fix: keep dry runs off the docker daemon and speed up the tests (#1349)

Without a docker daemon six `act/runner` tests failed instead of passing. A dry run still created and removed job networks, the job teardown dropped the dry-run flag along with every other context value, and the container mocks asked the daemon for the runner arch.

This also speeds up the tests and CI without dropping assertions. Most of the time went to git and docker round trips, which are now pinned, batched or coalesced, and the dind and e2e steps cache or overlap more of their setup. On CI, `act/runner` went from 77s to 55s and `test-dind` from 108s to 55s. Locally, `act/runner` without docker went from 53s to 7s.

Split out of https://gitea.com/gitea/runner/pulls/1348, which adds Windows CI on top. The go cache changes moved to https://gitea.com/gitea/runner/pulls/1350.

Written by Claude Code.

---------

Co-authored-by: bircni <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1349
Reviewed-by: bircni <[email protected]>
Co-authored-by: silverwind <[email protected]>

act/container/docker_proxy.go

@@ -48,13 +48,20 @@ var (
 	rawStreamPath = regexp.MustCompile(`^(/v[0-9.]+)?/(containers/[^/]+/attach|exec/[^/]+/start)$`)
 )
 
-var dockerProxyDirectories sync.Map
+var dockerProxyDirectories, dockerProxyProbes sync.Map
 
 type dockerProxyDirectory struct {
 	dir, daemonDir string
 	socket         os.FileInfo
 }
 
+type dockerProxyProbe struct {
+	directory dockerProxyDirectory
+	done      chan struct{}
+	seen      bool
+	err       error
+}
+
 func NewDockerProxy(ctx context.Context, job string) *DockerProxy {
 	if host := os.Getenv("DOCKER_HOST"); runtime.GOOS != "linux" || host != "" && !strings.HasPrefix(host, "unix://") {
 		return nil
@@ -79,16 +86,8 @@ func NewDockerProxy(ctx context.Context, job string) *DockerProxy {
 	cached, _ := dockerProxyDirectories.Load(key)
 	directory, seen := cached.(dockerProxyDirectory)
 	seen = seen && os.SameFile(directory.socket, socket) // daemon replacement invalidates cached paths
-	dir, daemonDir := directory.dir, directory.daemonDir
 	if !seen {
-		dir, daemonDir = runnerContainerWorkdir(probeCtx, cli)
-		seen = daemonDir != ""
-		if !seen {
-			if dir, err = filepath.Abs(os.TempDir()); err == nil {
-				daemonDir = dir
-				seen, err = daemonSeesDir(probeCtx, cli, dir, daemonDir)
-			}
-		}
+		directory, seen, err = probeDockerProxyDirectory(probeCtx, cli, key, socket)
 	}
 	if err != nil {
 		common.Logger(ctx).Infof("docker proxy probe failed, jobs get the daemon socket directly: %v", err)
@@ -101,17 +100,51 @@ func NewDockerProxy(ctx context.Context, job string) *DockerProxy {
 	if ctx.Err() != nil {
 		return nil
 	}
-	proxy, err := StartDockerProxy(daemonSocket, dir, job)
+	proxy, err := StartDockerProxy(daemonSocket, directory.dir, job)
 	if err != nil {
 		dockerProxyDirectories.Delete(key)
 		common.Logger(ctx).Warnf("docker proxy not started, the job gets the daemon socket directly: %v", err)
 		return nil
 	}
-	dockerProxyDirectories.Store(key, dockerProxyDirectory{dir, daemonDir, socket})
-	proxy.Socket = daemonDir + strings.TrimPrefix(proxy.Socket, dir)
+	dockerProxyDirectories.Store(key, directory)
+	proxy.Socket = directory.daemonDir + strings.TrimPrefix(proxy.Socket, directory.dir)
 	return proxy
 }
 
+func probeDockerProxyDirectory(ctx context.Context, cli client.APIClient, key [3]string, socket os.FileInfo) (dockerProxyDirectory, bool, error) {
+	for {
+		probe := &dockerProxyProbe{directory: dockerProxyDirectory{socket: socket}, done: make(chan struct{})}
+		actual, loaded := dockerProxyProbes.LoadOrStore(key, probe)
+		if loaded {
+			probe, _ = actual.(*dockerProxyProbe)
+			if !os.SameFile(probe.directory.socket, socket) {
+				dockerProxyProbes.CompareAndDelete(key, probe)
+				continue
+			}
+			select {
+			case <-ctx.Done():
+				return dockerProxyDirectory{}, false, ctx.Err()
+			case <-probe.done:
+			}
+			if ctx.Err() == nil && (errors.Is(probe.err, context.Canceled) || errors.Is(probe.err, context.DeadlineExceeded)) {
+				continue
+			}
+		} else {
+			probe.directory.dir, probe.directory.daemonDir = runnerContainerWorkdir(ctx, cli)
+			probe.seen = probe.directory.daemonDir != ""
+			if !probe.seen {
+				if probe.directory.dir, probe.err = filepath.Abs(key[2]); probe.err == nil {
+					probe.directory.daemonDir = probe.directory.dir
+					probe.seen, probe.err = daemonSeesDir(ctx, cli, probe.directory.dir, probe.directory.daemonDir)
+				}
+			}
+			dockerProxyProbes.CompareAndDelete(key, probe)
+			close(probe.done)
+		}
+		return probe.directory, probe.seen, probe.err
+	}
+}
+
 // runnerContainerWorkdir looks the runner's container up by hostname to find the daemon's path to its working directory.
 func runnerContainerWorkdir(ctx context.Context, cli client.APIClient) (workdir, daemonDir string) {
 	workdir, err := os.Getwd()
@@ -162,12 +195,18 @@ func daemonSeesDir(ctx context.Context, cli client.APIClient, dir, daemonDir str
 	if len(images.Items) == 0 {
 		return false, errors.New("no image available for the docker proxy probe")
 	}
-	created, err := cli.ContainerCreate(ctx, client.ContainerCreateOptions{
-		Config: &container.Config{Image: images.Items[0].ID, Cmd: []string{"true"}},
-		HostConfig: &container.HostConfig{Mounts: []mount.Mount{
-			{Type: mount.TypeBind, Source: daemonDir, Target: "/gitea-runner-probe", ReadOnly: true}, // not the marker itself, podman creates a missing bind source where docker rejects it
-		}},
-	})
+	var created client.ContainerCreateResult
+	for _, image := range images.Items {
+		created, err = cli.ContainerCreate(ctx, client.ContainerCreateOptions{
+			Config: &container.Config{Image: image.ID, Cmd: []string{"true"}},
+			HostConfig: &container.HostConfig{Mounts: []mount.Mount{
+				{Type: mount.TypeBind, Source: daemonDir, Target: "/gitea-runner-probe", ReadOnly: true}, // not the marker itself, podman creates a missing bind source where docker rejects it
+			}},
+		})
+		if !cerrdefs.IsNotFound(err) {
+			break
+		}
+	}
 	if cerrdefs.IsInvalidArgument(err) {
 		return false, nil
 	}