developping.

(edit) @101   15 years sdipsl

Better that way, less dependant to personnal environment.

(edit) @100   15 years sdipsl

Bugfix when printing UserChoices? from component card

(edit) @99   15 years sdipsl

Add SX9 to libIGCM. Work fine.

(edit) @98   15 years sdipsl
  • Try to avoid having stack files every where during post-processing job.
(edit) @97   15 years sdipsl

Remove a useless job

(edit) @96   15 years sdipsl

Use gawk instead of awk. Because gawk have the same behaviour on every machine, awk not.

(edit) @95   15 years sdipsl

Forgot to clean up in the ORCA2_LIM2_PISCES case.

(edit) @94   15 years sdipsl

Bugfix ulam : monitoring job must know whereis libIGCM on the computing machine.

(edit) @93   15 years sdipsl
  • Add ZNL atlas for ORCA2_LIM2 and ORCA2_LIM2_PISCES
  • Bugfix for vargas IGCM_sys_Qsub and IGCM_sys_QsubPost function
(edit) @92   15 years sdipsl

MAF :
On the way to finalyze vargas

  • adjustment to support large numbers of cpus (and files)
  • prepare another policy for rebuild : from workdir

Some cleanup

(edit) @91   15 years mmaipsl

Copy of the IGCM_Patch_20070220_histcom_time_axis.ksh, but for "double" time axis.

(edit) @90   15 years brocksce

PB: - Modification according to new option (--listcommand) in ATLAS/monitoring01

to allow remote listing through call to 'IGCM_sys_RshArchive ls'
since ulam cannot list archive file system directly.

(edit) @89   15 years sdipsl

Bug fix vargas with LoadLeveler? option

(edit) @88   15 years sdipsl

Bugfix with create_ts_begin_date and create_ts_next_date function du to ulam's ksh.
Do not put anymore typeset on a variable passed to a function that will modify it.
This bugs breaks leap and noleap calendar for post-processing on ulam.
Thanks to NEMO for reporting this nasty bug

(edit) @87   15 years sdipsl

NEMO config added to turn on atlas submission.
Do not use so called additionnal for NEMO config only "classic" atlas.

(edit) @86   15 years sdipsl

Switch back to mfget. IDRIS replace old mfget by tmfget and remove tmfget.

(edit) @85   15 years sdipsl

Bugfix : typo

(edit) @84   15 years sdipsl

Use tmfget instead of mfget. We will go back to mfget when IDRIS put tmfget as standard command.

(edit) @83   15 years sdipsl

Bugfix with aix6 headers

(edit) @82   15 years sdipsl

Bugfix with rebuild activation.

(edit) @81   15 years sdipsl
  • AA_create_ts' homogenization regarding array handling in AA_create_se
(edit) @80   15 years sdipsl

Explicit all MASTER name.
for exemple you can not rcp vargas92, but vargas you can

(edit) @79   15 years mmaipsl

MM: change the comment

(edit) @78   15 years sdipsl
  • Correct tyÃpo error in aix6 headers
(edit) @77   15 years sdipsl
  • Use MASTER=vargas in libIGCM_sys_vargas.ksh
  • Adapt IGCM_sys_Get_Master command for ulam
(edit) @76   15 years mmaipsl

MM: Add a patch to change _Fillvalue to missing_value attribute for some old compatibily.

(edit) @74   15 years sdipsl

AC : new atlas for diaptr (streamfucntion etc ...)

remove so called additional atlas
add NUM_PROC_OCE in libIGCM_sys_platine.ksh

(edit) @73   15 years sdipsl

Bugfix : ulam WORKDIR is not defined through an rsh from vargas --> tweak the ulam ${HOME} path

(edit) @72   15 years sdipsl

Export "MASTER" variable to post-processing job so they know where to get information

(edit) @71   15 years sdipsl
  • Add "MASTER" variables to the machine able to run models.
  • Clean up libIGCM_POST definition
  • Handle the fact that ulam can post-process for brodie or vargas (all having their own filesystem)
(edit) @70   15 years sdipsl

Add vargas

(edit) @69   15 years sdipsl

Remove libIGCM_sys_zahir.ksh

(edit) @68   15 years sdipsl
  • Add headers for vargas
  • Deleted svn property svn:executable when needed (chmod -x)
(edit) @67   15 years sdipsl

libIGCM is ready for ulam except concerning the monitoring

  • fix issue due to mfls (chaotic behaviour) do not use mfls anymore
  • change due to the fact that ulam do not see gaya : imply some libsys homogeneisation
  • atlas/monitoring01 needs an adaptation to ulam (because of "ls $homegaya" like command)
(edit) @66   15 years sdipsl
  • Adaptation to ulam. Need to handle array in a different way
(edit) @65   15 years sdipsl

Source ferret atlas/fast from common account. With database, climato files ..

(edit) @64   15 years sdipsl
  • Bug fix. ksh93 is very very strict
(edit) @63   15 years sdipsl
(edit) @62   15 years sdipsl
  • Adapt to the strange behaviour of ulam's ksh93
  • UnixMaxLimit? now depends on libIGCM_sys
(edit) @61   15 years sdipsl
  • Add UnixMaxLimit? variables inside libIGCM_sys
  • Don't use module with ulam for the moment
  • Adapt llsubmit procedure to have post-process job output when we want to
(edit) @60   15 years sdipsl
  • Add meaningfull JobName? for ulam
  • Heritate only usefull environments variables through LoadLeveler? option
(edit) @59   15 years sdipsl
  • change headers to force ksh in the shell
  • adapt rsync to ulam
  • full path for llsubmit
  • reflect the fact that ulam does not see gaya file system
  • Asynchronous rebuild OK, classic post-treatment not
(edit) @58   15 years sdipsl

Headers for LoadLeveler? 3.5

(edit) @57   15 years brocksce

PB: - Reformat run.card

  • Add monitoring card analysis feature
  • Rename Script_Output
  • Remove obsolete comments in run.card.init
(edit) @56   15 years sdipsl

Add module load for netcdf, nco and cdo

(edit) @55   15 years sdipsl

First draft for ulam

(edit) @54   15 years sdipsl

A few bug fixes and adaptation for ulam so that libIGCM auto-check
succeed on this machine.

(edit) @53   15 years sdipsl

Remove sys_Get LEVITUS_1m_Temperature_Pot_Ice_nomask.nc .

sys_Get LEVITUS_1m_Salinity_Ice_nomask.nc .

Those files must be found in $FER_DATA to have ocean additionnal diagnostic for CM5

(edit) @52   15 years sdipsl

Some tweak to avoid shell error message. Just cosmetic

(edit) @51   15 years sdipsl

SD+MAF Bugfix
one missing character break the libIGCM all calendar feature.

(edit) @50   15 years sdipsl

BugFix? 2
When simulation finished and asynchronous rebuild is on : submit rebuild anyway
Also when simulation finished and time series are on : submit create_ts anyway

(edit) @49   15 years sdipsl

Bug fix
When simulation finished do not produce time series if user do no want to ...
TimeSeriesFrequency?=NONE or
TimeSeriesFrequency?=

Note: See TracRevisionLog for help on using the revision log.