ci(guide): widen cd-develop smoke health-check window to ~5 min (#12) #13

2026-07-08T17:12:15-04:00

hiveiq commented

2026-07-08 17:12:15 -04:00

Widens the cd-develop smoke health-check window from 60 s to ~5 min so a normal Spring Boot restart after deploy isn't marked a failure.

Why

Run #6 (PR #11) deployed correctly — guide.bcos.dev served the new content and /guide/actuator/health returned 200 UP — but was flagged failure because the smoke step (12 × 5 s) expired before the freshly-restarted backend warmed up. Same false-failure class as #9 (Slack notify), one step earlier.

Change

seq 1 12 / sleep 5 → seq 1 30 / sleep 10 (30 attempts, ~5 min).
Echo each attempt's status for observability.
Deploy behaviour unchanged.

Note

This PR touches .gitea/workflows/cd-develop.yml, which is in the workflow's own paths trigger — merging it re-runs cd-develop, which serves as a live test of the wider window (expected: green).

Closes #12.

Widens the `cd-develop` smoke health-check window from **60 s** to **~5 min** so a normal Spring Boot restart after deploy isn't marked a failure. ### Why Run #6 (PR #11) deployed correctly — guide.bcos.dev served the new content and `/guide/actuator/health` returned `200 UP` — but was flagged **failure** because the smoke step (12 × 5 s) expired before the freshly-restarted backend warmed up. Same false-failure class as #9 (Slack notify), one step earlier. ### Change - `seq 1 12` / `sleep 5` → `seq 1 30` / `sleep 10` (30 attempts, ~5 min). - Echo each attempt's status for observability. - Deploy behaviour unchanged. ### Note This PR touches `.gitea/workflows/cd-develop.yml`, which is in the workflow's own `paths` trigger — **merging it re-runs `cd-develop`**, which serves as a live test of the wider window (expected: green). Closes #12.

hiveiq added 1 commit 2026-07-08 17:12:16 -04:00

ci(guide): widen cd-develop smoke health-check window to ~5 min e948180cf5

The smoke step polled /guide/actuator/health for only 60s, but the Spring
Boot backend restart after `compose up -d` needs longer to warm up — so a
successful deploy was marked FAILED (run #6, PR #11). Poll 30x10s (~5 min)
and echo attempts. No change to the deploy itself.

Closes #12

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>