Files
HealthLog/docs/ops/env-check.md
T
Marc-André Bombeck c986efab30 Correct operator runbooks against the current code
- attempt-number-semantics: name the real bucket column
  (consecutiveFailuresByKind) and drop a drifted line citation
- backup-restore: show the smoke-test response inside the standard
  { data, error } envelope
- deploy: the manual migration step now uses the bundled Prisma CLI
  via node (the image strips pnpm); WHOOP vars are already on the
  shipped compose whitelist
- encryption-key-rotation: point at the encrypted-column registry
  instead of a five-model list several releases behind
- env-check: the CI gate shipped (env-check.yml); the partial-APNs
  example now shows the synthetic all-or-none row and the non-zero exit
- password-reset: the admin API reset route exists; only the
  self-service route does not
- tls-cert-pin: stop asserting client-repo pin-set details that cannot
  be audited from this repository
- drop v141-followup-issues.md: a v1.4.1 punch list whose items are
  fixed, removed, or superseded; keeping it under docs/ops invites
  re-fixing shipped work
2026-08-27 12:18:34 +02:00

107 lines
4.0 KiB
Markdown

# Pre-deploy env-var sanity check
`pnpm check-env` reads `scripts/env-manifest.json` and verifies every
variable declared `required: true` is present + non-empty in the
runtime environment. Optional groups are surfaced too, so an operator
can spot the "3 out of 4 vars set" pattern that silently disables a
feature.
This is the answer to the v1.4.40 AP-2 silent-disable: the `.p8` file
went missing during a Coolify migration, three of four `APNS_*` vars
remained set, the app booted cleanly, and APNs notifications stopped
working — undetected for three days because the dispatcher's fallback
chain hid the gap behind the Telegram → ntfy → Web Push path.
## Usage
### Against the live shell
```sh
pnpm check-env
```
Runs against `process.env` (after the local `.env` / `.env.local` load,
if any). Exit code 0 on green, 1 on at least one required variable
missing, 2 on a malformed manifest.
### Against a Coolify export / .env.production
```sh
pnpm check-env --file /path/to/.env.production
```
The file is parsed locally (no shell execution, no variable expansion)
and checked against the manifest. Useful for spot-checking a Coolify
export without polluting the local shell.
## Output format
Output is one row per declared variable, grouped by section. Three
states:
| Marker | Meaning |
| -------------------- | -------------------------------------------------------------- |
| `[OK]` | Variable is set and non-empty. |
| `[MISSING-REQUIRED]` | Variable is required for boot. The script exits 1. |
| `[missing-optional]` | Variable is optional. The script reports it but does not fail. |
The grep-friendly format lets a CI pipeline do `grep -q
'\[MISSING-REQUIRED\]'` to gate a deploy. CI already does: the
`env-check.yml` workflow runs `pnpm check-env --file
.env.production.example` on every PR and fails on missing required
vars.
## All-or-none groups
Some groups (e.g. off-host backups) are marked `allOrNone: true` in
the manifest. When such a group is partially populated — some vars
set, some missing — the script emits a synthetic `<all-or-none>` row
with severity REQUIRED, even if the surrounding group is `required:
false`. This is the v1.4.40 AP-2 detection pattern.
Example output for a partial APNs config:
```
# APNs (iOS push)
[OK] APNS_KEY_ID
[OK] APNS_TEAM_ID
[OK] APNS_BUNDLE_ID
[missing-optional] APNS_KEY — Satisfied by any of: APNS_KEY, APNS_KEY_FILE
[MISSING-REQUIRED] <all-or-none> — Group is all-or-none but only 3/4 variables are set. Either set all or unset all.
```
The APNs group is `allOrNone: true`, so the partial config also emits
the synthetic `<all-or-none>` row at REQUIRED severity and the run
exits 1. A reviewer immediately sees that `APNS_KEY` (or its
`APNS_KEY_FILE` alternative) is missing while the other three APNs
vars are set — the exact configuration that silently disables iOS
push.
## Editing the manifest
The manifest lives at `scripts/env-manifest.json` and is git-tracked
so changes go through a PR. To add a new variable:
1. Pick the right group (or add a new one if the variable doesn't
belong to any existing category).
2. Decide whether the variable is required for boot, optional, or
part of an all-or-none feature group.
3. Document the variable's `purpose` in one sentence — the operator
reads this text when the var shows up as missing.
4. Run `pnpm check-env` locally against your live shell to confirm
the manifest still passes (or fails for the right reason).
5. Add a regression test under `scripts/__tests__/check-env.test.ts`
if the manifest change introduces a new classification branch
(anyOf, allOrNone, …).
## CI gate
The GitHub Actions workflow `.github/workflows/env-check.yml` runs
`pnpm check-env --file .env.production.example` on every PR, blocking
changes that introduce new required vars without updating the example
file.
Still an idea only: a Coolify pre-deploy hook that runs
`pnpm check-env` inside the target container and aborts the deploy on
exit code 1.