Merge branch 'main' into skiprange-6.1

pkg/storage/bloom/v1/builder.go

@@ -70,18 +70,25 @@ func NewBlockOptions(enc compression.Codec, maxBlockSizeBytes, maxBloomSizeBytes
 	opts := NewBlockOptionsFromSchema(Schema{
 		version:  CurrentSchemaVersion,
 		encoding: enc,
-	})
+	}, maxBloomSizeBytes)
 	opts.BlockSize = maxBlockSizeBytes
 	opts.UnencodedBlockOptions.MaxBloomSizeBytes = maxBloomSizeBytes
 	return opts
 }
 
-func NewBlockOptionsFromSchema(s Schema) BlockOptions {
+func NewBlockOptionsFromSchema(s Schema, maxBloomSizeBytes uint64) BlockOptions {
 	return BlockOptions{
 		Schema: s,
 		// TODO(owen-d): benchmark and find good defaults
-		SeriesPageSize: 4 << 10,   // 4KB, typical page size
-		BloomPageSize:  256 << 10, // 256KB, no idea what to make this
+		SeriesPageSize: 4 << 10, // 4KB, typical page size
+
+		// Allow one bloom page to fit either several small blooms or one large
+		// bloom at max size.
+		//
+		// Previously this value was fixed at 256KB, which is smaller than most
+		// blooms. Setting this value less than maxBloomSizeBytes means that most
+		// pages will consist of a single oversized bloom.
+		BloomPageSize: maxBloomSizeBytes,
 	}
 }
 

pkg/storage/bloom/v1/builder_test.go

@@ -23,6 +23,22 @@ var blockEncodings = []compression.Codec{
 	compression.Zstd,
 }
 
+func TestBlockOptions_BloomPageSize(t *testing.T) {
+	t.Parallel()
+
+	var (
+		maxBlockSizeBytes = uint64(50 << 10)
+		maxBloomSizeBytes = uint64(10 << 10)
+	)
+
+	opts := NewBlockOptions(compression.None, maxBlockSizeBytes, maxBloomSizeBytes)
+
+	require.GreaterOrEqual(
+		t, opts.BloomPageSize, maxBloomSizeBytes,
+		"opts.BloomPageSize should be greater or equal to the maximum bloom size to avoid having too many overfilled pages",
+	)
+}
+
 func TestBlockOptions_RoundTrip(t *testing.T) {
 	t.Parallel()
 	opts := BlockOptions{