Manish R Jain | 02eca14 | 2018-12-30 19:35:11 | [diff] [blame] | 1 | /* |
| 2 | * Copyright 2018 Dgraph Labs, Inc. and Contributors |
| 3 | * |
| 4 | * Licensed under the Apache License, Version 2.0 (the "License"); |
| 5 | * you may not use this file except in compliance with the License. |
| 6 | * You may obtain a copy of the License at |
| 7 | * |
| 8 | * http://www.apache.org/licenses/LICENSE-2.0 |
| 9 | * |
| 10 | * Unless required by applicable law or agreed to in writing, software |
| 11 | * distributed under the License is distributed on an "AS IS" BASIS, |
| 12 | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| 13 | * See the License for the specific language governing permissions and |
| 14 | * limitations under the License. |
| 15 | */ |
| 16 | |
| 17 | package badger |
| 18 | |
| 19 | import ( |
| 20 | "context" |
| 21 | "fmt" |
| 22 | "io/ioutil" |
| 23 | "math" |
Manish R Jain | 02eca14 | 2018-12-30 19:35:11 | [diff] [blame] | 24 | "strconv" |
| 25 | "strings" |
| 26 | "testing" |
| 27 | |
aman bansal | b69163b | 2021-01-13 05:25:37 | [diff] [blame] | 28 | "github.com/dgraph-io/badger/v3/pb" |
| 29 | bpb "github.com/dgraph-io/badger/v3/pb" |
| 30 | "github.com/dgraph-io/badger/v3/y" |
Manish R Jain | ab8b5d9 | 2020-11-25 18:44:31 | [diff] [blame] | 31 | "github.com/dgraph-io/ristretto/z" |
Manish R Jain | 8d26d52 | 2020-10-13 21:31:31 | [diff] [blame] | 32 | "github.com/golang/protobuf/proto" |
Manish R Jain | 02eca14 | 2018-12-30 19:35:11 | [diff] [blame] | 33 | "github.com/stretchr/testify/require" |
| 34 | ) |
| 35 | |
Manish R Jain | 02eca14 | 2018-12-30 19:35:11 | [diff] [blame] | 36 | func keyWithPrefix(prefix string, k int) []byte { |
| 37 | return []byte(fmt.Sprintf("%s-%d", prefix, k)) |
| 38 | } |
| 39 | |
| 40 | func keyToInt(k []byte) (string, int) { |
| 41 | splits := strings.Split(string(k), "-") |
| 42 | key, err := strconv.Atoi(splits[1]) |
| 43 | y.Check(err) |
| 44 | return splits[0], key |
| 45 | } |
| 46 | |
| 47 | func value(k int) []byte { |
| 48 | return []byte(fmt.Sprintf("%08d", k)) |
| 49 | } |
| 50 | |
| 51 | type collector struct { |
| 52 | kv []*bpb.KV |
| 53 | } |
| 54 | |
Manish R Jain | ab8b5d9 | 2020-11-25 18:44:31 | [diff] [blame] | 55 | func (c *collector) Send(buf *z.Buffer) error { |
| 56 | list, err := BufferToKVList(buf) |
| 57 | if err != nil { |
| 58 | return err |
| 59 | } |
Manish R Jain | 8d26d52 | 2020-10-13 21:31:31 | [diff] [blame] | 60 | for _, kv := range list.Kv { |
| 61 | if kv.StreamDone == true { |
Manish R Jain | ab8b5d9 | 2020-11-25 18:44:31 | [diff] [blame] | 62 | return nil |
Manish R Jain | 8d26d52 | 2020-10-13 21:31:31 | [diff] [blame] | 63 | } |
| 64 | cp := proto.Clone(kv).(*bpb.KV) |
| 65 | c.kv = append(c.kv, cp) |
| 66 | } |
Manish R Jain | ab8b5d9 | 2020-11-25 18:44:31 | [diff] [blame] | 67 | return err |
Manish R Jain | 02eca14 | 2018-12-30 19:35:11 | [diff] [blame] | 68 | } |
| 69 | |
| 70 | var ctxb = context.Background() |
| 71 | |
| 72 | func TestStream(t *testing.T) { |
Ashish Goswami | 8c85c27 | 2019-06-19 08:21:22 | [diff] [blame] | 73 | dir, err := ioutil.TempDir("", "badger-test") |
Manish R Jain | 02eca14 | 2018-12-30 19:35:11 | [diff] [blame] | 74 | require.NoError(t, err) |
Ibrahim Jarif | 00039ea | 2019-10-17 13:02:32 | [diff] [blame] | 75 | defer removeDir(dir) |
Manish R Jain | 02eca14 | 2018-12-30 19:35:11 | [diff] [blame] | 76 | |
Francesc Campoy | 2bb87c0 | 2019-06-24 21:57:56 | [diff] [blame] | 77 | db, err := OpenManaged(DefaultOptions(dir)) |
Manish R Jain | 02eca14 | 2018-12-30 19:35:11 | [diff] [blame] | 78 | require.NoError(t, err) |
| 79 | |
| 80 | var count int |
| 81 | for _, prefix := range []string{"p0", "p1", "p2"} { |
| 82 | txn := db.NewTransactionAt(math.MaxUint64, true) |
| 83 | for i := 1; i <= 100; i++ { |
Ashish Goswami | e9447c9 | 2019-05-28 05:28:37 | [diff] [blame] | 84 | require.NoError(t, txn.SetEntry(NewEntry(keyWithPrefix(prefix, i), value(i)))) |
Manish R Jain | 02eca14 | 2018-12-30 19:35:11 | [diff] [blame] | 85 | count++ |
| 86 | } |
| 87 | require.NoError(t, txn.CommitAt(5, nil)) |
| 88 | } |
| 89 | |
| 90 | stream := db.NewStreamAt(math.MaxUint64) |
| 91 | stream.LogPrefix = "Testing" |
| 92 | c := &collector{} |
Martin Martinez Rivera | c892251 | 2020-07-13 18:29:38 | [diff] [blame] | 93 | stream.Send = c.Send |
Manish R Jain | 02eca14 | 2018-12-30 19:35:11 | [diff] [blame] | 94 | |
| 95 | // Test case 1. Retrieve everything. |
| 96 | err = stream.Orchestrate(ctxb) |
| 97 | require.NoError(t, err) |
| 98 | require.Equal(t, 300, len(c.kv), "Expected 300. Got: %d", len(c.kv)) |
| 99 | |
| 100 | m := make(map[string]int) |
| 101 | for _, kv := range c.kv { |
| 102 | prefix, ki := keyToInt(kv.Key) |
| 103 | expected := value(ki) |
| 104 | require.Equal(t, expected, kv.Value) |
| 105 | m[prefix]++ |
| 106 | } |
| 107 | require.Equal(t, 3, len(m)) |
| 108 | for pred, count := range m { |
| 109 | require.Equal(t, 100, count, "Count mismatch for pred: %s", pred) |
| 110 | } |
| 111 | |
| 112 | // Test case 2. Retrieve only 1 predicate. |
| 113 | stream.Prefix = []byte("p1") |
| 114 | c.kv = c.kv[:0] |
| 115 | err = stream.Orchestrate(ctxb) |
| 116 | require.NoError(t, err) |
| 117 | require.Equal(t, 100, len(c.kv), "Expected 100. Got: %d", len(c.kv)) |
| 118 | |
| 119 | m = make(map[string]int) |
| 120 | for _, kv := range c.kv { |
| 121 | prefix, ki := keyToInt(kv.Key) |
| 122 | expected := value(ki) |
| 123 | require.Equal(t, expected, kv.Value) |
| 124 | m[prefix]++ |
| 125 | } |
| 126 | require.Equal(t, 1, len(m)) |
| 127 | for pred, count := range m { |
| 128 | require.Equal(t, 100, count, "Count mismatch for pred: %s", pred) |
| 129 | } |
| 130 | |
| 131 | // Test case 3. Retrieve select keys within the predicate. |
| 132 | c.kv = c.kv[:0] |
| 133 | stream.ChooseKey = func(item *Item) bool { |
| 134 | _, k := keyToInt(item.Key()) |
| 135 | return k%2 == 0 |
| 136 | } |
| 137 | err = stream.Orchestrate(ctxb) |
| 138 | require.NoError(t, err) |
| 139 | require.Equal(t, 50, len(c.kv), "Expected 50. Got: %d", len(c.kv)) |
| 140 | |
| 141 | m = make(map[string]int) |
| 142 | for _, kv := range c.kv { |
| 143 | prefix, ki := keyToInt(kv.Key) |
| 144 | expected := value(ki) |
| 145 | require.Equal(t, expected, kv.Value) |
| 146 | m[prefix]++ |
| 147 | } |
| 148 | require.Equal(t, 1, len(m)) |
| 149 | for pred, count := range m { |
| 150 | require.Equal(t, 50, count, "Count mismatch for pred: %s", pred) |
| 151 | } |
| 152 | |
| 153 | // Test case 4. Retrieve select keys from all predicates. |
| 154 | c.kv = c.kv[:0] |
| 155 | stream.Prefix = []byte{} |
| 156 | err = stream.Orchestrate(ctxb) |
| 157 | require.NoError(t, err) |
| 158 | require.Equal(t, 150, len(c.kv), "Expected 150. Got: %d", len(c.kv)) |
| 159 | |
| 160 | m = make(map[string]int) |
| 161 | for _, kv := range c.kv { |
| 162 | prefix, ki := keyToInt(kv.Key) |
| 163 | expected := value(ki) |
| 164 | require.Equal(t, expected, kv.Value) |
| 165 | m[prefix]++ |
| 166 | } |
| 167 | require.Equal(t, 3, len(m)) |
| 168 | for pred, count := range m { |
| 169 | require.Equal(t, 50, count, "Count mismatch for pred: %s", pred) |
| 170 | } |
Ibrahim Jarif | 00039ea | 2019-10-17 13:02:32 | [diff] [blame] | 171 | require.NoError(t, db.Close()) |
Manish R Jain | 02eca14 | 2018-12-30 19:35:11 | [diff] [blame] | 172 | } |
Martin Martinez Rivera | 6eaa500 | 2020-05-22 17:45:26 | [diff] [blame] | 173 | |
| 174 | func TestStreamWithThreadId(t *testing.T) { |
| 175 | dir, err := ioutil.TempDir("", "badger-test") |
| 176 | require.NoError(t, err) |
| 177 | defer removeDir(dir) |
| 178 | |
| 179 | db, err := OpenManaged(DefaultOptions(dir)) |
| 180 | require.NoError(t, err) |
| 181 | |
| 182 | var count int |
| 183 | for _, prefix := range []string{"p0", "p1", "p2"} { |
| 184 | txn := db.NewTransactionAt(math.MaxUint64, true) |
| 185 | for i := 1; i <= 100; i++ { |
| 186 | require.NoError(t, txn.SetEntry(NewEntry(keyWithPrefix(prefix, i), value(i)))) |
| 187 | count++ |
| 188 | } |
| 189 | require.NoError(t, txn.CommitAt(5, nil)) |
| 190 | } |
| 191 | |
| 192 | stream := db.NewStreamAt(math.MaxUint64) |
| 193 | stream.LogPrefix = "Testing" |
| 194 | stream.KeyToList = func(key []byte, itr *Iterator) ( |
| 195 | *bpb.KVList, error) { |
| 196 | require.Less(t, itr.ThreadId, stream.NumGo) |
| 197 | return stream.ToList(key, itr) |
| 198 | } |
| 199 | c := &collector{} |
Martin Martinez Rivera | c892251 | 2020-07-13 18:29:38 | [diff] [blame] | 200 | stream.Send = c.Send |
Martin Martinez Rivera | 6eaa500 | 2020-05-22 17:45:26 | [diff] [blame] | 201 | |
| 202 | err = stream.Orchestrate(ctxb) |
| 203 | require.NoError(t, err) |
| 204 | require.Equal(t, 300, len(c.kv), "Expected 300. Got: %d", len(c.kv)) |
| 205 | |
| 206 | m := make(map[string]int) |
| 207 | for _, kv := range c.kv { |
| 208 | prefix, ki := keyToInt(kv.Key) |
| 209 | expected := value(ki) |
| 210 | require.Equal(t, expected, kv.Value) |
| 211 | m[prefix]++ |
| 212 | } |
| 213 | require.Equal(t, 3, len(m)) |
| 214 | for pred, count := range m { |
| 215 | require.Equal(t, 100, count, "Count mismatch for pred: %s", pred) |
| 216 | } |
| 217 | require.NoError(t, db.Close()) |
| 218 | } |
Martin Martinez Rivera | c892251 | 2020-07-13 18:29:38 | [diff] [blame] | 219 | |
| 220 | func TestBigStream(t *testing.T) { |
Rahul Gurnani | 774a99d | 2020-11-02 23:39:09 | [diff] [blame] | 221 | if !*manual { |
| 222 | t.Skip("Skipping test meant to be run manually.") |
| 223 | return |
| 224 | } |
Martin Martinez Rivera | c892251 | 2020-07-13 18:29:38 | [diff] [blame] | 225 | // Set the maxStreamSize to 1MB for the duration of the test so that the it can use a smaller |
| 226 | // dataset than it would otherwise need. |
| 227 | originalMaxStreamSize := maxStreamSize |
| 228 | maxStreamSize = 1 << 20 |
| 229 | defer func() { |
| 230 | maxStreamSize = originalMaxStreamSize |
| 231 | }() |
| 232 | |
| 233 | testSize := int(1e6) |
| 234 | dir, err := ioutil.TempDir("", "badger-big-test") |
| 235 | require.NoError(t, err) |
| 236 | defer removeDir(dir) |
| 237 | |
| 238 | db, err := OpenManaged(DefaultOptions(dir)) |
| 239 | require.NoError(t, err) |
| 240 | |
| 241 | var count int |
| 242 | wb := db.NewWriteBatchAt(5) |
| 243 | for _, prefix := range []string{"p0", "p1", "p2"} { |
| 244 | for i := 1; i <= testSize; i++ { |
| 245 | require.NoError(t, wb.SetEntry(NewEntry(keyWithPrefix(prefix, i), value(i)))) |
| 246 | count++ |
| 247 | } |
| 248 | } |
| 249 | require.NoError(t, wb.Flush()) |
| 250 | |
| 251 | stream := db.NewStreamAt(math.MaxUint64) |
| 252 | stream.LogPrefix = "Testing" |
| 253 | c := &collector{} |
| 254 | stream.Send = c.Send |
| 255 | |
| 256 | // Test case 1. Retrieve everything. |
| 257 | err = stream.Orchestrate(ctxb) |
| 258 | require.NoError(t, err) |
| 259 | require.Equal(t, 3*testSize, len(c.kv), "Expected 30000. Got: %d", len(c.kv)) |
| 260 | |
| 261 | m := make(map[string]int) |
| 262 | for _, kv := range c.kv { |
| 263 | prefix, ki := keyToInt(kv.Key) |
| 264 | expected := value(ki) |
| 265 | require.Equal(t, expected, kv.Value) |
| 266 | m[prefix]++ |
| 267 | } |
| 268 | require.Equal(t, 3, len(m)) |
| 269 | for pred, count := range m { |
| 270 | require.Equal(t, testSize, count, "Count mismatch for pred: %s", pred) |
| 271 | } |
| 272 | require.NoError(t, db.Close()) |
| 273 | } |
Manish R Jain | b94b561 | 2020-11-11 21:02:30 | [diff] [blame] | 274 | |
| 275 | // There was a bug in the stream writer code which would cause allocators to be |
| 276 | // freed up twice if the default keyToList was not used. This test verifies that issue. |
| 277 | func TestStreamCustomKeyToList(t *testing.T) { |
| 278 | dir, err := ioutil.TempDir("", "badger-test") |
| 279 | require.NoError(t, err) |
| 280 | defer removeDir(dir) |
| 281 | |
| 282 | db, err := OpenManaged(DefaultOptions(dir)) |
| 283 | require.NoError(t, err) |
| 284 | |
| 285 | var count int |
| 286 | for _, key := range []string{"p0", "p1", "p2"} { |
| 287 | for i := 1; i <= 100; i++ { |
| 288 | txn := db.NewTransactionAt(math.MaxUint64, true) |
| 289 | require.NoError(t, txn.SetEntry(NewEntry([]byte(key), value(i)))) |
| 290 | count++ |
| 291 | require.NoError(t, txn.CommitAt(uint64(i), nil)) |
| 292 | } |
| 293 | } |
| 294 | |
| 295 | stream := db.NewStreamAt(math.MaxUint64) |
| 296 | stream.LogPrefix = "Testing" |
| 297 | stream.KeyToList = func(key []byte, itr *Iterator) (*pb.KVList, error) { |
| 298 | item := itr.Item() |
| 299 | val, err := item.ValueCopy(nil) |
| 300 | if err != nil { |
| 301 | return nil, err |
| 302 | } |
Manish R Jain | a75cb55 | 2020-12-04 20:59:40 | [diff] [blame] | 303 | kv := &pb.KV{ |
Manish R Jain | b94b561 | 2020-11-11 21:02:30 | [diff] [blame] | 304 | Key: y.Copy(item.Key()), |
| 305 | Value: val, |
| 306 | } |
| 307 | return &pb.KVList{ |
| 308 | Kv: []*pb.KV{kv}, |
| 309 | }, nil |
| 310 | } |
| 311 | res := map[string]struct{}{"p0": {}, "p1": {}, "p2": {}} |
Manish R Jain | ab8b5d9 | 2020-11-25 18:44:31 | [diff] [blame] | 312 | stream.Send = func(buf *z.Buffer) error { |
| 313 | list, err := BufferToKVList(buf) |
| 314 | require.NoError(t, err) |
Manish R Jain | b94b561 | 2020-11-11 21:02:30 | [diff] [blame] | 315 | for _, kv := range list.Kv { |
| 316 | key := string(kv.Key) |
| 317 | if _, ok := res[key]; !ok { |
| 318 | panic(fmt.Sprintf("%s key not found", key)) |
| 319 | } |
| 320 | delete(res, key) |
| 321 | } |
| 322 | return nil |
| 323 | } |
| 324 | require.NoError(t, stream.Orchestrate(ctxb)) |
| 325 | require.Zero(t, len(res)) |
| 326 | } |