Skip to content

fix(git): exclude temporary files from snapshots #413

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
joshfriend merged 1 commit into from
Oct 7, 2026
Merged

fix(git): exclude temporary files from snapshots #413

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 8 additions & 4 deletions internal/strategy/git/snapshot.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -220,6 +220,10 @@ func (s *Strategy) withSnapshotClone(ctx context.Context, repo *gitclone.Reposit
return fn(workDir)
}

func snapshotExcludePatterns(gitDir string) []string {
return []string{gitDir + "/*.lock", gitDir + "/objects/pack/tmp_*"}
}

func (s *Strategy) generateAndUploadSnapshot(ctx context.Context, repo *gitclone.Repository) (commit string, returnErr error) {
upstream := repo.UpstreamURL()
ctx, span := tracer.Start(ctx, "git.snapshot.generate",
Expand Down Expand Up @@ -266,7 +270,7 @@ func (s *Strategy) generateAndUploadSnapshot(ctx context.Context, repo *gitclone
extraHeaders := http.Header{}
extraHeaders.Set("X-Cachew-Snapshot-Commit", headSHA)

return snapshot.Create(ctx, s.cache, cacheKey, workDir, 0, nil, s.config.ZstdThreads, extraHeaders)
return snapshot.Create(ctx, s.cache, cacheKey, workDir, 0, snapshotExcludePatterns("./.git"), s.config.ZstdThreads, extraHeaders)
}); err != nil {
return "", errors.Wrap(err, "create snapshot")
}
Expand Down Expand Up @@ -306,7 +310,7 @@ func (s *Strategy) generateAndUploadMirrorSnapshot(ctx context.Context, repo *gi
defer mu.Unlock()

cacheKey := mirrorSnapshotCacheKey(upstream)
excludePatterns := []string{"*.lock"}
excludePatterns := snapshotExcludePatterns(".")

// Hold the fetch semaphore while tar-ing the bare mirror directory.
// Without this, a concurrent git fetch can replace packed-refs mid-read,
Expand Down Expand Up @@ -1174,7 +1178,7 @@ func (s *Strategy) streamSnapshotDirect(w http.ResponseWriter, r *http.Request,
w.Header().Set("Content-Type", "application/zstd")
w.Header().Set("Content-Disposition", fmt.Sprintf("attachment; filename=%q", filepath.Base(repoDir)+".tar.zst"))

return errors.Wrap(snapshot.StreamTo(ctx, w, repoDir, nil, s.config.ZstdThreads), "stream snapshot to client")
return errors.Wrap(snapshot.StreamTo(ctx, w, repoDir, snapshotExcludePatterns("./.git"), s.config.ZstdThreads), "stream snapshot to client")
}

// prepareSnapshotSpool creates the spool and clones the mirror into a temp directory,
Expand Down Expand Up @@ -1237,7 +1241,7 @@ func (s *Strategy) writeSnapshotSpool(w http.ResponseWriter, r *http.Request, re
w.Header().Set("Content-Disposition", fmt.Sprintf("attachment; filename=%q", filepath.Base(repoDir)+".tar.zst"))

tw := NewSpoolTeeWriter(w, spool)
streamErr := snapshot.StreamTo(ctx, tw, repoDir, nil, s.config.ZstdThreads)
streamErr := snapshot.StreamTo(ctx, tw, repoDir, snapshotExcludePatterns("./.git"), s.config.ZstdThreads)
if streamErr != nil {
spool.MarkError(streamErr)
s.metrics.recordSpoolWriter(ctx, repoName, "error", time.Since(writerStart))
Expand Down
54 changes: 50 additions & 4 deletions internal/strategy/git/snapshot_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -109,6 +109,7 @@ func TestSnapshotOnDemandGenerationViaHTTP(t *testing.T) {
mirrorRoot := filepath.Join(tmpDir, "mirrors")
mirrorPath := filepath.Join(mirrorRoot, "github.com", "org", "repo")
createTestMirrorRepo(t, mirrorPath)
addSnapshotGarbage(t, mirrorPath)

memCache, err := cache.NewMemory(ctx, cache.MemoryConfig{MaxTTL: time.Hour})
assert.NoError(t, err)
Expand Down Expand Up @@ -136,6 +137,42 @@ func TestSnapshotOnDemandGenerationViaHTTP(t *testing.T) {
// Allow background goroutines (spool cleanup, cache backfill) to finish
// before TempDir cleanup runs.
time.Sleep(2 * time.Second)

restoreDir := t.TempDir()
assert.NoError(t, snapshot.Extract(ctx, w.Body, restoreDir, 0))
assertSnapshotGitFiles(t, mirrorPath, filepath.Join(restoreDir, ".git"))
}

func addSnapshotGarbage(t *testing.T, mirrorPath string) {
t.Helper()
output, err := exec.Command("git", "-C", mirrorPath, "repack", "-ad").CombinedOutput()
assert.NoError(t, err, string(output))
for _, name := range []string{"objects/pack/tmp_pack_unused", "objects/pack/tmp_idx_unused", "objects/pack/stale.lock"} {
assert.NoError(t, os.WriteFile(filepath.Join(mirrorPath, name), []byte("garbage"), 0o644))
}
}

func assertSnapshotGitFiles(t *testing.T, mirrorPath, restoredGitDir string) {
t.Helper()
for _, name := range []string{"objects/pack/tmp_pack_unused", "objects/pack/tmp_idx_unused", "objects/pack/stale.lock"} {
_, err := os.Stat(filepath.Join(restoredGitDir, name))
assert.True(t, os.IsNotExist(err), "snapshot should exclude %s", name)
data, err := os.ReadFile(filepath.Join(mirrorPath, name))
assert.NoError(t, err)
assert.Equal(t, "garbage", string(data))
}
packs, err := filepath.Glob(filepath.Join(mirrorPath, "objects/pack/pack-*"))
assert.NoError(t, err)
assert.True(t, len(packs) >= 2, "expected a real pack and index")
for _, pack := range packs {
want, err := os.ReadFile(pack)
assert.NoError(t, err)
got, err := os.ReadFile(filepath.Join(restoredGitDir, "objects/pack", filepath.Base(pack)))
assert.NoError(t, err)
assert.Equal(t, want, got)
}
output, err := exec.CommandContext(t.Context(), "git", "-C", restoredGitDir, "-c", "core.fsmonitor=false", "fsck", "--full").CombinedOutput()
assert.NoError(t, err, string(output))
}

// createTestMirrorRepo creates a bare mirror-style repo at mirrorPath with one commit.
Expand Down Expand Up @@ -460,7 +497,7 @@ func createTestMirrorRepoWithHistory(t *testing.T, mirrorPath string) string {
return historicalBlob
}

func TestSnapshotGenerationIncludesTrackedLockFiles(t *testing.T) {
func TestSnapshotGenerationExcludesGitGarbage(t *testing.T) {
if _, err := exec.LookPath("git"); err != nil {
t.Skip("git not found in PATH")
}
Expand All @@ -472,10 +509,12 @@ func TestSnapshotGenerationIncludesTrackedLockFiles(t *testing.T) {

mirrorPath := filepath.Join(mirrorRoot, "github.com", "org", "repo")
createTestMirrorRepoWithFiles(t, mirrorPath, map[string]string{
"hello.txt": "hello\n",
"package-lock.json": "{\n \"name\": \"repo\"\n}\n",
"subdir/Gemfile.lock": "GEM\n",
"hello.txt": "hello\n",
"package-lock.json": "{\n \"name\": \"repo\"\n}\n",
"subdir/Gemfile.lock": "GEM\n",
"objects/pack/tmp_pack_test": "tracked fixture\n",
})
addSnapshotGarbage(t, mirrorPath)

memCache, err := cache.NewMemory(ctx, cache.MemoryConfig{MaxTTL: time.Hour})
assert.NoError(t, err)
Expand Down Expand Up @@ -507,6 +546,11 @@ func TestSnapshotGenerationIncludesTrackedLockFiles(t *testing.T) {
data, err = os.ReadFile(filepath.Join(restoreDir, "subdir", "Gemfile.lock"))
assert.NoError(t, err)
assert.Equal(t, "GEM\n", string(data))

data, err = os.ReadFile(filepath.Join(restoreDir, "objects", "pack", "tmp_pack_test"))
assert.NoError(t, err)
assert.Equal(t, "tracked fixture\n", string(data))
assertSnapshotGitFiles(t, mirrorPath, filepath.Join(restoreDir, ".git"))
}

func TestMirrorSnapshotRestoreDirectly(t *testing.T) {
Expand All @@ -522,6 +566,7 @@ func TestMirrorSnapshotRestoreDirectly(t *testing.T) {
// Create a mirror repo and generate a mirror snapshot (bare tarball).
mirrorPath := filepath.Join(mirrorRoot, "github.com", "org", "repo")
createTestMirrorRepo(t, mirrorPath)
addSnapshotGarbage(t, mirrorPath)

memCache, err := cache.NewMemory(ctx, cache.MemoryConfig{MaxTTL: time.Hour})
assert.NoError(t, err)
Expand All @@ -545,6 +590,7 @@ func TestMirrorSnapshotRestoreDirectly(t *testing.T) {
cacheKey := cache.NewKey(upstreamURL + ".mirror-snapshot")
err = snapshot.Restore(ctx, memCache, cacheKey, restoreDir, 0)
assert.NoError(t, err)
assertSnapshotGitFiles(t, mirrorPath, restoreDir)

// Should be bare already (no .git subdir).
_, err = os.Stat(filepath.Join(restoreDir, ".git"))
Expand Down
Loading