fix: bound blocking calls and stop failing silently (#1174)

Jobs occasionally go silent ([example](https://gitea.com/gitea/runner/actions/runs/805045/jobs/1055123)) mid-run and Gitea reaped them after `ZOMBIE_TASK_TIMEOUT`, with no error in the log. This contains a number of related fixes, all with full test coverage:

1. Bound every RPC to Gitea with a timeout, a stalled report otherwise parked logs and heartbeats for the whole job.
2. Cap `runner.fetch_timeout` at that ceiling.
3. Let only the daemon loop close its own channel, the race panicked the process.
4. Stop the job on any terminal server result, not just `RESULT_CANCELLED`.
5. Report that result instead of relabelling it as cancelled.
6. Log reporting failures once at each end of an outage instead of discarding them.
7. Clamp the acknowledged log index, a too-large ack panicked on a slice bound.
8. Stop reading server health from a `FetchTask` deadline, it marked the runner healthy and reset the error backoff on a timeout.
9. Return an error from the Docker version probe instead of a `logrus` panic.
10. Pass the context to go-git's fetch and pull.
11. Fail the clone when a refresh dies on a cancelled context.
12. Set `terminationGracePeriodSeconds` in the Kubernetes examples.

Also contains a deprecation fix for goreleaser.

Reviewed-on: https://gitea.com/gitea/runner/pulls/1174
Reviewed-by: bircni <bircni@icloud.com>
Co-authored-by: silverwind <me@silverwind.io>

act/common/git/git.go

@@ -345,6 +345,16 @@ func gitOptions(token string) (fetchOptions git.FetchOptions, pullOptions git.Pu
 	return fetchOptions, pullOptions
 }
 
+// staleRefreshErr reports why a failed refresh must abort: the resolve and
+// checkout that follow are local and succeed on a cancelled context, which
+// would hand back the cached revision as if it were fresh.
+func staleRefreshErr(ctx context.Context, err error) error {
+	if err == nil || errors.Is(err, git.NoErrAlreadyUpToDate) {
+		return nil
+	}
+	return ctx.Err()
+}
+
 // NewGitCloneExecutor creates an executor to clone git repos
 func NewGitCloneExecutor(input NewGitCloneExecutorInput) common.Executor {
 	return func(ctx context.Context) error {
@@ -385,7 +395,7 @@ func NewGitCloneExecutor(input NewGitCloneExecutorInput) common.Executor {
 		}
 
 		if !isOfflineMode {
-			err = r.Fetch(&fetchOptions)
+			err = r.FetchContext(ctx, &fetchOptions)
 			if err != nil && !errors.Is(err, git.NoErrAlreadyUpToDate) {
 				return err
 			}
@@ -454,9 +464,12 @@ func NewGitCloneExecutor(input NewGitCloneExecutorInput) common.Executor {
 		switch {
 		case !isOfflineMode && !shallow:
 			// In shallow mode the depth-limited fetch above already advanced the ref.
-			if err = w.Pull(&pullOptions); err != nil && err != git.NoErrAlreadyUpToDate {
+			if err = w.PullContext(ctx, &pullOptions); err != nil && !errors.Is(err, git.NoErrAlreadyUpToDate) {
 				logger.Debugf("Unable to pull %s: %v", refName, err)
 			}
+			if err := staleRefreshErr(ctx, err); err != nil {
+				return err
+			}
 		case isOfflineMode && reused:
 			reusedMsg = " (reused in offline mode)"
 		}

act/common/git/git_test.go

@@ -6,18 +6,24 @@ package git
 
 import (
 	"context"
+	"errors"
 	"fmt"
+	"net/http"
+	"net/http/httptest"
 	"os"
 	"os/exec"
 	"path/filepath"
 	"strconv"
 	"strings"
+	"sync"
 	"syscall"
 	"testing"
 	"time"
 
 	"gitea.com/gitea/runner/act/common"
 
+	gogit "github.com/go-git/go-git/v5"
+	gogitconfig "github.com/go-git/go-git/v5/config"
 	log "github.com/sirupsen/logrus"
 	logrustest "github.com/sirupsen/logrus/hooks/test"
 	"github.com/stretchr/testify/assert"
@@ -610,3 +616,55 @@ func TestAcquireCloneLock(t *testing.T) {
 		}
 	})
 }
+
+// An unresponsive remote must not pin a job: the refresh has to be interruptible.
+func TestNewGitCloneExecutorFetchHonoursContext(t *testing.T) {
+	block := make(chan struct{})
+	reached := make(chan struct{})
+	var once sync.Once
+	server := httptest.NewServer(http.HandlerFunc(func(_ http.ResponseWriter, _ *http.Request) {
+		once.Do(func() { close(reached) })
+		<-block
+	}))
+	t.Cleanup(func() {
+		close(block)
+		server.Close()
+	})
+
+	dir := filepath.Join(t.TempDir(), "cached-action")
+	repo, err := gogit.PlainInit(dir, false)
+	require.NoError(t, err)
+	_, err = repo.CreateRemote(&gogitconfig.RemoteConfig{Name: "origin", URLs: []string{server.URL}})
+	require.NoError(t, err)
+
+	ctx, cancel := context.WithCancel(t.Context())
+	defer cancel()
+
+	done := make(chan error, 1)
+	go func() {
+		done <- NewGitCloneExecutor(NewGitCloneExecutorInput{URL: server.URL, Ref: "main", Dir: dir})(ctx)
+	}()
+
+	select {
+	case <-reached:
+	case <-time.After(10 * time.Second):
+		t.Fatal("the executor never reached the remote")
+	}
+	cancel()
+
+	select {
+	case err := <-done:
+		require.Error(t, err)
+	case <-time.After(10 * time.Second):
+		t.Fatal("fetch ignored context cancellation")
+	}
+}
+
+func TestStaleRefreshErr(t *testing.T) {
+	ctx, cancel := context.WithCancel(t.Context())
+	require.NoError(t, staleRefreshErr(ctx, errors.New("remote hung up")))
+
+	cancel()
+	require.ErrorIs(t, staleRefreshErr(ctx, errors.New("remote hung up")), context.Canceled)
+	require.NoError(t, staleRefreshErr(ctx, gogit.NoErrAlreadyUpToDate))
+}

act/container/docker_run.go

@@ -88,14 +88,14 @@ func (cr *containerReference) connectToNetwork(name string, aliases []string) co
 	}
 }
 
-// supportsContainerImagePlatform returns true if the underlying Docker server
-// API version is 1.41 and beyond
-func supportsContainerImagePlatform(ctx context.Context, cli client.APIClient) bool {
+// supportsContainerImagePlatform reports whether the Docker server API version
+// is 1.41 and beyond
+func supportsContainerImagePlatform(ctx context.Context, cli client.APIClient) (bool, error) {
 	ver, err := cli.ServerVersion(ctx, client.ServerVersionOptions{})
 	if err != nil {
-		common.Logger(ctx).Panicf("Failed to get Docker API Version: %s", err)
+		return false, fmt.Errorf("get docker API version: %w", err)
 	}
-	return versions.GreaterThanOrEqualTo(ver.APIVersion, "1.41")
+	return versions.GreaterThanOrEqualTo(ver.APIVersion, "1.41"), nil
 }
 
 func (cr *containerReference) Create(capAdd, capDrop []string) common.Executor {
@@ -682,11 +682,17 @@ func (cr *containerReference) create(capAdd, capDrop []string) common.Executor {
 		}
 
 		var platSpecs *specs.Platform
-		if cr.input.Platform != "" && supportsContainerImagePlatform(ctx, cr.cli) {
-			platSpecs, err = parsePlatform(cr.input.Platform)
+		if cr.input.Platform != "" {
+			// Dropping the platform silently would build for the host arch.
+			supported, err := supportsContainerImagePlatform(ctx, cr.cli)
 			if err != nil {
 				return err
 			}
+			if supported {
+				if platSpecs, err = parsePlatform(cr.input.Platform); err != nil {
+					return err
+				}
+			}
 		}
 
 		hostConfig := &container.HostConfig{

act/container/docker_run_test.go

@@ -79,6 +79,11 @@ type mockDockerClient struct {
 	mock.Mock
 }
 
+func (m *mockDockerClient) ServerVersion(ctx context.Context, opts mobyclient.ServerVersionOptions) (mobyclient.ServerVersionResult, error) {
+	args := m.Called(ctx, opts)
+	return args.Get(0).(mobyclient.ServerVersionResult), args.Error(1)
+}
+
 func (m *mockDockerClient) ExecCreate(ctx context.Context, id string, opts mobyclient.ExecCreateOptions) (mobyclient.ExecCreateResult, error) {
 	args := m.Called(ctx, id, opts)
 	return args.Get(0).(mobyclient.ExecCreateResult), args.Error(1)
@@ -930,3 +935,15 @@ func TestMergeContainerConfigsVolumesReplaceRunnerMounts(t *testing.T) {
 	assert.Equal(t, []string{"/var/run/docker.sock:/var/run/docker.sock", "/host/tools:/opt/hostedtoolcache"}, hostConf.Binds)
 	assert.Empty(t, hostConf.Mounts)
 }
+
+// A dead daemon must fail the job, not panic through logrus and not silently
+// drop the requested platform.
+func TestSupportsContainerImagePlatformDaemonError(t *testing.T) {
+	cli := &mockDockerClient{}
+	cli.On("ServerVersion", mock.Anything, mock.Anything).
+		Return(mobyclient.ServerVersionResult{}, errors.New("cannot connect to the Docker daemon"))
+
+	supported, err := supportsContainerImagePlatform(t.Context(), cli)
+	require.ErrorContains(t, err, "cannot connect to the Docker daemon")
+	assert.False(t, supported)
+}