Like Prometheus, but for logs.
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 
 
 
 
loki/pkg/dataset/array/codec_zstd_test.go

301 lines
8.0 KiB

package array_test
import (
"errors"
"fmt"
"io"
"math/rand"
"testing"
"github.com/stretchr/testify/require"
"github.com/grafana/loki/v3/pkg/columnar"
"github.com/grafana/loki/v3/pkg/columnar/columnartest"
"github.com/grafana/loki/v3/pkg/columnar/types"
"github.com/grafana/loki/v3/pkg/dataset/array"
"github.com/grafana/loki/v3/pkg/dataset/buffer"
"github.com/grafana/loki/v3/pkg/memory"
)
func TestZstdCodec_Validation(t *testing.T) {
tt := []struct {
name string
spec array.Spec
typ types.Type
expectError bool
}{
{
name: "accepts valid non-nullable utf8",
spec: &array.SpecZstd{Offsets: &array.SpecPlain{}},
typ: &types.UTF8{Nullable: false},
expectError: false,
},
{
name: "rejects missing offsets spec",
spec: &array.SpecZstd{Offsets: nil},
typ: &types.UTF8{Nullable: false},
expectError: true,
},
{
name: "rejects non-nullable utf8 with validity spec",
spec: &array.SpecZstd{Offsets: &array.SpecPlain{}, Validity: &array.SpecBool{}},
typ: &types.UTF8{Nullable: false},
expectError: true,
},
{
name: "rejects nullable utf8 with no validity spec",
spec: &array.SpecZstd{Offsets: &array.SpecPlain{}},
typ: &types.UTF8{Nullable: true},
expectError: true,
},
{
name: "accepts nullable utf8 with validity spec",
spec: &array.SpecZstd{Offsets: &array.SpecPlain{}, Validity: &array.SpecBool{}},
typ: &types.UTF8{Nullable: true},
expectError: false,
},
{
name: "rejects unsupported type bool",
spec: &array.SpecZstd{Offsets: &array.SpecPlain{}},
typ: &types.Bool{Nullable: false},
expectError: true,
},
{
name: "rejects unsupported type int64",
spec: &array.SpecZstd{Offsets: &array.SpecPlain{}},
typ: &types.Int64{Nullable: false},
expectError: true,
},
}
for _, tc := range tt {
t.Run(tc.name, func(t *testing.T) {
var alloc memory.Allocator
_, err := array.NewWriter(&alloc, tc.spec, tc.typ)
if tc.expectError {
require.Error(t, err)
} else {
require.NoError(t, err)
}
})
}
}
func TestZstdCodec_NonNullable(t *testing.T) {
var alloc memory.Allocator
var store buffer.MemoryStore
var (
spec = &array.SpecZstd{Offsets: &array.SpecPlain{}}
typ = &types.UTF8{Nullable: false}
)
w, err := array.NewWriter(&alloc, spec, typ)
require.NoError(t, err)
input := []columnar.Array{
columnartest.Array(t, types.KindUTF8, &alloc, "hello", "world", "", "foo", "bar"),
columnartest.Array(t, types.KindUTF8, &alloc, "baz", "qux", "quux", "corge", "grault"),
}
for _, arr := range input {
require.NoError(t, w.Append(arr))
}
result, err := w.Flush(t.Context(), &store)
require.NoError(t, err)
require.Equal(t, 10, result.RowCount)
require.Equal(t, 0, result.Stats.NullCount)
r, err := array.NewReader(&alloc, result, &store)
require.NoError(t, err)
expect := columnartest.Array(
t, types.KindUTF8, &alloc,
"hello", "world", "", "foo", "bar",
"baz", "qux", "quux", "corge", "grault",
)
actual := readBatches(t, &alloc, r, 2) // Read in a small batch size to test reading multiple times
columnartest.RequireArraysEqual(t, expect, actual, memory.Bitmap{})
// Reading again should produce a EOF.
_, err = r.Read(t.Context(), &alloc, 1)
require.ErrorIs(t, err, io.EOF)
}
func TestZstdCodec_Nullable(t *testing.T) {
var alloc memory.Allocator
var store buffer.MemoryStore
var (
spec = &array.SpecZstd{Offsets: &array.SpecPlain{}, Validity: &array.SpecBool{}}
typ = &types.UTF8{Nullable: true}
)
w, err := array.NewWriter(&alloc, spec, typ)
require.NoError(t, err)
input := []columnar.Array{
columnartest.Array(t, types.KindUTF8, &alloc, "hello", nil, "world", nil, "foo"),
columnartest.Array(t, types.KindUTF8, &alloc, "bar", "baz", "qux", "quux", "corge"),
columnartest.Array(t, types.KindUTF8, &alloc, nil),
}
for _, arr := range input {
require.NoError(t, w.Append(arr))
}
result, err := w.Flush(t.Context(), &store)
require.NoError(t, err)
require.Equal(t, 11, result.RowCount)
require.Equal(t, 3, result.Stats.NullCount)
r, err := array.NewReader(&alloc, result, &store)
require.NoError(t, err)
expect := columnartest.Array(
t, types.KindUTF8, &alloc,
"hello", nil, "world", nil, "foo",
"bar", "baz", "qux", "quux", "corge",
nil,
)
actual := readBatches(t, &alloc, r, 2) // Read in a small batch size to test reading multiple times
columnartest.RequireArraysEqual(t, expect, actual, memory.Bitmap{})
// Reading again should produce a EOF.
_, err = r.Read(t.Context(), &alloc, 1)
require.ErrorIs(t, err, io.EOF)
}
func TestZstdCodec_SlicedInput(t *testing.T) {
var alloc memory.Allocator
var store buffer.MemoryStore
var (
spec = &array.SpecZstd{Offsets: &array.SpecPlain{}}
typ = &types.UTF8{Nullable: false}
)
// Build a full array then slice out a middle section, so the input has
// non-zero-based offsets and a shared data buffer.
full := columnartest.Array(t, types.KindUTF8, &alloc,
"alpha", "bravo", "charlie", "delta", "echo",
)
sliced := full.Slice(1, 4) // ["bravo", "charlie", "delta"]
w, err := array.NewWriter(&alloc, spec, typ)
require.NoError(t, err)
require.NoError(t, w.Append(sliced))
result, err := w.Flush(t.Context(), &store)
require.NoError(t, err)
require.Equal(t, 3, result.RowCount)
r, err := array.NewReader(&alloc, result, &store)
require.NoError(t, err)
expect := columnartest.Array(t, types.KindUTF8, &alloc,
"bravo", "charlie", "delta",
)
actual := readBatches(t, &alloc, r, 2) // Read in a small batch size to test reading multiple times
columnartest.RequireArraysEqual(t, expect, actual, memory.Bitmap{})
}
func BenchmarkZstdCodec(b *testing.B) {
var (
store buffer.MemoryStore
spec = &array.SpecZstd{Offsets: &array.SpecPlain{}}
typ = &types.UTF8{Nullable: false}
)
const valuesPerPage = 1 << 16
type scenario struct {
name string
valueCount int
encoded array.Array
}
build := func(name string, valueCount int, valueAt func(i int) []byte) scenario {
var alloc memory.Allocator
w, err := array.NewWriter(&alloc, spec, typ)
require.NoError(b, err)
builder := columnar.NewUTF8Builder(&alloc)
builder.Grow(valueCount)
for i := range valueCount {
builder.AppendValue(valueAt(i))
}
require.NoError(b, w.Append(builder.Build()))
arr, err := w.Flush(b.Context(), &store)
require.NoError(b, err)
return scenario{name: name, valueCount: valueCount, encoded: arr}
}
scenarios := []scenario{
build("variance=constant", valuesPerPage, func(int) []byte { return []byte("hello") }),
func() scenario {
rnd := rand.New(rand.NewSource(0))
return build("variance=random", valuesPerPage, func(int) []byte {
buf := make([]byte, 5+rnd.Intn(20))
rnd.Read(buf)
return buf
})
}(),
}
batchSizes := []int{256, 1024, 4096}
for _, sc := range scenarios {
b.Run(sc.name, func(b *testing.B) {
b.Run(fmt.Sprintf("values_per_page=%d", sc.valueCount), func(b *testing.B) {
for _, batchSize := range batchSizes {
b.Run(fmt.Sprintf("batch_size=%d", batchSize), func(b *testing.B) {
var alloc memory.Allocator
b.ReportAllocs()
// Approximate bytes: data + offsets
decodedBytesPerOp := int64(sc.encoded.RowCount) * 15 // ~15 bytes avg string
b.SetBytes(decodedBytesPerOp)
for b.Loop() {
alloc.Reset()
r, _ := array.NewReader(&alloc, sc.encoded, &store)
var decoded int
for {
arr, err := r.Read(b.Context(), &alloc, batchSize)
if arr != nil {
decoded += arr.Len()
}
if errors.Is(err, io.EOF) {
break
} else if err != nil {
b.Fatal(err)
}
}
if decoded != sc.valueCount {
b.Fatalf("decoded %d values, expected %d", decoded, sc.valueCount)
}
}
elapsed := b.Elapsed()
if elapsed > 0 {
totalDecoded := int64(sc.valueCount) * int64(b.N)
b.ReportMetric(float64(totalDecoded)/elapsed.Seconds(), "rows/s")
}
})
}
})
})
}
}