v0.67.0: prune runner dind leftovers after e2e

v0.66.0 — E2E results in Gitea job summary (#141)

.gitea/workflows/ci.yml

@@ -189,10 +189,56 @@ jobs:
           # in-network app host out of any proxy path.
           NO_PROXY: app,localhost,127.0.0.1,.kkhost.pl
           no_proxy: app,localhost,127.0.0.1,.kkhost.pl
+          # Machine-readable e2e results for the job-summary step below (Playwright's built-in
+          # json reporter; without this env var it would dump to stdout). test-results/ is git-ignored.
+          PLAYWRIGHT_JSON_OUTPUT_FILE: test-results/e2e.json
         # Connectivity diagnostics now run post-deploy in e2e/ci-diagnostics.ts (globalSetup).
         # --reporter=list prints each test as it completes (name + ✓/× + duration) for progress
-        # visibility in the job log; failure error context/snapshots are unchanged.
+        # visibility in the job log; failure error context/snapshots are unchanged. The json
-        run: npx playwright test --reporter=list
+        # reporter adds a machine-readable report (PLAYWRIGHT_JSON_OUTPUT_FILE above) — no
+        # execution behavior changes, reporters only observe.
+        run: npx playwright test --reporter=list,json
+
+      # Appends e2e results to the job-summary panel — mirrors how unit tests get there
+      # (vitest's built-in github-actions reporter auto-appends $GITHUB_STEP_SUMMARY in CI).
+      # if: always() so a FAILED suite still gets summarized; non-fatal by design — a missing
+      # or malformed report must never mask the real e2e exit code. The script is single-quoted
+      # for bash, so it contains no apostrophes.
+      - name: E2E job summary
+        if: always()
+        run: |
+          node -e '
+            const fs = require("fs");
+            let report;
+            try {
+              report = JSON.parse(fs.readFileSync("test-results/e2e.json", "utf8"));
+            } catch {
+              process.exit(0); // e2e never ran (earlier step failed) or no report — nothing to summarize
+            }
+            let passed = 0, skipped = 0;
+            const failed = [];
+            const walk = (suite) => {
+              for (const spec of suite.specs ?? []) {
+                if (!spec.ok) failed.push(`${suite.title} › ${spec.title}`);
+                else if ((spec.tests ?? []).length && spec.tests.every((t) => t.expectedStatus === "skipped")) skipped++;
+                else passed++;
+              }
+              for (const sub of suite.suites ?? []) walk(sub);
+            };
+            for (const s of report.suites ?? []) walk(s);
+            const lines = [
+              "", "## E2E results", "",
+              `- Passed: ${passed}`,
+              `- Failed: ${failed.length}`,
+              `- Skipped: ${skipped}`,
+            ];
+            if (failed.length) {
+              lines.push("", "### Failed specs");
+              for (const f of failed) lines.push(`- \`${f}\``);
+            }
+            const summaryFile = process.env.GITHUB_STEP_SUMMARY;
+            if (summaryFile) fs.appendFileSync(summaryFile, lines.join("\n") + "\n");
+          ' || true
 
       # The daemon is shared with production — never leave our stack or network behind.
       # -v drops the named DB volume so every run seeds fresh (no schema drift between PRs).
@@ -201,3 +247,13 @@ jobs:
         run: |
           docker compose -f docker-compose.ci.yml -p flight-ci down --remove-orphans -v || true
           docker network remove flight-ci-e2e 2>/dev/null || true
+
+      # The shared runner dind still accumulates between runs what the steps above can't reach:
+      # dangling images/layers from the per-run flight-ops-ci rebuilds, BuildKit cache, and job
+      # containers leaked by interrupted runs. Prune them here (after stack + network cleanup).
+      # Deliberately NO --volumes: the daemon is shared with production and named volumes may hold
+      # other workflows' persistent data — this project's own flight-ci_flight-db volume is already
+      # dropped by the steps above. Non-fatal so a prune failure can never mask or break the run.
+      - name: Prune runner dind leftovers
+        if: always()
+        run: docker system prune -af || echo "::warning::dind prune failed — see logs"

CHANGELOG.md

@@ -1,5 +1,11 @@
 # Changelog
 
+## v0.67.0 — 2026-09-23 18:03
+- **Prune runner dind leftovers after e2e.** The CI runner's full-dind daemon is shared with production and other workflows, so between runs it accumulates what the existing cleanup steps can't reach: dangling images/layers from the per-run `flight-ops-ci` rebuilds, BuildKit cache (persists by design), and job containers leaked by interrupted runs. A new non-fatal "Prune runner dind leftovers" step (`if: always()`, placed after the existing "Cleanup e2e stack + network") runs `docker system prune -af || echo "::warning::dind prune failed — see logs"` so a prune failure can never mask or break the run's real result. Deliberately NO `--volumes`: named volumes on the shared daemon may hold other workflows' persistent data — this project's own `flight-ci_flight-db` volume is already dropped by the existing cleanup steps. No app/TS changes; CI validates the workflow itself.
+
+## v0.66.0 — 2026-09-23 17:06
+- **E2E results in the Gitea job summary (#141).** The run page's "Job summaries" panel showed vitest results (vitest's built-in `github-actions` reporter auto-appends `$GITHUB_STEP_SUMMARY` when `GITHUB_ACTIONS=true`) but nothing about the Playwright suite, so a failed e2e run required digging into raw logs to see which specs broke. The CI E2E step now also runs Playwright's built-in `json` reporter (`--reporter=list,json`, output pinned to git-ignored `test-results/e2e.json` via `PLAYWRIGHT_JSON_OUTPUT_FILE` — no new dependencies, no config/TS changes), and a new non-fatal "E2E job summary" step (`if: always()`) parses that report with an inline node script and appends spec-level passed/failed/skipped counts plus the file + title of every failed spec to `$GITHUB_STEP_SUMMARY` — mirroring exactly how vitest gets there. The append is guarded (`|| true`, missing/malformed report = silent no-op) so it can never fail or mask the e2e step's real exit code; test behavior, timeouts, workers and retries are untouched.
+
 ## v0.65.0 — 2026-09-23 12:58
 - **Per-airport job pools with real-time refresh.** The market was a global bag of jobs (top-up to `MARKET_TARGET_OPEN_PACKAGES = 10` + coverage spawn at uncovered grounded airports), so the board showed work you could never take and hid nothing. Now open jobs are generated PER AIRPORT where a plane can actually pick them up — the pool is the set of grounded planes' `locationIcao`, flying planes' `destIcao` and queued-leg ICAOs — queued legs count even while their plane is in the shop (the work will be waiting at its next stop); a fleet fully repairing with empty queues has no pool, and orphaned pools simply expire out. Each pool airport holds exactly `POOL_TARGET_PER_AIRPORT = 10` open jobs in a fixed weight composition (`POOL_BANDS`: 2×[500,4000) + 6×[4000,18000) + 1×[18000,48000] + 1×(48000,80000] kg), so the board always mixes jobs you can take now with heavier ones that are what you buy better planes for. `spawnPoolPackages` in `lib/sim/engine.ts` replaces both old spawners and runs on every tick (including frozen ones) like v0.64.0's top-up; seeded and spawned packages come from the same new `generatePoolPackage` helper in `lib/game/seed.ts`, so they are indistinguishable — `buildSeed` now seeds a full pool at the starter's home hub, and `PACKAGE_COUNT_MIN/MAX` + `MARKET_TARGET_OPEN_PACKAGES` are retired.
 - **Open jobs live on the wall clock.** New `created_real_ms INTEGER NOT NULL DEFAULT 0` column (guarded ALTER backfilled with `Date.now()` for existing rows): an open job expires `JOB_LIFETIME_MS = 1 h` after creation regardless of time scale or fleet state — checked every tick OUTSIDE the advance gate, so a paused world still refreshes its board. Loaded jobs keep their sim-clock deadline logic unchanged (delivery check, expiry floor, abandon penalty all untouched). Taken slots (loaded OR ignored) stamp `slot_free_at_real_ms` and hold their pool slot for a refill cooldown — `REFILL_DELAY_MS = 15 min`, or `SMALL_REFILL_DELAY_MS = 1 min` for the <4 t band — so the board visibly refreshes instead of instantly refilling; expired/delivered/abandoned never block. ponytail: one shared per-airport cooldown (the max stamp) rather than a per-band table, add when per-band refill timing matters.

package.json

@@ -1,6 +1,6 @@
 {
   "name": "flight-ops",
-  "version": "0.65.0",
+  "version": "0.67.0",
   "private": true,
   "scripts": {
     "dev": "next dev -p 3000",