test(e2e): cover the register/daemon CLI and failing jobs (#1346)

The e2e suite started the runner in-process and never exercised failing jobs. This adds a scenario that builds the binary and drives `register` then `daemon` against the Gitea fixture, and one for failed, skipped, continue-on-error and timed-out jobs.

Verified locally against gitea/gitea:main-nightly.

Assisted-by: Codet:codet
Reviewed-on: https://gitea.com/gitea/runner/pulls/1346
Reviewed-by: Zettat123 <[email protected]>

DEVELOPMENT.md

@@ -59,5 +59,8 @@ Each lives in `internal/pkg/action/<name>` and is registered in `builtinActions`
 It requires Docker and is excluded from `make test`. CI runs stable and nightly variants in
 parallel.
 
+The `cli_register_daemon` scenario builds the binary and drives `register` and `daemon` as a user would. The
+`failure_paths` scenario covers failed, skipped, tolerated and timed-out jobs.
+
 The suite shares one Gitea and regular runner. Cache and ephemeral scenarios use isolated
 repository runners. A run path is `<workflow>@<ref>` and a log row is `<timestamp>Z <payload>`.

e2e/cli_test.go

@@ -0,0 +1,121 @@
+// Copyright 2026 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+//go:build e2e
+
+package e2e
+
+import (
+	"bytes"
+	"context"
+	"fmt"
+	"os"
+	"os/exec"
+	"path/filepath"
+	"sync"
+	"syscall"
+	"testing"
+	"time"
+)
+
+type syncBuffer struct {
+	mu  sync.Mutex
+	buf bytes.Buffer
+}
+
+func (b *syncBuffer) Write(p []byte) (int, error) {
+	b.mu.Lock()
+	defer b.mu.Unlock()
+	return b.buf.Write(p)
+}
+
+func (b *syncBuffer) String() string {
+	b.mu.Lock()
+	defer b.mu.Unlock()
+	return b.buf.String()
+}
+
+// testCLIRegisterDaemon drives the built binary the way a user does: register, then daemon.
+func testCLIRegisterDaemon(t *testing.T) {
+	t.Parallel()
+	ctx := t.Context()
+
+	api, repo := newScenario(t)
+	token, err := fixture.RegistrationToken(ctx, repo)
+	if err != nil {
+		t.Fatalf("get registration token: %v", err)
+	}
+
+	dir := t.TempDir()
+	bin := filepath.Join(dir, "gitea-runner")
+	if out, err := exec.CommandContext(ctx, "go", "build", "-o", bin, "..").CombinedOutput(); err != nil {
+		t.Fatalf("build runner: %v\n%s", err, out)
+	}
+
+	cfgFile := filepath.Join(dir, "config.yaml")
+	cfg := fmt.Sprintf(`runner:
+  file: %s
+  fetch_interval: 250ms
+  fetch_interval_max: 250ms
+  state_report_interval: 500ms
+  log_report_interval: 500ms
+cache:
+  dir: %s
+`, filepath.Join(dir, ".runner"), filepath.Join(dir, "cache"))
+	if fixture.network != "" {
+		cfg += "container:\n  network: " + fixture.network + "\n"
+	}
+	if err := os.WriteFile(cfgFile, []byte(cfg), 0o600); err != nil {
+		t.Fatalf("write config: %v", err)
+	}
+
+	label := "e2e-cli:docker://" + os.Getenv("E2E_JOB_IMAGE")
+	out, err := exec.CommandContext(ctx, bin, "register", "--no-interactive", "-c", cfgFile,
+		"--instance", fixture.baseURL, "--token", token, "--name", repo, "--labels", label).CombinedOutput()
+	if err != nil {
+		t.Fatalf("register: %v\n%s", err, out)
+	}
+	if _, err := os.Stat(filepath.Join(dir, ".runner")); err != nil {
+		t.Fatalf("register did not write the runner file: %v\n%s", err, out)
+	}
+
+	daemonLog := &syncBuffer{}
+	daemon := exec.Command(bin, "daemon", "-c", cfgFile)
+	daemon.Stdout, daemon.Stderr = daemonLog, daemonLog
+	if err := daemon.Start(); err != nil {
+		t.Fatalf("start daemon: %v", err)
+	}
+	exited := make(chan error, 1)
+	go func() { exited <- daemon.Wait() }()
+	t.Cleanup(func() {
+		_ = daemon.Process.Kill() // no-op once the daemon has exited
+	})
+
+	pushWorkflow(t, api, repo, "cli.yml")
+	wfRun := waitForRun(t, api, repo)
+	requireSuccessOrDumpDaemon(t, api, repo, wfRun.ID, daemonLog)
+	if logs := runLogs(t, api, repo, wfRun.ID); commandRow(logs, "e2e-cli-job-ran") == "" {
+		t.Errorf("job logs are missing the marker:\n%s", logs)
+	}
+
+	if err := daemon.Process.Signal(syscall.SIGTERM); err != nil {
+		t.Fatalf("signal daemon: %v", err)
+	}
+	select {
+	case err := <-exited:
+		if err != nil {
+			t.Errorf("daemon did not exit cleanly on SIGTERM: %v\n%s", err, daemonLog)
+		}
+	case <-time.After(30 * time.Second):
+		t.Errorf("daemon did not exit within 30s of SIGTERM\n%s", daemonLog)
+	}
+}
+
+func requireSuccessOrDumpDaemon(t *testing.T, api *GiteaAPI, repo string, runID int64, daemonLog *syncBuffer) {
+	t.Helper()
+	completed, err := api.WaitForRunConclusion(context.WithoutCancel(t.Context()), repo, runID, runTimeout)
+	if err != nil || completed.Conclusion != "success" {
+		t.Logf("daemon output:\n%s", daemonLog)
+		requireSuccess(t, api, repo, runID)
+	}
+}

e2e/compatibility_test.go

@@ -32,6 +32,8 @@ func TestCompatibility(t *testing.T) {
 	t.Run("builtin_checkout", testBuiltinCheckout)
 	t.Run("moving_action_ref", testMovingActionRef)
 	t.Run("ephemeral", testEphemeralRunner)
+	t.Run("failure_paths", testFailurePaths)
+	t.Run("cli_register_daemon", testCLIRegisterDaemon)
 	t.Run("kubernetes", testKubernetes)
 }
 

e2e/failure_test.go

@@ -0,0 +1,73 @@
+// Copyright 2026 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+//go:build e2e
+
+package e2e
+
+import (
+	"strings"
+	"testing"
+)
+
+func testFailurePaths(t *testing.T) {
+	t.Parallel()
+	ctx := t.Context()
+
+	api, repo := newScenario(t)
+	pushWorkflow(t, api, repo, "failure.yml")
+
+	wfRun := waitForRun(t, api, repo)
+	completed, err := api.WaitForRunConclusion(ctx, repo, wfRun.ID, runTimeout)
+	if err != nil {
+		dumpRunLogs(t, api, repo, wfRun.ID)
+		t.Fatalf("wait for run: %v", err)
+	}
+	if completed.Conclusion != "failure" {
+		t.Errorf("run concluded %q, want failure", completed.Conclusion)
+	}
+
+	jobs, err := api.Jobs(ctx, repo, wfRun.ID)
+	if err != nil {
+		t.Fatalf("list jobs: %v", err)
+	}
+	want := map[string]string{
+		"fail":      "failure",
+		"tolerated": "success",
+		"dependent": "skipped",
+		"timeout":   "failure",
+	}
+	for _, job := range jobs {
+		if w, ok := want[job.Name]; ok && job.Conclusion != w {
+			t.Errorf("job %q concluded %q, want %q", job.Name, job.Conclusion, w)
+		}
+		delete(want, job.Name)
+	}
+	for name := range want {
+		t.Errorf("job %q not found", name)
+	}
+
+	var all strings.Builder
+	for _, job := range jobs {
+		if job.Conclusion == "skipped" { // a skipped job never started, so it has no log
+			continue
+		}
+		jobLogs, err := api.JobLogs(ctx, repo, job.ID)
+		if err != nil {
+			t.Fatalf("job logs for %s: %v", job.Name, err)
+		}
+		all.WriteString(jobLogs)
+	}
+	logs := all.String()
+	for _, marker := range []string{"e2e-before-fail", "e2e-failure-cleanup", "e2e-tolerated-next", "e2e-timeout-started"} {
+		if commandRow(logs, marker) != marker {
+			t.Errorf("logs are missing %q", marker)
+		}
+	}
+	if commandRow(logs, "e2e-skipped-step") != "" {
+		t.Error("a success() step ran after a failed step")
+	}
+	if t.Failed() {
+		t.Logf("full stored logs:\n%s", logs)
+	}
+}

e2e/testdata/workflows/cli.yml

@@ -0,0 +1,7 @@
+name: cli
+on: push
+jobs:
+  hello:
+    runs-on: e2e-cli
+    steps:
+      - run: echo e2e-cli-job-ran

e2e/testdata/workflows/failure.yml

@@ -0,0 +1,30 @@
+name: failure
+on: push
+jobs:
+  fail:
+    runs-on: ubuntu-latest
+    steps:
+      - run: echo e2e-before-fail
+      - run: exit 3
+      - if: success()
+        run: echo e2e-skipped-step
+      - if: failure()
+        run: echo e2e-failure-cleanup
+  tolerated:
+    runs-on: ubuntu-latest
+    steps:
+      - continue-on-error: true
+        run: exit 1
+      - run: echo e2e-tolerated-next
+  dependent:
+    needs: fail
+    runs-on: ubuntu-latest
+    steps:
+      - run: echo dependent
+  timeout:
+    runs-on: ubuntu-latest
+    timeout-minutes: 1
+    steps:
+      - run: |
+          echo e2e-timeout-started
+          sleep 300