You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Model log files can take up a significant amount of storage. For a single year of ESM1.5 simulation, the sub-models produce:
Atmosphere: ~50M
Ocean: ~60M
Ice: ~900M
which can add up over long simulations.
Previous discussions raised the idea of compressing the log files during the archive step. @aidanheerdegen mentioned a good place to start would be the CICE driver, and if successful it could be expanded to a general (perhaps optional) step across models.
A couple of implementation questions/concerns:
Would it be preferable to compress each log file separately, or combine all the logs for a single sub-model into a tar.gz?
People may have workflows set up for extracting information from the logs (e.g. subroutine timing information) which would be affected.
The text was updated successfully, but these errors were encountered:
So I think a compressed tar archive makes most sense.
The question of what to archive? Probably best to have a list in the model driver of patterns to match files. Simple case would be just anything that beginswith a list of strings.
Model log files can take up a significant amount of storage. For a single year of ESM1.5 simulation, the sub-models produce:
which can add up over long simulations.
Previous discussions raised the idea of compressing the log files during the
archive
step. @aidanheerdegen mentioned a good place to start would be the CICE driver, and if successful it could be expanded to a general (perhaps optional) step across models.A couple of implementation questions/concerns:
tar.gz
?The text was updated successfully, but these errors were encountered: