ci(nightly): post-deploy smoke test for /api/admin/import-status #531

New Issue

marcel · 2026-05-11T20:13:52+02:00

marcel commented

2026-05-11 20:13:52 +02:00

Type: CI / smoke test
Priority: P2-medium — production correctness of an admin-only endpoint that no current smoke step exercises
Source: review of #526 by Sara Holt (QA) — comment #8649, finding 2
Parent PR: #526 (mass-import bind mount)

Summary

Add a post-deploy smoke step to nightly.yml (and release.yml) that authenticates as the staging admin, hits GET /api/admin/import-status, asserts HTTP 200, and asserts state is one of {IDLE, RUNNING, DONE, FAILED}.

Context

#526 added structural regression guards (compose-config grep) for the /import bind mount, but didn't add a behavioural smoke check that the mass-import endpoint actually responds correctly after a fresh deploy. The current nightly.yml smoke step only verifies HSTS, Permissions-Policy, and /actuator blocking — nothing exercises an admin-authenticated endpoint.

A future regression that breaks admin auth, or somehow corrupts the MassImportService bean wiring, would not be caught until an operator manually clicks the import card.

Required

Extend the "Smoke test deployed environment" step in .gitea/workflows/nightly.yml with a check against /api/admin/import-status.
The check must:
- Authenticate as the staging admin (uses STAGING_APP_ADMIN_USERNAME / STAGING_APP_ADMIN_PASSWORD already available as secrets)
- Hit GET /api/admin/import-status
- Assert HTTP 200
- Assert response body has state in {IDLE, RUNNING, DONE, FAILED} (parsed via jq)
Failure must be a hard stop with a clear ::error:: annotation.

Notes

Unauthenticated /api/admin/* should return 401/403, not 200 — so the smoke test must include auth. Easiest: Basic Auth header (the SecurityConfig already supports it from #520).
This step lands after up -d --wait, which already guarantees backend health, so the auth + endpoint test is meaningful additional coverage rather than redundant.

Acceptance criteria

Step added to nightly.yml
On the next nightly: step passes and the run log shows the response body
Step is also added to release.yml for production (or a follow-up issue tracks that)
Failure produces a clear ::error:: annotation in the workflow log

Linked NFRs

Reliability: Every prod-reachable admin endpoint MUST have a post-deploy smoke check that authenticates and asserts a structured response.
Observability: Smoke failures MUST surface as actionable annotations, not silent step failures.

Definition of Ready

Endpoint + assertions specified
Auth mechanism + secret names identified
Failure semantics defined
Acceptance criteria testable

🤖 Generated with Claude Code during /implement on #526

**Type:** CI / smoke test **Priority:** P2-medium — production correctness of an admin-only endpoint that no current smoke step exercises **Source:** review of #526 by Sara Holt (QA) — comment [#8649](https://git.raddatz.cloud/marcel/familienarchiv/pulls/526#issuecomment-8649), finding 2 **Parent PR:** #526 (mass-import bind mount) ## Summary Add a post-deploy smoke step to `nightly.yml` (and `release.yml`) that authenticates as the staging admin, hits `GET /api/admin/import-status`, asserts HTTP 200, and asserts `state` is one of `{IDLE, RUNNING, DONE, FAILED}`. ## Context #526 added structural regression guards (compose-config grep) for the `/import` bind mount, but didn't add a behavioural smoke check that the mass-import endpoint actually responds correctly after a fresh deploy. The current `nightly.yml` smoke step only verifies HSTS, Permissions-Policy, and `/actuator` blocking — nothing exercises an admin-authenticated endpoint. A future regression that breaks admin auth, or somehow corrupts the `MassImportService` bean wiring, would not be caught until an operator manually clicks the import card. ## Required 1. Extend the "Smoke test deployed environment" step in `.gitea/workflows/nightly.yml` with a check against `/api/admin/import-status`. 2. The check must: - Authenticate as the staging admin (uses `STAGING_APP_ADMIN_USERNAME` / `STAGING_APP_ADMIN_PASSWORD` already available as secrets) - Hit `GET /api/admin/import-status` - Assert HTTP 200 - Assert response body has `state` in `{IDLE, RUNNING, DONE, FAILED}` (parsed via `jq`) 3. Failure must be a hard stop with a clear `::error::` annotation. ## Notes - Unauthenticated `/api/admin/*` should return 401/403, not 200 — so the smoke test must include auth. Easiest: Basic Auth header (the SecurityConfig already supports it from #520). - This step lands *after* `up -d --wait`, which already guarantees backend health, so the auth + endpoint test is meaningful additional coverage rather than redundant. ## Acceptance criteria - [ ] Step added to `nightly.yml` - [ ] On the next nightly: step passes and the run log shows the response body - [ ] Step is also added to `release.yml` for production (or a follow-up issue tracks that) - [ ] Failure produces a clear `::error::` annotation in the workflow log ## Linked NFRs - **Reliability:** Every prod-reachable admin endpoint MUST have a post-deploy smoke check that authenticates and asserts a structured response. - **Observability:** Smoke failures MUST surface as actionable annotations, not silent step failures. ## Definition of Ready - [x] Endpoint + assertions specified - [x] Auth mechanism + secret names identified - [x] Failure semantics defined - [x] Acceptance criteria testable 🤖 Generated with [Claude Code](https://claude.com/claude-code) during /implement on #526

marcel referenced this issue

2026-05-11 20:14:42 +02:00

feat(infra): bind-mount /import for backend mass-import endpoint #526

marcel added the P2-medium devops test labels 2026-05-11 20:36:41 +02:00

marcel closed this issue

2026-05-14 11:53:21 +02:00

Sign in to join this conversation.

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: marcel/familienarchiv#531