Changes between Version 3 and Version 4 of user/flavoni/libIGCM_v3.6/list_of_possible_errors


Ignore:
Timestamp:
2015-10-30T22:20:41+01:00 (5 years ago)
Author:
flavoni
Comment:

Legend:

Unmodified
Added
Removed
Modified
  • user/flavoni/libIGCM_v3.6/list_of_possible_errors

    v3 v4  
    55  rm Job_name/run.card;  rm Script_Output* rm everything created during the simulation in IGCM_OUT 
    66 
    7   but do not cancel run.card.'''    init'''     file, if not run.card file is not re-created. 
     7  but do not cancel run.card.'''      init'''       file, if not run.card file is not re-created. 
    88 
    9   ( File run.card.init is created with ins_job script )[[BR]]   [[BR]]   Submit the main job --- OR: cd EXPERIENCE[[BR]]   ../../../libIGCM/'''   clean_month.job''' 
     9  ( File run.card.init is created with ins_job script )[[BR]]     [[BR]]     Submit the main job --- OR: cd EXPERIENCE[[BR]]     ../../../libIGCM/'''     clean_month.job''' 
    1010 
    1111=== '''How to continue a simulation ''' === 
    12  
    13   Update config.card with the new !DateEnd   . Do not change !DateBegin   .'''    '''   Change !PeriodState    in run.card to !OnQueue    (or let clean_month job do it for you)'''    '''   Submit the main job 
     12  Update config.card with the new !DateEnd     . Do not change !DateBegin     .'''      '''     Change !PeriodState      in run.card to !OnQueue      (or let clean_month job do it for you)'''      '''     Submit the main job 
    1413 
    1514=== ''' How to continue a simulation that crashed previously''' === 
    16  
    1715  If you want to relaunch a crashed simulation you need first to remove all files produced for the crashed period. Use the script modipsl/libIGCM/clean_month.job and answer yes to delete files and change run.card Submit the main job 
    1816 
    1917=== '''!RebuildFromArchive=NONE''' === 
     18  In MPI it is good to use Rebuild in "archive" machine, to not ovrerload calcul machine. [[BR]]        You have to put in config.card in POST Session: 
    2019 
    21   In MPI it is good to use Rebuild in "archive" machine, to not ovrerload calcul machine. [[BR]]      You have to put in config.card in POST Session: 
     20  !RebuildFrequency        =1Y [[BR]]        !RebuildFromArchive        =true 
    2221 
    23   !RebuildFrequency      =1Y [[BR]]      !RebuildFromArchive      =true 
    24  
    25   NOTA BENE: If you put !RebuildFrequency      =1Y and you forget to set !RebuildFromArchive      =true, it is very dangerous. NEVER DO IT! 
     22  NOTA BENE: If you put !RebuildFrequency        =1Y and you forget to set !RebuildFromArchive        =true, it is very dangerous. NEVER DO IT! 
    2623 
    2724=== '''How to do a long job without exit from queue''' === 
     25  You can play with Duration Experience, PeriodLenght         and PeriodNb        , elapstim_req in this way :[[BR]]        in '''        config.card'''         Duration Experience is = !DateBegin         - !DateEnd         [[BR]]        in '''        config.card'''         !PeriodLenght         is how many time you do restart files[[BR]]        in '''        Job_Name'''         !PeriodNb         is  in '''        Job_Name'''         elapstim_req is elapsed time limit 
    2826 
    29   You can play with Duration Experience, PeriodLenght       and PeriodNb      , elapstim_req in this way :[[BR]]      in '''      config.card'''       Duration Experience is = !DateBegin       - !DateEnd       [[BR]]      in '''      config.card'''       !PeriodLenght       is how many time you do restart files[[BR]]      in '''      Job_Name'''       !PeriodNb       is  in '''      Job_Name'''       elapstim_req is elapsed time limit 
     27  Your job is out from queue every time that is finished a PeriodNb        , so you can do : 
    3028 
    31   Your job is out from queue every time that is finished a PeriodNb      , so you can do : 
    32  
    33   !PeriodNb       = ( Duration Experience ) / !PeriodLenght 
     29  !PeriodNb         = ( Duration Experience ) / !PeriodLenght 
    3430 
    3531  but be careful with elapstim_req, because this is a limit, you've to be sure that it exists a queue that can hold you.