good day! in dagster documentation there are a couple of examples, that use yaml configuration, like:

configuration.yaml

loggers:
  console:
    config:
      log_level: ERROR

i’ ve found the only way to use this configuration in Python program: 1. load dictionary from yaml

import yaml
with open('configuration.yaml'. 'r\) as f:
    log_config = yaml.load(f)

2. pass this log_config into execute_pipeline function

import dagster as d

d.execute_pipeline(pipeline, run_config=log_config)

but this way is not suitable - way better to use repositories. not using yaml configuration generates new problems - in my example, my custom logger won’t overwrite default dagster colored_logger and a won’t reach expected effect. so, the question is - how to load yaml configuration and use it when using repositories? thanks in advance!

With the current APIs, you can do this using presets. You can use the `.from_files`(link) method to create a preset from yaml, and pass that to a pipeline. This will make it available from every execution entrypoint.

thank you, Chris! i will try to implement this into code. new api with graph’s and op’s is going to be OP :)

feel free to let us know if any other questions come up !

@chris, i would like to clarify one more question about presets - is there a way to use custom PresetDefinition by default in schedule. I’m using my custom logger, but it works only when i’m running pipeline in playground, but in schedule runs it doesn’t work. Thanks in advance!

We don't have a way to use a preset in a schedule directly - the pattern you'll see a lot is pulling the run config off of their preset and returning it from the schedule:

@schedule(...)
def my_schedule():
    return my_preset.run_config