github.com/chain5j/chain5j-pkg@v1.0.7/collection/trees/tree/sync_test.go (about) 1 // Copyright 2015 The go-ethereum Authors 2 // This file is part of the go-ethereum library. 3 // 4 // The go-ethereum library is free software: you can redistribute it and/or modify 5 // it under the terms of the GNU Lesser General Public License as published by 6 // the Free Software Foundation, either version 3 of the License, or 7 // (at your option) any later version. 8 // 9 // The go-ethereum library is distributed in the hope that it will be useful, 10 // but WITHOUT ANY WARRANTY; without even the implied warranty of 11 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 12 // GNU Lesser General Public License for more details. 13 // 14 // You should have received a copy of the GNU Lesser General Public License 15 // along with the go-ethereum library. If not, see <http://www.gnu.org/licenses/>. 16 17 package tree 18 19 import ( 20 "bytes" 21 "testing" 22 23 "github.com/chain5j/chain5j-pkg/database/kvstore/memorydb" 24 "github.com/chain5j/chain5j-pkg/types" 25 "github.com/chain5j/chain5j-pkg/util/hexutil" 26 ) 27 28 // makeTestTrie create a sample test trie to test node-wise reconstruction. 29 func makeTestTrie() (*Database, *Trie, map[string][]byte) { 30 // Create an empty trie 31 triedb := NewDatabase(memorydb.New()) 32 trie, _ := New(types.Hash{}, triedb) 33 34 // Fill it with some arbitrary data 35 content := make(map[string][]byte) 36 for i := byte(0); i < 255; i++ { 37 // Map the same data under multiple keys 38 key, val := hexutil.LeftPadBytes([]byte{1, i}, 32), []byte{i} 39 content[string(key)] = val 40 trie.Update(key, val) 41 42 key, val = hexutil.LeftPadBytes([]byte{2, i}, 32), []byte{i} 43 content[string(key)] = val 44 trie.Update(key, val) 45 46 // Add some other data to inflate the trie 47 for j := byte(3); j < 13; j++ { 48 key, val = hexutil.LeftPadBytes([]byte{j, i}, 32), []byte{j, i} 49 content[string(key)] = val 50 trie.Update(key, val) 51 } 52 } 53 trie.Commit(nil) 54 55 // Return the generated trie 56 return triedb, trie, content 57 } 58 59 // checkTrieContents cross references a reconstructed trie with an expected data 60 // content map. 61 func checkTrieContents(t *testing.T, db *Database, root []byte, content map[string][]byte) { 62 // Check root availability and trie contents 63 trie, err := New(types.BytesToHash(root), db) 64 if err != nil { 65 t.Fatalf("failed to create trie at %x: %v", root, err) 66 } 67 if err := checkTrieConsistency(db, types.BytesToHash(root)); err != nil { 68 t.Fatalf("inconsistent trie at %x: %v", root, err) 69 } 70 for key, val := range content { 71 if have := trie.Get([]byte(key)); !bytes.Equal(have, val) { 72 t.Errorf("entry %x: content mismatch: have %x, want %x", key, have, val) 73 } 74 } 75 } 76 77 // checkTrieConsistency checks that all nodes in a trie are indeed present. 78 func checkTrieConsistency(db *Database, root types.Hash) error { 79 // Create and iterate a trie rooted in a subnode 80 trie, err := New(root, db) 81 if err != nil { 82 return nil // Consider a non existent state consistent 83 } 84 it := trie.NodeIterator(nil) 85 for it.Next(true) { 86 } 87 return it.Error() 88 } 89 90 // Tests that an empty trie is not scheduled for syncing. 91 func TestEmptySync(t *testing.T) { 92 dbA := NewDatabase(memorydb.New()) 93 dbB := NewDatabase(memorydb.New()) 94 emptyA, _ := New(types.Hash{}, dbA) 95 emptyB, _ := New(emptyRoot, dbB) 96 97 for i, trie := range []*Trie{emptyA, emptyB} { 98 if req := NewSync(trie.Hash(), memorydb.New(), nil, NewSyncBloom(1, memorydb.New())).Missing(1); len(req) != 0 { 99 t.Errorf("test %d: content requested for empty trie: %v", i, req) 100 } 101 } 102 } 103 104 // Tests that given a root hash, a trie can sync iteratively on a single thread, 105 // requesting retrieval tasks and returning all of them in one go. 106 func TestIterativeSyncIndividual(t *testing.T) { testIterativeSync(t, 1) } 107 func TestIterativeSyncBatched(t *testing.T) { testIterativeSync(t, 100) } 108 109 func testIterativeSync(t *testing.T, count int) { 110 // Create a random trie to copy 111 srcDb, srcTrie, srcData := makeTestTrie() 112 113 // Create a destination trie and sync with the scheduler 114 diskdb := memorydb.New() 115 triedb := NewDatabase(diskdb) 116 sched := NewSync(srcTrie.Hash(), diskdb, nil, NewSyncBloom(1, diskdb)) 117 118 queue := append([]types.Hash{}, sched.Missing(count)...) 119 for len(queue) > 0 { 120 results := make([]SyncResult, len(queue)) 121 for i, hash := range queue { 122 data, err := srcDb.Node(hash) 123 if err != nil { 124 t.Fatalf("failed to retrieve node data for %x: %v", hash, err) 125 } 126 results[i] = SyncResult{hash, data} 127 } 128 if _, index, err := sched.Process(results); err != nil { 129 t.Fatalf("failed to process result #%d: %v", index, err) 130 } 131 batch := diskdb.NewBatch() 132 if err := sched.Commit(batch); err != nil { 133 t.Fatalf("failed to commit data: %v", err) 134 } 135 batch.Write() 136 queue = append(queue[:0], sched.Missing(count)...) 137 } 138 // Cross check that the two tries are in sync 139 checkTrieContents(t, triedb, srcTrie.Hash().Bytes(), srcData) 140 } 141 142 // Tests that the trie scheduler can correctly reconstruct the state even if only 143 // partial results are returned, and the others sent only later. 144 func TestIterativeDelayedSync(t *testing.T) { 145 // Create a random trie to copy 146 srcDb, srcTrie, srcData := makeTestTrie() 147 148 // Create a destination trie and sync with the scheduler 149 diskdb := memorydb.New() 150 triedb := NewDatabase(diskdb) 151 sched := NewSync(srcTrie.Hash(), diskdb, nil, NewSyncBloom(1, diskdb)) 152 153 queue := append([]types.Hash{}, sched.Missing(10000)...) 154 for len(queue) > 0 { 155 // Sync only half of the scheduled nodes 156 results := make([]SyncResult, len(queue)/2+1) 157 for i, hash := range queue[:len(results)] { 158 data, err := srcDb.Node(hash) 159 if err != nil { 160 t.Fatalf("failed to retrieve node data for %x: %v", hash, err) 161 } 162 results[i] = SyncResult{hash, data} 163 } 164 if _, index, err := sched.Process(results); err != nil { 165 t.Fatalf("failed to process result #%d: %v", index, err) 166 } 167 batch := diskdb.NewBatch() 168 if err := sched.Commit(batch); err != nil { 169 t.Fatalf("failed to commit data: %v", err) 170 } 171 batch.Write() 172 queue = append(queue[len(results):], sched.Missing(10000)...) 173 } 174 // Cross check that the two tries are in sync 175 checkTrieContents(t, triedb, srcTrie.Hash().Bytes(), srcData) 176 } 177 178 // Tests that given a root hash, a trie can sync iteratively on a single thread, 179 // requesting retrieval tasks and returning all of them in one go, however in a 180 // random order. 181 func TestIterativeRandomSyncIndividual(t *testing.T) { testIterativeRandomSync(t, 1) } 182 func TestIterativeRandomSyncBatched(t *testing.T) { testIterativeRandomSync(t, 100) } 183 184 func testIterativeRandomSync(t *testing.T, count int) { 185 // Create a random trie to copy 186 srcDb, srcTrie, srcData := makeTestTrie() 187 188 // Create a destination trie and sync with the scheduler 189 diskdb := memorydb.New() 190 triedb := NewDatabase(diskdb) 191 sched := NewSync(srcTrie.Hash(), diskdb, nil, NewSyncBloom(1, diskdb)) 192 193 queue := make(map[types.Hash]struct{}) 194 for _, hash := range sched.Missing(count) { 195 queue[hash] = struct{}{} 196 } 197 for len(queue) > 0 { 198 // Fetch all the queued nodes in a random order 199 results := make([]SyncResult, 0, len(queue)) 200 for hash := range queue { 201 data, err := srcDb.Node(hash) 202 if err != nil { 203 t.Fatalf("failed to retrieve node data for %x: %v", hash, err) 204 } 205 results = append(results, SyncResult{hash, data}) 206 } 207 // Feed the retrieved results back and queue new tasks 208 if _, index, err := sched.Process(results); err != nil { 209 t.Fatalf("failed to process result #%d: %v", index, err) 210 } 211 batch := diskdb.NewBatch() 212 if err := sched.Commit(batch); err != nil { 213 t.Fatalf("failed to commit data: %v", err) 214 } 215 batch.Write() 216 queue = make(map[types.Hash]struct{}) 217 for _, hash := range sched.Missing(count) { 218 queue[hash] = struct{}{} 219 } 220 } 221 // Cross check that the two tries are in sync 222 checkTrieContents(t, triedb, srcTrie.Hash().Bytes(), srcData) 223 } 224 225 // Tests that the trie scheduler can correctly reconstruct the state even if only 226 // partial results are returned (Even those randomly), others sent only later. 227 func TestIterativeRandomDelayedSync(t *testing.T) { 228 // Create a random trie to copy 229 srcDb, srcTrie, srcData := makeTestTrie() 230 231 // Create a destination trie and sync with the scheduler 232 diskdb := memorydb.New() 233 triedb := NewDatabase(diskdb) 234 sched := NewSync(srcTrie.Hash(), diskdb, nil, NewSyncBloom(1, diskdb)) 235 236 queue := make(map[types.Hash]struct{}) 237 for _, hash := range sched.Missing(10000) { 238 queue[hash] = struct{}{} 239 } 240 for len(queue) > 0 { 241 // Sync only half of the scheduled nodes, even those in random order 242 results := make([]SyncResult, 0, len(queue)/2+1) 243 for hash := range queue { 244 data, err := srcDb.Node(hash) 245 if err != nil { 246 t.Fatalf("failed to retrieve node data for %x: %v", hash, err) 247 } 248 results = append(results, SyncResult{hash, data}) 249 250 if len(results) >= cap(results) { 251 break 252 } 253 } 254 // Feed the retrieved results back and queue new tasks 255 if _, index, err := sched.Process(results); err != nil { 256 t.Fatalf("failed to process result #%d: %v", index, err) 257 } 258 batch := diskdb.NewBatch() 259 if err := sched.Commit(batch); err != nil { 260 t.Fatalf("failed to commit data: %v", err) 261 } 262 batch.Write() 263 for _, result := range results { 264 delete(queue, result.Hash) 265 } 266 for _, hash := range sched.Missing(10000) { 267 queue[hash] = struct{}{} 268 } 269 } 270 // Cross check that the two tries are in sync 271 checkTrieContents(t, triedb, srcTrie.Hash().Bytes(), srcData) 272 } 273 274 // Tests that a trie sync will not request nodes multiple times, even if they 275 // have such references. 276 func TestDuplicateAvoidanceSync(t *testing.T) { 277 // Create a random trie to copy 278 srcDb, srcTrie, srcData := makeTestTrie() 279 280 // Create a destination trie and sync with the scheduler 281 diskdb := memorydb.New() 282 triedb := NewDatabase(diskdb) 283 sched := NewSync(srcTrie.Hash(), diskdb, nil, NewSyncBloom(1, diskdb)) 284 285 queue := append([]types.Hash{}, sched.Missing(0)...) 286 requested := make(map[types.Hash]struct{}) 287 288 for len(queue) > 0 { 289 results := make([]SyncResult, len(queue)) 290 for i, hash := range queue { 291 data, err := srcDb.Node(hash) 292 if err != nil { 293 t.Fatalf("failed to retrieve node data for %x: %v", hash, err) 294 } 295 if _, ok := requested[hash]; ok { 296 t.Errorf("hash %x already requested once", hash) 297 } 298 requested[hash] = struct{}{} 299 300 results[i] = SyncResult{hash, data} 301 } 302 if _, index, err := sched.Process(results); err != nil { 303 t.Fatalf("failed to process result #%d: %v", index, err) 304 } 305 batch := diskdb.NewBatch() 306 if err := sched.Commit(batch); err != nil { 307 t.Fatalf("failed to commit data: %v", err) 308 } 309 batch.Write() 310 queue = append(queue[:0], sched.Missing(0)...) 311 } 312 // Cross check that the two tries are in sync 313 checkTrieContents(t, triedb, srcTrie.Hash().Bytes(), srcData) 314 } 315 316 // Tests that at any point in time during a sync, only complete sub-tries are in 317 // the database. 318 func TestIncompleteSync(t *testing.T) { 319 // Create a random trie to copy 320 srcDb, srcTrie, _ := makeTestTrie() 321 322 // Create a destination trie and sync with the scheduler 323 diskdb := memorydb.New() 324 triedb := NewDatabase(diskdb) 325 sched := NewSync(srcTrie.Hash(), diskdb, nil, NewSyncBloom(1, diskdb)) 326 327 var added []types.Hash 328 queue := append([]types.Hash{}, sched.Missing(1)...) 329 for len(queue) > 0 { 330 // Fetch a batch of trie nodes 331 results := make([]SyncResult, len(queue)) 332 for i, hash := range queue { 333 data, err := srcDb.Node(hash) 334 if err != nil { 335 t.Fatalf("failed to retrieve node data for %x: %v", hash, err) 336 } 337 results[i] = SyncResult{hash, data} 338 } 339 // Process each of the trie nodes 340 if _, index, err := sched.Process(results); err != nil { 341 t.Fatalf("failed to process result #%d: %v", index, err) 342 } 343 batch := diskdb.NewBatch() 344 if err := sched.Commit(batch); err != nil { 345 t.Fatalf("failed to commit data: %v", err) 346 } 347 batch.Write() 348 for _, result := range results { 349 added = append(added, result.Hash) 350 } 351 // Check that all known sub-tries in the synced trie are complete 352 for _, root := range added { 353 if err := checkTrieConsistency(triedb, root); err != nil { 354 t.Fatalf("trie inconsistent: %v", err) 355 } 356 } 357 // Fetch the next batch to retrieve 358 queue = append(queue[:0], sched.Missing(1)...) 359 } 360 // Sanity check that removing any node from the database is detected 361 for _, node := range added[1:] { 362 key := node.Bytes() 363 value, _ := diskdb.Get(key) 364 365 diskdb.Delete(key) 366 if err := checkTrieConsistency(triedb, added[0]); err == nil { 367 t.Fatalf("trie inconsistency not caught, missing: %x", key) 368 } 369 diskdb.Put(key, value) 370 } 371 }