search

geoschem / GCHP

The "superproject" wrapper repository for GCHP, the high-performance instance of the GEOS-Chem chemical-transport model.
https://gchp.readthedocs.io
Other
21 stars 25 forks source link

[DISCUSSION] Default content of GCHP logs files #238

Closed lizziel closed 3 months ago

lizziel commented 2 years ago

This discussion is about what to include in GCHP log files. This includes the output of pFlogger that is configured in logging.yml as well as the main GCHP log file. As of the 14.0 release logging.yml is set to use level WARNING for root-level logging of MAPL and CAP.EXTDATA, and to use level INFO for CAP, GCHP, and GCHPctmEnv. The main log is set to include all GEOS-Chem standard output with debug prints off, and HEMCO standard output with WARNING set to 1 and VERBOSE set to 0. There is no HEMCO.log file in GCHP 14.0. These default settings are not consistent with each other as some include informational information and some only warnings.

Regarding logging.ymlsettings: The logging output of this file is sent to the file set in logging.yml (allPEs.txt in 14.0). Currently the WARNING level is not really used in GCHP, or at least I haven't encountered it yet. I have not done a global search looking for it. The INFO level is used but is suppressed by default for MAPL and CAP.EXTDATA. We used to have the MAPL level set to INFO for root but changed it due to excessive writing to the file (https://github.com/geoschem/geos-chem/pull/1262).

Question: should we adjust how INFO is used so that when enabled it minimizes file size but maximizes usefulness, and then make the INFO level on by default? This will require some code work.

Regarding the main log file: In general the file should be inspected to see if any of it could be slimmed down, or if there are HEMCO verbose prints that would be useful to have on.

Question: Do you think the current standard output log of GCHP too large? If yes, what do you think is not needed? What do you think it could/should have that it does not?

General: We need to balance having a useful log file in the event of a run problem with having reasonably sized log files. This will be an ongoing process that even if we settle upon now would need to be revisited periodically as new prints make their way into the model.

lizziel commented 3 months ago

I am closing this discussion as we anticipate the logging for MAPL to be cleaned up in MAPL3, to be included in GCHP 15.0. We also did work in the last few versions to reduce the HEMCO and GEOS-Chem output to log.