mirror of
https://github.com/dredx/prole.git
synced 2026-09-27 16:54:30 +00:00
b905cbe584
- Add centralized config resolver/activator with per-environment layering and stable symlink entrypoint\n- Bootstrap missing env dirs and preserve non-symlink legacy prole.cfg by seeding into inferred env\n- Wire installer UI/backend + shell helpers to shared resolution path for explicit, safe env switching\n- Harden k3d: registry network/DNS wiring and ArgoCD repo-server hostPath permission init\n- Add docs + regression tests for env switching, namespace stability, k3d registry, and ArgoCD rollout Co-authored-by: Junie <junie@jetbrains.com>
45 lines
2.1 KiB
Markdown
45 lines
2.1 KiB
Markdown
### Prole configuration environments
|
||
|
||
Prole treats `$PROLE_CONF` as an explicit, switchable set of isolated environment configs.
|
||
|
||
#### Directory layout
|
||
|
||
Under `$PROLE_CONF`:
|
||
|
||
```text
|
||
$PROLE_CONF/prole.cfg -> symlink entrypoint (active env)
|
||
$PROLE_CONF/dev/prole.cfg -> base config (dev)
|
||
$PROLE_CONF/service/prole.cfg -> base config (service)
|
||
$PROLE_CONF/prod/prole.cfg -> base config (prod)
|
||
$PROLE_CONF/test/prole.cfg -> base config (test)
|
||
|
||
# Optional overrides (conf.d-style) per environment:
|
||
$PROLE_CONF/<env>/*.cfg -> applied as layered overrides
|
||
```
|
||
|
||
Existing code continues to read `.../prole.cfg` via the single stable entrypoint path: `$PROLE_CONF/prole.cfg`.
|
||
|
||
#### Activation (selecting the active environment)
|
||
|
||
- The active environment is determined by what `$PROLE_CONF/prole.cfg` points to.
|
||
- The Cluster Environment screen activates the selected env by updating the symlink (via `installer/prole_conf.py`).
|
||
- Switching environments only changes the symlink target; it does not overwrite any other environment’s curated files.
|
||
|
||
If the entrypoint is a regular file (legacy layout), activation preserves it as `prole.cfg.legacy.<timestamp>` and switches to the symlink-based layout.
|
||
|
||
#### Override layering (deterministic merge)
|
||
|
||
For a selected environment directory `$PROLE_CONF/<env>/`:
|
||
|
||
1. Load `$PROLE_CONF/<env>/prole.cfg` (base)
|
||
2. Load every other `*.cfg` file in the same directory, in lexicographic filename order
|
||
|
||
Later files override earlier values (last write wins). This makes overrides predictable and reviewable.
|
||
|
||
#### Why this prevents accidental namespace drift
|
||
|
||
- Tests can target an isolated environment (`test`) with its own base config and overrides.
|
||
- Environment switching is explicit (symlink target changes) and stable (entrypoint path stays the same).
|
||
- Updates that need to modify config write through the entrypoint symlink to the active environment’s base file, avoiding hidden cross-environment mutations.
|
||
- The `test` environment defaults to a non-`default` namespace (`prole-test`), reducing the chance that unit/integration runs revert to Kubernetes `default`.
|