search

PhasesResearchLab / ESPEI-datasets

Datasets to be used by ESPEI. CC-BY-4.0
Creative Commons Attribution 4.0 International
12 stars 17 forks source link

Work-stealing problem #15

Closed June813 closed 5 years ago

June813 commented 5 years ago

Hi Richard, I'm running the CU-MG example and have an issue about the work-stealing setting. After I finished the installation of Espei, I have already set the work-stealing value as false. I successfully generated the Cu-Mg_dft.tdb and no error was reported. When I move on to the MCMC optimization and try to generate the Cu-Mg_mcmc.tdb following the instruction, the value error is reported as "work-stealing is on in dask". But I checked the distributed.YAML, work-stealing status is still false. I don't know how to fix this, hope to hear from you soon. work-stealing value error

bocklund commented 5 years ago

Hey, @June813, in YAML files, any text after a # is a comment, so this file is actually just accepting the default value in dask, which is to set work stealing.

To change this file, delete the comment and first space so at least the following lines are un-commented:

distributed:
  version: 2
  scheduler:
    work-stealing: False
June813 commented 5 years ago

Hey, @June813, in YAML files, any text after a # is a comment, so this file is actually just accepting the default value in dask, which is to set work stealing.

To change this file, delete the comment and first space so at least the following lines are un-commented:

distributed:
  version: 2
  scheduler:
    work-stealing: False

Thank you! It works!

bocklund commented 5 years ago

Great, glad it works for you! FYI our Gitter chat, https://gitter.im/PhasesResearchLab/ESPEI, is another good place for these kind of questions.