Environments¶

Environments are configuration layers that override the base config. A common use case is maintaining separate settings for development vs. production, but they can represent any named set of overrides (a testing setup, a resource-tuned configuration, a specific modeling suite branch, etc.).

For example, in development you might use a dev Docker image tag and lower resource allocations, while production uses a pinned version tag and higher resources. Environments let you express this cleanly.

Using environments¶

You select an environment per command using the --env flag. If you don't specify --env, it defaults to default.

$ epycloud --env dev <command>          # Use dev environment (default)
$ epycloud --env prod <command>         # Use production environment
$ epycloud --env prod config show       # Show resolved prod config

What belongs in an environment¶

Environment files only need to contain keys that differ from the base config. Everything else is inherited. Typical settings include:

• Docker image tag (docker.image_tag)
• Modeling suite branch (github.modeling_suite_ref)
• Resource allocations (google_cloud.batch.stage_b.cpu_milli, etc.)
• Parallelism limits (google_cloud.batch.max_parallelism)

For project-specific settings (experiment repo, storage prefix), consider using profiles instead.

Examples¶

A dev environment with lower resources and a development image:

# Build epymodelingsuite from dev branch and tag the image as "dev"
docker:
  image_tag: "dev"

github:
  modeling_suite_ref: "dev"

# Lower resources for development
google_cloud:
  batch:
    max_parallelism: 50
    stage_b:
      cpu_milli: 1000
      memory_mib: 4096

A prod environment with pinned versions and higher resources:

# Pin epymodelingsuite to a release tag
docker:
  image_tag: "v1.0.0"

github:
  modeling_suite_ref: "v1.0.0"

# Higher resources for production
google_cloud:
  batch:
    max_parallelism: 200
    stage_b:
      cpu_milli: 4000
      memory_mib: 16384

Usage¶

$ epycloud build cloud                              # Builds with dev env (default)
$ epycloud run workflow --exp-id test                # Runs with dev env

$ epycloud --env prod build cloud                   # Builds with prod env
$ epycloud --env prod run workflow --exp-id study    # Runs with prod env

$ epycloud --env prod config show                   # Inspect resolved config

For all available configuration keys, see the Configuration Variables Reference.