perf: optimize git clone/fetch for large repos with many tags (#533)

apps/backend/internal/repoclone/clone.go

@@ -16,7 +16,10 @@ import (
 )
 
 // ghCredentialHelper is the git credential helper command that delegates to gh CLI.
-const ghCredentialHelper = "!gh auth git-credential"
+const (
+	ghCredentialHelper = "!gh auth git-credential"
+	gitNoTags          = "--no-tags"
+)
 
 // Config holds configuration for the repository cloner.
 type Config struct {
@@ -122,7 +125,7 @@ func (c *Cloner) gitCmd(ctx context.Context, args ...string) *exec.Cmd {
 
 func (c *Cloner) fetch(ctx context.Context, repoPath string) {
 	c.logger.Debug("repository already cloned, fetching", zap.String("path", repoPath))
-	cmd := c.gitCmd(ctx, "-C", repoPath, "fetch", "--all", "--prune", "--force")
+	cmd := c.gitCmd(ctx, "-C", repoPath, "fetch", "--all", "--prune", "--force", gitNoTags)
 	if out, err := cmd.CombinedOutput(); err != nil {
 		c.logger.Warn("git fetch failed (non-fatal)",
 			zap.String("path", repoPath),
@@ -153,7 +156,7 @@ func (c *Cloner) clone(ctx context.Context, cloneURL, targetPath, owner, name st
 	}
 
 	// Fallback: git clone with credential helper.
-	cmd := c.gitCmd(ctx, "clone", cloneURL, targetPath)
+	cmd := c.gitCmd(ctx, "clone", "--filter=blob:none", gitNoTags, cloneURL, targetPath)
 	if out, err := cmd.CombinedOutput(); err != nil {
 		return fmt.Errorf("git clone failed: %s: %w", string(out), err)
 	}
@@ -164,7 +167,7 @@ func (c *Cloner) clone(ctx context.Context, cloneURL, targetPath, owner, name st
 // automatically via the user's gh CLI session.
 func (c *Cloner) ghClone(ctx context.Context, owner, name, targetPath string) error {
 	nwo := owner + "/" + name
-	cmd := exec.CommandContext(ctx, "gh", "repo", "clone", nwo, targetPath)
+	cmd := exec.CommandContext(ctx, "gh", "repo", "clone", nwo, targetPath, "--", "--filter=blob:none", gitNoTags)
 	cmd.Env = append(os.Environ(), "GIT_TERMINAL_PROMPT=0")
 	out, err := cmd.CombinedOutput()
 	if err != nil {

apps/backend/internal/repoclone/clone_test.go

@@ -0,0 +1,77 @@
+package repoclone
+
+import (
+	"context"
+	"os"
+	"os/exec"
+	"path/filepath"
+	"testing"
+
+	"github.com/kandev/kandev/internal/common/logger"
+)
+
+func TestClone_PreservesNonDefaultRemoteBranches(t *testing.T) {
+	t.Parallel()
+
+	originPath := initBareRepoWithReleaseBranch(t)
+	targetPath := filepath.Join(t.TempDir(), "clone")
+
+	cloner := NewCloner(Config{}, ProtocolSSH, t.TempDir(), logger.Default())
+	if err := cloner.clone(context.Background(), originPath, targetPath, "", ""); err != nil {
+		t.Fatalf("clone() unexpected error: %v", err)
+	}
+
+	if !gitRefExists(t, targetPath, "refs/remotes/origin/release") {
+		t.Fatal("expected cloned repo to contain origin/release for downstream worktree base branches")
+	}
+}
+
+func initBareRepoWithReleaseBranch(t *testing.T) string {
+	t.Helper()
+
+	root := t.TempDir()
+	originPath := filepath.Join(root, "origin.git")
+	workPath := filepath.Join(root, "work")
+
+	runGit(t, root, "init", "--bare", "-b", "main", originPath)
+	runGit(t, root, "clone", originPath, workPath)
+	runGit(t, workPath, "config", "user.email", "test@example.com")
+	runGit(t, workPath, "config", "user.name", "Test User")
+
+	readmePath := filepath.Join(workPath, "README.md")
+	if err := os.WriteFile(readmePath, []byte("main\n"), 0o644); err != nil {
+		t.Fatalf("write README.md: %v", err)
+	}
+	runGit(t, workPath, "add", "README.md")
+	runGit(t, workPath, "commit", "-m", "main commit")
+	runGit(t, workPath, "push", "origin", "main")
+
+	runGit(t, workPath, "checkout", "-b", "release")
+	if err := os.WriteFile(readmePath, []byte("release\n"), 0o644); err != nil {
+		t.Fatalf("write release README.md: %v", err)
+	}
+	runGit(t, workPath, "commit", "-am", "release commit")
+	runGit(t, workPath, "push", "origin", "release")
+
+	return originPath
+}
+
+func gitRefExists(t *testing.T, repoPath, ref string) bool {
+	t.Helper()
+
+	cmd := exec.Command("git", "show-ref", "--verify", "--quiet", ref)
+	cmd.Dir = repoPath
+	return cmd.Run() == nil
+}
+
+func runGit(t *testing.T, repoPath string, args ...string) string {
+	t.Helper()
+
+	cmd := exec.Command("git", args...)
+	cmd.Dir = repoPath
+	output, err := cmd.CombinedOutput()
+	if err != nil {
+		t.Fatalf("git %v failed: %v\n%s", args, err, output)
+	}
+	return string(output)
+}

apps/backend/internal/worktree/manager.go

@@ -18,8 +18,9 @@ import (
 )
 
 const (
-	defaultGitFetchTimeout = 60 * time.Second
+	defaultGitFetchTimeout = 90 * time.Second
 	defaultGitPullTimeout  = 60 * time.Second
+	gitNoTags              = "--no-tags"
 )
 
 // repoLockEntry tracks a repository lock and its reference count.
@@ -443,10 +444,10 @@ func (m *Manager) fetchBranchToLocal(ctx context.Context, repoPath, branch strin
 		zap.String("repo_path", repoPath))
 
 	// Try to fetch from origin to get the latest version.
-	fetchCtx, cancelFetch := context.WithTimeout(ctx, 30*time.Second)
+	fetchCtx, cancelFetch := context.WithTimeout(ctx, m.fetchTimeout)
 	defer cancelFetch()
 
-	fetchCmd := m.newNonInteractiveGitCmd(fetchCtx, repoPath, "fetch", "origin", branch+":"+branch)
+	fetchCmd := m.newNonInteractiveGitCmd(fetchCtx, repoPath, "fetch", gitNoTags, "origin", branch+":"+branch)
 	if output, err := fetchCmd.CombinedOutput(); err != nil {
 		outputStr := string(output)
 
@@ -454,7 +455,7 @@ func (m *Manager) fetchBranchToLocal(ctx context.Context, repoPath, branch strin
 		// the local ref. Retry by fetching only the remote-tracking ref (origin/branch),
 		// which is always safe regardless of worktree state.
 		if isFetchRefusedCheckedOut(outputStr) {
-			retryCmd := m.newNonInteractiveGitCmd(fetchCtx, repoPath, "fetch", "origin", branch)
+			retryCmd := m.newNonInteractiveGitCmd(fetchCtx, repoPath, "fetch", gitNoTags, "origin", branch)
 			if _, retryErr := retryCmd.CombinedOutput(); retryErr == nil {
 				m.logger.Info("fetched via remote-tracking ref (branch checked out elsewhere)",
 					zap.String("branch", branch))
@@ -1272,7 +1273,7 @@ func (m *Manager) pullBaseBranch(ctx context.Context, repoPath, baseBranch strin
 	fetchCtx, cancelFetch := context.WithTimeout(ctx, m.fetchTimeout)
 	defer cancelFetch()
 
-	fetchArgs := []string{"fetch", "origin"}
+	fetchArgs := []string{"fetch", gitNoTags, "origin"}
 	if localBranch != "" {
 		fetchArgs = append(fetchArgs, localBranch)
 	}