Environment variable interpolation

Only configuration properties under the config section that accept string values can be populated from environment variables at startup. This allows deployment-specific configuration, including credentials from Kubernetes Secrets, to be supplied without being hardcoded in the Helm chart.

An environment variable reference is declared as ${NAME}. The reference must occupy the complete property value; partial variable substitution is not supported. For example, bucket-${STORAGE_SERVICE_IDX}-id is not interpolated.

For example, the following configuration declares a Redis client, imports its address from a Kubernetes Secret, and supplies the address through an environment variable reference:

extraEnvs:
  - name: STORAGE_SERVICE_REDIS_ADDRESS
    valueFrom:
      secretKeyRef:
        name: storage-service-redis
        key: address
config:
  redis:
    enabled: true
    address: "${STORAGE_SERVICE_REDIS_ADDRESS}"

For example, the following configuration is invalid:

config:
  storage:
    s3:
      endpoints:
        "${STORAGE_SERVICE_S3_BUCKET}": # See the explanation below.
          anonymous: true
  statCache:
    enabled: "${STORAGE_SERVICE_STAT_CACHE_ENABLED}" # Only string configuration properties support interpolation.
  listCache:
    capacity: "${STORAGE_SERVICE_LIST_CACHE_CAPACITY}MiB" # Partial variable substitution is not supported.

Known issues

S3 bucket endpoints and Azure Blob storage account endpoints cannot be replaced with environment variable placeholders. Their names must be specified verbatim.