Configuration Overview

Volundr loads configuration from multiple sources. Higher priority sources override lower ones.

Priority order (highest to lowest):

1. Constructor arguments — testing only
2. Environment variables — double underscore nesting: DATABASE__HOST=postgres.local
3. YAML config file — ./config.yaml or /etc/volundr/config.yaml
4. /run/secrets files — Kubernetes secrets mounted as files

Two Config Contexts

There are two ways to configure Volundr depending on how you run it.

Config file (config.yaml / config.yaml.example) is for running from source or local CLI. Uses snake_case throughout.

Helm values (values.yaml) is for Kubernetes deployments. The chart generates a ConfigMap mounted at /etc/volundr/config.yaml. Uses camelCase.

The naming differs between them. Helm uses camelCase (podManager), config.yaml uses snake_case (pod_manager). The chart templates handle the translation.

Dynamic Adapter Pattern

Most infrastructure components are configured by specifying a fully-qualified Python class path and constructor kwargs. You can swap backends without code changes — just update the config.

# Example: configure pod manager
pod_manager:
  adapter: "volundr.adapters.outbound.flux.FluxPodManager"
  kwargs:
    namespace: "default"
    chart_name: "skuld"

Adding a new adapter = write the class + update config. No code changes in the container.

See Dynamic Adapters for full details on how the pattern works.