Skip to content

Namespaces and env_id

podCI isolates caches and artifacts by deriving a deterministic namespace and env_id.

Why it exists

Without namespacing, caches become shared global state: - builds “randomly” become fast/slow, - stale outputs leak across unrelated repos, - safe prune becomes impossible.

Behavior

  • namespace identifies the logical project space.
  • env_id identifies an execution environment derived from stable inputs (profile + toolchain/image identity).

The derivation mechanism is intentionally strict. Do not hand-edit derived identifiers.

Operational impact

  • Caches are safe to prune by namespace without risking other projects.
  • Build/test results are more predictable across machines.

Inputs that affect env_id

env_id is derived from stable configuration inputs. Changing any of these will intentionally produce a new env_id (and therefore new cache volume names):

  • profile.container
  • profile environment (profiles.<name>.env)
  • step ordering and step argv
  • step workdir
  • step environment (jobs.<job>.steps.<step>.env)