Comparaison des versions

Légende

  • Ces lignes ont été ajoutées. Ce mot a été ajouté.
  • Ces lignes ont été supprimées. Ce mot a été supprimé.
  • La mise en forme a été modifiée.

The recipe below only works for simulations that were originally started with 'Chunk_lance'. If you need to restart a simulation originally started with 'Um_lance' contact me: winger.katja@uqam.ca

Sommaire

Restart

...

a simulation ...

... from the beginning

To restart a simulation that already started running from the beginning you need to add the key '-start':

    Chunk_lance -start

If you are starting a simulation for the very first time it is not necessary to add this key.
However, if your simulation already started but stopped in the executable of the first month and you want to make changes to the config files you need to restart the simulation from the beginning, so you need to add the key '-start'.

... from where it stopped

If the simulation was stopped (job was killed) or crashed it can often get re-submitted by executing again the command:

    Chunk_lance

... from previous restart files

Sometimes But sometimes it happens that part of the restart files gets overwritten before a month is finished or that the restart files are corrupted or one would simply like to rerun part of a simulation. In that case these cases one cannot just relaunch the month with Chunk_lance but has to get restart the original simulation from previous (uncorrupted) restart files back (from the month previous to the one one wants to rerun) and restart the simulation from there.When the model job stops for whatever reason, there is an automatic check in Chunk_lance if the restart files are still the original ones or not. If they got already modified the following message will appear in the "chunk_job listing" (!!!) not in the model listing:
   At least one of the restart files got already rewritten
   Therefore the model could not get restarted automatically
   You have to restart your simulation starting from the previous restart files
            ----- ABORT -----
 
If you see this message one cannot simply restart the simulation with Chunk_lance but as said, one need to restart it from the previous restart file - see below.

...

.

Note: In the description below replace all text in cursive/italic and all '...' with the actual names!

  1. Make sure the restart files of the previous month are in the execution directory :
           ~/MODEL_EXEC_RUN/${TRUE_HOST}
    If the are not there you need to put them back. You will find all previous restart files under ${CLIMAT_archdir}/Restarts. To "put them back" you need to:

    1. Go into the execution directory:
          cd ~/MODEL_EXEC_RUN/$TRUE_HOST
    2. Untar the restart file from which you want to restart your simulation. With something like:
          tar xvf ${CLIMAT_archdir}/Restarts/tar_file_for_previous_month.tarz

    Click on the following link to learn more about restart files.

  2. Go into the config file directory
  3. Edit the log file 'chunk_job.log':
    (First I suggest to make a backup copy of the log file. Just in case.)
    The Then remove all lines concerning the month you want to rerun and all following lines
  4. So the last line should contain something like:
        ... previous_month_M finished at ...
    Make sure the last line is not an empty line!

    Click on the following link to learn more about the file 'chunk_job.log'.
  5. Make sure the "Scripts-job",
  6. In your config file directory you also need the script called:
        ${GEM_exp%_
  7. exp
  8. *}_
  9. S, is in the config directory
    In case this file is not
  10. month-to-rerun_S
    If you do not have this script anymore in your config file directory
  11. anymore
  12. you can find it in
  13. your
  14. the archive
  15. under
  16. in the file:
  17.       
  18.     ${CLIMAT_archdir}/Listings/jobs_
  19. *
  20. ....zip
  21. Execute "Chunk_lance" again (without the '-start'!!!)