fix(ci): e2e-api health-wait 30s→300s + migration completion gate (resolves #2205) #2207

Merged
claude-ceo-assistant merged 1 commits from fix/2205-e2e-api-health-wait-migration-gate into main 2026-06-04 11:57:15 +00:00
+19 -5
View File
@@ -327,7 +327,12 @@ jobs:
# start-redis steps point at this run's per-run host ports.
./platform-server > platform.log 2>&1 &
echo $! > platform.pid
- name: Wait for /health
- name: Wait for /health (with migration completion gate)
# Issue #2205: 30 one-second probes is insufficient when the migration
# chain is still running; /health can flip true before migrations
# finish, so subsequent steps that touch the DB fail. Hybrid fix:
# bump timeout to 300s AND gate exit on the same workspaces-table
# existence check the downstream "Assert migrations applied" uses.
if: needs.detect-changes.outputs.api == 'true'
run: |
# Readiness signal: the platform binds /health only AFTER the full
@@ -343,13 +348,21 @@ jobs:
# background platform-server process has exited (e.g. a broken
# migration crashed it), we stop and fail loudly at once instead of
# waiting out the whole budget.
DEADLINE_SECS=180 # cold-start + full migration chain headroom
#
# Issue #2205: /health can flip true before migrations finish on a
# growing chain, so we gate exit on the workspaces-table existence
# check the downstream "Assert migrations applied" uses.
DEADLINE_SECS=300 # cold-start + full migration chain headroom
PLATFORM_PID="$(cat workspace-server/platform.pid 2>/dev/null || true)"
start=$(date +%s)
while :; do
if curl -sf "$BASE/health" > /dev/null; then
echo "Platform healthy after $(( $(date +%s) - start ))s"
exit 0
tables=$(docker exec "$PG_CONTAINER" psql -U dev -d molecule -tAc \
"SELECT count(*) FROM information_schema.tables WHERE table_schema='public' AND table_name='workspaces'" 2>/dev/null || echo "0")
if [ "$tables" = "1" ]; then
echo "Platform healthy + migrations applied after $(( $(date +%s) - start ))s"
exit 0
fi
fi
# Fast-fail: if the platform process died, /health will never come.
if [ -n "$PLATFORM_PID" ] && ! kill -0 "$PLATFORM_PID" 2>/dev/null; then
@@ -358,12 +371,13 @@ jobs:
exit 1
fi
if [ "$(( $(date +%s) - start ))" -ge "$DEADLINE_SECS" ]; then
echo "::error::Platform did not become healthy within ${DEADLINE_SECS}s — see log below"
echo "::error::Platform did not become healthy with migrations applied within ${DEADLINE_SECS}s — see log below"
cat workspace-server/platform.log || true
exit 1
fi
sleep 1
done
- name: Assert migrations applied
if: needs.detect-changes.outputs.api == 'true'
run: |