Remove compression of .tar bloom blocks

Decompression is a CPU intensive task, especially un-gzipping. The gain
of compressing a tar archive of storage optimized binary blocks is
rather neglectable.

Signed-off-by: Christian Haudum <christian.haudum@gmail.com>

pkg/bloombuild/planner/planner_test.go

@@ -202,7 +202,7 @@ func genBlock(ref bloomshipper.BlockRef) (bloomshipper.Block, error) {
 	block := v1.NewBlock(reader, v1.NewMetrics(nil))
 
 	buf := bytes.NewBuffer(nil)
-	if err := v1.TarGz(buf, block.Reader()); err != nil {
+	if err := v1.Tar(buf, block.Reader()); err != nil {
 		return bloomshipper.Block{}, err
 	}
 
@@ -1019,7 +1019,7 @@ func Test_deleteOutdatedMetas(t *testing.T) {
 	} {
 		t.Run(tc.name, func(t *testing.T) {
 			logger := log.NewNopLogger()
-			//logger := log.NewLogfmtLogger(os.Stdout)
+			// logger := log.NewLogfmtLogger(os.Stdout)
 
 			cfg := Config{
 				PlanningInterval:        1 * time.Hour,

pkg/bloomgateway/bloomgateway_test.go

@@ -165,7 +165,7 @@ func TestBloomGateway_FilterChunkRefs(t *testing.T) {
 			Through: now,
 			Refs:    groupRefs(t, chunkRefs),
 			Plan:    plan.QueryPlan{AST: expr},
-			Blocks:  []string{"bloom/invalid/block.tar.gz"},
+			Blocks:  []string{"bloom/invalid/block.tar"},
 		}
 
 		ctx := user.InjectOrgID(context.Background(), tenantID)

pkg/storage/stores/shipper/bloomshipper/cache_test.go

@@ -64,7 +64,7 @@ func Test_LoadBlocksDirIntoCache(t *testing.T) {
 	wd := t.TempDir()
 
 	// plain file
-	fp, _ := os.Create(filepath.Join(wd, "regular-file.tar.gz"))
+	fp, _ := os.Create(filepath.Join(wd, "regular-file.tar"))
 	fp.Close()
 
 	// invalid directory
@@ -100,7 +100,7 @@ func Test_LoadBlocksDirIntoCache(t *testing.T) {
 
 	require.Equal(t, 1, len(c.entries))
 
-	key := validDir + v1.ExtTarGz // cache key must not contain directory prefix
+	key := validDir + v1.ExtTar // cache key must not contain directory prefix
 	elem, found := c.entries[key]
 	require.True(t, found)
 	blockDir := elem.Value.(*Entry).Value

pkg/storage/stores/shipper/bloomshipper/client.go

@@ -227,10 +227,10 @@ func BlockFrom(tenant, table string, blk *v1.Block) (Block, error) {
 
 	// TODO(owen-d): pool
 	buf := bytes.NewBuffer(nil)
-	err := v1.TarGz(buf, blk.Reader())
+	err := v1.Tar(buf, blk.Reader())
 
 	if err != nil {
-		return Block{}, errors.Wrap(err, "archiving+compressing block")
+		return Block{}, err
 	}
 
 	reader := bytes.NewReader(buf.Bytes())
@@ -321,14 +321,14 @@ func (b *BloomClient) GetBlock(ctx context.Context, ref BlockRef) (BlockDirector
 	defer rc.Close()
 
 	path := b.fsResolver.Block(ref).LocalPath()
-	// the block directory should not contain the .tar.gz extension
-	path = strings.TrimSuffix(path, v1.ExtTarGz)
+	// the block directory should not contain the .tar extension
+	path = strings.TrimSuffix(path, v1.ExtTar)
 	err = util.EnsureDirectory(path)
 	if err != nil {
 		return BlockDirectory{}, fmt.Errorf("failed to create block directory %s: %w", path, err)
 	}
 
-	err = v1.UnTarGz(path, rc)
+	err = v1.UnTar(path, rc)
 	if err != nil {
 		return BlockDirectory{}, fmt.Errorf("failed to extract block file %s: %w", key, err)
 	}

pkg/storage/stores/shipper/bloomshipper/client_test.go

@@ -201,13 +201,13 @@ func putBlock(t *testing.T, c *BloomClient, tenant string, start model.Time, min
 	day := start.Unix() / step
 
 	tmpDir := t.TempDir()
-	fp, _ := os.CreateTemp(t.TempDir(), "*.tar.gz")
+	fp, _ := os.CreateTemp(t.TempDir(), "*.tar")
 
 	blockWriter := v1.NewDirectoryBlockWriter(tmpDir)
 	err := blockWriter.Init()
 	require.NoError(t, err)
 
-	err = v1.TarGz(fp, v1.NewDirectoryBlockReader(tmpDir))
+	err = v1.Tar(fp, v1.NewDirectoryBlockReader(tmpDir))
 	require.NoError(t, err)
 
 	_, _ = fp.Seek(0, 0)
@@ -277,13 +277,13 @@ func TestBloomClient_PutBlock(t *testing.T) {
 	start := parseTime("2024-02-05 12:00")
 
 	tmpDir := t.TempDir()
-	fp, _ := os.CreateTemp(t.TempDir(), "*.tar.gz")
+	fp, _ := os.CreateTemp(t.TempDir(), "*.tar")
 
 	blockWriter := v1.NewDirectoryBlockWriter(tmpDir)
 	err := blockWriter.Init()
 	require.NoError(t, err)
 
-	err = v1.TarGz(fp, v1.NewDirectoryBlockReader(tmpDir))
+	err = v1.Tar(fp, v1.NewDirectoryBlockReader(tmpDir))
 	require.NoError(t, err)
 
 	block := Block{

pkg/storage/stores/shipper/bloomshipper/fetcher.go

@@ -408,9 +408,9 @@ func (f *Fetcher) loadBlocksFromFS(_ context.Context, refs []BlockRef) ([]BlockD
 
 	for _, ref := range refs {
 		path := f.localFSResolver.Block(ref).LocalPath()
-		// the block directory does not contain the .tar.gz extension
+		// the block directory does not contain the .tar extension
 		// since it is stripped when the archive is extracted into a folder
-		path = strings.TrimSuffix(path, v1.ExtTarGz)
+		path = strings.TrimSuffix(path, v1.ExtTar)
 		if ok, clean := f.isBlockDir(path); ok {
 			blockDirs = append(blockDirs, NewBlockDirectory(ref, path))
 		} else {

pkg/storage/stores/shipper/bloomshipper/fetcher_test.go

@@ -336,9 +336,9 @@ func TestFetcher_LoadBlocksFromFS(t *testing.T) {
 		{Ref: Ref{TenantID: "tenant", TableName: "12345", Bounds: v1.NewBounds(0x2000, 0x2fff)}},
 	}
 	dirs := []string{
-		strings.TrimSuffix(resolver.Block(refs[0]).LocalPath(), v1.ExtTarGz),
-		strings.TrimSuffix(resolver.Block(refs[1]).LocalPath(), v1.ExtTarGz),
-		strings.TrimSuffix(resolver.Block(refs[2]).LocalPath(), v1.ExtTarGz),
+		strings.TrimSuffix(resolver.Block(refs[0]).LocalPath(), v1.ExtTar),
+		strings.TrimSuffix(resolver.Block(refs[1]).LocalPath(), v1.ExtTar),
+		strings.TrimSuffix(resolver.Block(refs[2]).LocalPath(), v1.ExtTar),
 	}
 
 	createBlockDir(t, dirs[1])

pkg/storage/stores/shipper/bloomshipper/resolver.go

@@ -18,7 +18,7 @@ const (
 	BlocksPrefix = "blocks"
 
 	metaExtension  = ".json"
-	blockExtension = v1.ExtTarGz
+	blockExtension = v1.ExtTar
 )
 
 // KeyResolver is an interface for resolving keys to locations.

pkg/storage/stores/shipper/bloomshipper/resolver_test.go

@@ -46,7 +46,7 @@ func TestResolver_ParseBlockKey(t *testing.T) {
 	// encode block ref as string
 	loc := r.Block(ref)
 	path := loc.LocalPath()
-	require.Equal(t, "bloom/table_1/tenant/blocks/0000000000000000-000000000000ffff/0-3600000-abcd.tar.gz", path)
+	require.Equal(t, "bloom/table_1/tenant/blocks/0000000000000000-000000000000ffff/0-3600000-abcd.tar", path)
 
 	// parse encoded string into block ref
 	parsed, err := r.ParseBlockKey(key(path))
@@ -87,7 +87,7 @@ func TestResolver_ShardedPrefixedResolver(t *testing.T) {
 		loc := r.Meta(metaRef)
 		require.Equal(t, "prefix/bloom/table_1/tenant/metas/0000000000000000-000000000000ffff-abcd.json", loc.LocalPath())
 		loc = r.Block(blockRef)
-		require.Equal(t, "prefix/bloom/table_1/tenant/blocks/0000000000000000-000000000000ffff/0-3600000-bcde.tar.gz", loc.LocalPath())
+		require.Equal(t, "prefix/bloom/table_1/tenant/blocks/0000000000000000-000000000000ffff/0-3600000-bcde.tar", loc.LocalPath())
 	})
 
 	t.Run("multiple prefixes", func(t *testing.T) {
@@ -96,6 +96,6 @@ func TestResolver_ShardedPrefixedResolver(t *testing.T) {
 		loc := r.Meta(metaRef)
 		require.Equal(t, "b/bloom/table_1/tenant/metas/0000000000000000-000000000000ffff-abcd.json", loc.LocalPath())
 		loc = r.Block(blockRef)
-		require.Equal(t, "d/bloom/table_1/tenant/blocks/0000000000000000-000000000000ffff/0-3600000-bcde.tar.gz", loc.LocalPath())
+		require.Equal(t, "d/bloom/table_1/tenant/blocks/0000000000000000-000000000000ffff/0-3600000-bcde.tar", loc.LocalPath())
 	})
 }

pkg/storage/stores/shipper/bloomshipper/store_test.go

@@ -109,13 +109,13 @@ func createMetaInStorage(store *BloomStore, tenant string, start model.Time, min
 
 func createBlockInStorage(t *testing.T, store *BloomStore, tenant string, start model.Time, minFp, maxFp model.Fingerprint) (Block, error) {
 	tmpDir := t.TempDir()
-	fp, _ := os.CreateTemp(t.TempDir(), "*.tar.gz")
+	fp, _ := os.CreateTemp(t.TempDir(), "*.tar")
 
 	blockWriter := v1.NewDirectoryBlockWriter(tmpDir)
 	err := blockWriter.Init()
 	require.NoError(t, err)
 
-	err = v1.TarGz(fp, v1.NewDirectoryBlockReader(tmpDir))
+	err = v1.Tar(fp, v1.NewDirectoryBlockReader(tmpDir))
 	require.NoError(t, err)
 
 	_, _ = fp.Seek(0, 0)