]>
Commit | Line | Data |
---|---|---|
7c673cae | 1 | // Copyright (c) 2011-present, Facebook, Inc. All rights reserved. |
11fdf7f2 TL |
2 | // This source code is licensed under both the GPLv2 (found in the |
3 | // COPYING file in the root directory) and Apache 2.0 License | |
4 | // (found in the LICENSE.Apache file in the root directory). | |
7c673cae FG |
5 | |
6 | #include <functional> | |
7 | ||
8 | #include "db/db_test_util.h" | |
20effc67 | 9 | #include "db/version_edit.h" |
7c673cae FG |
10 | #include "port/port.h" |
11 | #include "port/stack_trace.h" | |
12 | #include "rocksdb/sst_file_writer.h" | |
1e59de90 | 13 | #include "test_util/testharness.h" |
f67539c2 | 14 | #include "test_util/testutil.h" |
20effc67 TL |
15 | #include "util/random.h" |
16 | #include "utilities/fault_injection_env.h" | |
7c673cae | 17 | |
f67539c2 | 18 | namespace ROCKSDB_NAMESPACE { |
7c673cae FG |
19 | |
20 | #ifndef ROCKSDB_LITE | |
494da23a TL |
21 | class ExternalSSTFileBasicTest |
22 | : public DBTestBase, | |
23 | public ::testing::WithParamInterface<std::tuple<bool, bool>> { | |
7c673cae | 24 | public: |
20effc67 | 25 | ExternalSSTFileBasicTest() |
1e59de90 TL |
26 | : DBTestBase("external_sst_file_basic_test", /*env_do_fsync=*/true) { |
27 | sst_files_dir_ = dbname_ + "_sst_files/"; | |
20effc67 | 28 | fault_injection_test_env_.reset(new FaultInjectionTestEnv(env_)); |
7c673cae | 29 | DestroyAndRecreateExternalSSTFilesDir(); |
1e59de90 TL |
30 | |
31 | // Check if the Env supports RandomRWFile | |
32 | std::string file_path = sst_files_dir_ + "test_random_rw_file"; | |
33 | std::unique_ptr<WritableFile> wfile; | |
34 | assert(env_->NewWritableFile(file_path, &wfile, EnvOptions()).ok()); | |
35 | wfile.reset(); | |
36 | std::unique_ptr<RandomRWFile> rwfile; | |
37 | Status s = env_->NewRandomRWFile(file_path, &rwfile, EnvOptions()); | |
38 | if (s.IsNotSupported()) { | |
39 | random_rwfile_supported_ = false; | |
40 | } else { | |
41 | EXPECT_OK(s); | |
42 | random_rwfile_supported_ = true; | |
43 | } | |
44 | rwfile.reset(); | |
45 | EXPECT_OK(env_->DeleteFile(file_path)); | |
7c673cae FG |
46 | } |
47 | ||
48 | void DestroyAndRecreateExternalSSTFilesDir() { | |
1e59de90 TL |
49 | ASSERT_OK(DestroyDir(env_, sst_files_dir_)); |
50 | ASSERT_OK(env_->CreateDir(sst_files_dir_)); | |
7c673cae FG |
51 | } |
52 | ||
53 | Status DeprecatedAddFile(const std::vector<std::string>& files, | |
54 | bool move_files = false, | |
55 | bool skip_snapshot_check = false) { | |
56 | IngestExternalFileOptions opts; | |
57 | opts.move_files = move_files; | |
58 | opts.snapshot_consistency = !skip_snapshot_check; | |
59 | opts.allow_global_seqno = false; | |
60 | opts.allow_blocking_flush = false; | |
61 | return db_->IngestExternalFile(files, opts); | |
62 | } | |
63 | ||
20effc67 TL |
64 | Status AddFileWithFileChecksum( |
65 | const std::vector<std::string>& files, | |
66 | const std::vector<std::string>& files_checksums, | |
67 | const std::vector<std::string>& files_checksum_func_names, | |
68 | bool verify_file_checksum = true, bool move_files = false, | |
69 | bool skip_snapshot_check = false, bool write_global_seqno = true) { | |
70 | IngestExternalFileOptions opts; | |
71 | opts.move_files = move_files; | |
72 | opts.snapshot_consistency = !skip_snapshot_check; | |
73 | opts.allow_global_seqno = false; | |
74 | opts.allow_blocking_flush = false; | |
75 | opts.write_global_seqno = write_global_seqno; | |
76 | opts.verify_file_checksum = verify_file_checksum; | |
77 | ||
78 | IngestExternalFileArg arg; | |
79 | arg.column_family = db_->DefaultColumnFamily(); | |
80 | arg.external_files = files; | |
81 | arg.options = opts; | |
82 | arg.files_checksums = files_checksums; | |
83 | arg.files_checksum_func_names = files_checksum_func_names; | |
84 | return db_->IngestExternalFiles({arg}); | |
85 | } | |
86 | ||
7c673cae | 87 | Status GenerateAndAddExternalFile( |
11fdf7f2 TL |
88 | const Options options, std::vector<int> keys, |
89 | const std::vector<ValueType>& value_types, | |
90 | std::vector<std::pair<int, int>> range_deletions, int file_id, | |
494da23a | 91 | bool write_global_seqno, bool verify_checksums_before_ingest, |
7c673cae | 92 | std::map<std::string, std::string>* true_data) { |
11fdf7f2 | 93 | assert(value_types.size() == 1 || keys.size() == value_types.size()); |
1e59de90 | 94 | std::string file_path = sst_files_dir_ + std::to_string(file_id); |
7c673cae FG |
95 | SstFileWriter sst_file_writer(EnvOptions(), options); |
96 | ||
97 | Status s = sst_file_writer.Open(file_path); | |
98 | if (!s.ok()) { | |
99 | return s; | |
100 | } | |
11fdf7f2 TL |
101 | for (size_t i = 0; i < range_deletions.size(); i++) { |
102 | // Account for the effect of range deletions on true_data before | |
103 | // all point operators, even though sst_file_writer.DeleteRange | |
104 | // must be called before other sst_file_writer methods. This is | |
105 | // because point writes take precedence over range deletions | |
106 | // in the same ingested sst. | |
107 | std::string start_key = Key(range_deletions[i].first); | |
108 | std::string end_key = Key(range_deletions[i].second); | |
109 | s = sst_file_writer.DeleteRange(start_key, end_key); | |
110 | if (!s.ok()) { | |
111 | sst_file_writer.Finish(); | |
112 | return s; | |
113 | } | |
114 | auto start_key_it = true_data->find(start_key); | |
115 | if (start_key_it == true_data->end()) { | |
116 | start_key_it = true_data->upper_bound(start_key); | |
117 | } | |
118 | auto end_key_it = true_data->find(end_key); | |
119 | if (end_key_it == true_data->end()) { | |
120 | end_key_it = true_data->upper_bound(end_key); | |
121 | } | |
122 | true_data->erase(start_key_it, end_key_it); | |
123 | } | |
124 | for (size_t i = 0; i < keys.size(); i++) { | |
125 | std::string key = Key(keys[i]); | |
1e59de90 | 126 | std::string value = Key(keys[i]) + std::to_string(file_id); |
11fdf7f2 TL |
127 | ValueType value_type = |
128 | (value_types.size() == 1 ? value_types[0] : value_types[i]); | |
129 | switch (value_type) { | |
130 | case ValueType::kTypeValue: | |
131 | s = sst_file_writer.Put(key, value); | |
132 | (*true_data)[key] = value; | |
133 | break; | |
134 | case ValueType::kTypeMerge: | |
135 | s = sst_file_writer.Merge(key, value); | |
136 | // we only use TestPutOperator in this test | |
137 | (*true_data)[key] = value; | |
138 | break; | |
139 | case ValueType::kTypeDeletion: | |
140 | s = sst_file_writer.Delete(key); | |
141 | true_data->erase(key); | |
142 | break; | |
143 | default: | |
144 | return Status::InvalidArgument("Value type is not supported"); | |
145 | } | |
7c673cae FG |
146 | if (!s.ok()) { |
147 | sst_file_writer.Finish(); | |
148 | return s; | |
149 | } | |
150 | } | |
151 | s = sst_file_writer.Finish(); | |
152 | ||
153 | if (s.ok()) { | |
154 | IngestExternalFileOptions ifo; | |
155 | ifo.allow_global_seqno = true; | |
494da23a TL |
156 | ifo.write_global_seqno = write_global_seqno; |
157 | ifo.verify_checksums_before_ingest = verify_checksums_before_ingest; | |
7c673cae FG |
158 | s = db_->IngestExternalFile({file_path}, ifo); |
159 | } | |
7c673cae FG |
160 | return s; |
161 | } | |
162 | ||
11fdf7f2 TL |
163 | Status GenerateAndAddExternalFile( |
164 | const Options options, std::vector<int> keys, | |
165 | const std::vector<ValueType>& value_types, int file_id, | |
494da23a | 166 | bool write_global_seqno, bool verify_checksums_before_ingest, |
11fdf7f2 | 167 | std::map<std::string, std::string>* true_data) { |
494da23a TL |
168 | return GenerateAndAddExternalFile( |
169 | options, keys, value_types, {}, file_id, write_global_seqno, | |
170 | verify_checksums_before_ingest, true_data); | |
11fdf7f2 TL |
171 | } |
172 | ||
173 | Status GenerateAndAddExternalFile( | |
174 | const Options options, std::vector<int> keys, const ValueType value_type, | |
494da23a TL |
175 | int file_id, bool write_global_seqno, bool verify_checksums_before_ingest, |
176 | std::map<std::string, std::string>* true_data) { | |
177 | return GenerateAndAddExternalFile( | |
178 | options, keys, std::vector<ValueType>(1, value_type), file_id, | |
179 | write_global_seqno, verify_checksums_before_ingest, true_data); | |
11fdf7f2 TL |
180 | } |
181 | ||
1e59de90 TL |
182 | ~ExternalSSTFileBasicTest() override { |
183 | DestroyDir(env_, sst_files_dir_).PermitUncheckedError(); | |
184 | } | |
7c673cae FG |
185 | |
186 | protected: | |
187 | std::string sst_files_dir_; | |
f67539c2 | 188 | std::unique_ptr<FaultInjectionTestEnv> fault_injection_test_env_; |
1e59de90 | 189 | bool random_rwfile_supported_; |
7c673cae FG |
190 | }; |
191 | ||
192 | TEST_F(ExternalSSTFileBasicTest, Basic) { | |
193 | Options options = CurrentOptions(); | |
194 | ||
195 | SstFileWriter sst_file_writer(EnvOptions(), options); | |
196 | ||
197 | // Current file size should be 0 after sst_file_writer init and before open a | |
198 | // file. | |
199 | ASSERT_EQ(sst_file_writer.FileSize(), 0); | |
200 | ||
201 | // file1.sst (0 => 99) | |
202 | std::string file1 = sst_files_dir_ + "file1.sst"; | |
203 | ASSERT_OK(sst_file_writer.Open(file1)); | |
204 | for (int k = 0; k < 100; k++) { | |
11fdf7f2 | 205 | ASSERT_OK(sst_file_writer.Put(Key(k), Key(k) + "_val")); |
7c673cae FG |
206 | } |
207 | ExternalSstFileInfo file1_info; | |
208 | Status s = sst_file_writer.Finish(&file1_info); | |
1e59de90 | 209 | ASSERT_OK(s) << s.ToString(); |
7c673cae FG |
210 | |
211 | // Current file size should be non-zero after success write. | |
212 | ASSERT_GT(sst_file_writer.FileSize(), 0); | |
213 | ||
214 | ASSERT_EQ(file1_info.file_path, file1); | |
215 | ASSERT_EQ(file1_info.num_entries, 100); | |
216 | ASSERT_EQ(file1_info.smallest_key, Key(0)); | |
217 | ASSERT_EQ(file1_info.largest_key, Key(99)); | |
11fdf7f2 TL |
218 | ASSERT_EQ(file1_info.num_range_del_entries, 0); |
219 | ASSERT_EQ(file1_info.smallest_range_del_key, ""); | |
220 | ASSERT_EQ(file1_info.largest_range_del_key, ""); | |
20effc67 TL |
221 | ASSERT_EQ(file1_info.file_checksum, kUnknownFileChecksum); |
222 | ASSERT_EQ(file1_info.file_checksum_func_name, kUnknownFileChecksumFuncName); | |
223 | // sst_file_writer already finished, cannot add this value | |
224 | s = sst_file_writer.Put(Key(100), "bad_val"); | |
1e59de90 | 225 | ASSERT_NOK(s) << s.ToString(); |
20effc67 | 226 | s = sst_file_writer.DeleteRange(Key(100), Key(200)); |
1e59de90 | 227 | ASSERT_NOK(s) << s.ToString(); |
20effc67 TL |
228 | |
229 | DestroyAndReopen(options); | |
230 | // Add file using file path | |
231 | s = DeprecatedAddFile({file1}); | |
1e59de90 | 232 | ASSERT_OK(s) << s.ToString(); |
20effc67 TL |
233 | ASSERT_EQ(db_->GetLatestSequenceNumber(), 0U); |
234 | for (int k = 0; k < 100; k++) { | |
235 | ASSERT_EQ(Get(Key(k)), Key(k) + "_val"); | |
236 | } | |
237 | ||
238 | DestroyAndRecreateExternalSSTFilesDir(); | |
239 | } | |
240 | ||
241 | class ChecksumVerifyHelper { | |
242 | private: | |
243 | Options options_; | |
244 | ||
245 | public: | |
246 | ChecksumVerifyHelper(Options& options) : options_(options) {} | |
247 | ~ChecksumVerifyHelper() {} | |
248 | ||
249 | Status GetSingleFileChecksumAndFuncName( | |
250 | const std::string& file_path, std::string* file_checksum, | |
251 | std::string* file_checksum_func_name) { | |
252 | Status s; | |
253 | EnvOptions soptions; | |
254 | std::unique_ptr<SequentialFile> file_reader; | |
255 | s = options_.env->NewSequentialFile(file_path, &file_reader, soptions); | |
256 | if (!s.ok()) { | |
257 | return s; | |
258 | } | |
259 | std::unique_ptr<char[]> scratch(new char[2048]); | |
260 | Slice result; | |
261 | FileChecksumGenFactory* file_checksum_gen_factory = | |
262 | options_.file_checksum_gen_factory.get(); | |
263 | if (file_checksum_gen_factory == nullptr) { | |
264 | *file_checksum = kUnknownFileChecksum; | |
265 | *file_checksum_func_name = kUnknownFileChecksumFuncName; | |
266 | return Status::OK(); | |
267 | } else { | |
268 | FileChecksumGenContext gen_context; | |
269 | std::unique_ptr<FileChecksumGenerator> file_checksum_gen = | |
270 | file_checksum_gen_factory->CreateFileChecksumGenerator(gen_context); | |
271 | *file_checksum_func_name = file_checksum_gen->Name(); | |
272 | s = file_reader->Read(2048, &result, scratch.get()); | |
273 | if (!s.ok()) { | |
274 | return s; | |
275 | } | |
276 | while (result.size() != 0) { | |
277 | file_checksum_gen->Update(scratch.get(), result.size()); | |
278 | s = file_reader->Read(2048, &result, scratch.get()); | |
279 | if (!s.ok()) { | |
280 | return s; | |
281 | } | |
282 | } | |
283 | file_checksum_gen->Finalize(); | |
284 | *file_checksum = file_checksum_gen->GetChecksum(); | |
285 | } | |
286 | return Status::OK(); | |
287 | } | |
288 | }; | |
289 | ||
290 | TEST_F(ExternalSSTFileBasicTest, BasicWithFileChecksumCrc32c) { | |
291 | Options options = CurrentOptions(); | |
292 | options.file_checksum_gen_factory = GetFileChecksumGenCrc32cFactory(); | |
293 | ChecksumVerifyHelper checksum_helper(options); | |
294 | ||
295 | SstFileWriter sst_file_writer(EnvOptions(), options); | |
296 | ||
297 | // Current file size should be 0 after sst_file_writer init and before open a | |
298 | // file. | |
299 | ASSERT_EQ(sst_file_writer.FileSize(), 0); | |
300 | ||
301 | // file1.sst (0 => 99) | |
302 | std::string file1 = sst_files_dir_ + "file1.sst"; | |
303 | ASSERT_OK(sst_file_writer.Open(file1)); | |
304 | for (int k = 0; k < 100; k++) { | |
305 | ASSERT_OK(sst_file_writer.Put(Key(k), Key(k) + "_val")); | |
306 | } | |
307 | ExternalSstFileInfo file1_info; | |
308 | Status s = sst_file_writer.Finish(&file1_info); | |
1e59de90 | 309 | ASSERT_OK(s) << s.ToString(); |
20effc67 TL |
310 | std::string file_checksum, file_checksum_func_name; |
311 | ASSERT_OK(checksum_helper.GetSingleFileChecksumAndFuncName( | |
312 | file1, &file_checksum, &file_checksum_func_name)); | |
313 | ||
314 | // Current file size should be non-zero after success write. | |
315 | ASSERT_GT(sst_file_writer.FileSize(), 0); | |
316 | ||
317 | ASSERT_EQ(file1_info.file_path, file1); | |
318 | ASSERT_EQ(file1_info.num_entries, 100); | |
319 | ASSERT_EQ(file1_info.smallest_key, Key(0)); | |
320 | ASSERT_EQ(file1_info.largest_key, Key(99)); | |
321 | ASSERT_EQ(file1_info.num_range_del_entries, 0); | |
322 | ASSERT_EQ(file1_info.smallest_range_del_key, ""); | |
323 | ASSERT_EQ(file1_info.largest_range_del_key, ""); | |
324 | ASSERT_EQ(file1_info.file_checksum, file_checksum); | |
325 | ASSERT_EQ(file1_info.file_checksum_func_name, file_checksum_func_name); | |
7c673cae | 326 | // sst_file_writer already finished, cannot add this value |
11fdf7f2 | 327 | s = sst_file_writer.Put(Key(100), "bad_val"); |
1e59de90 | 328 | ASSERT_NOK(s) << s.ToString(); |
11fdf7f2 | 329 | s = sst_file_writer.DeleteRange(Key(100), Key(200)); |
1e59de90 | 330 | ASSERT_NOK(s) << s.ToString(); |
7c673cae FG |
331 | |
332 | DestroyAndReopen(options); | |
333 | // Add file using file path | |
334 | s = DeprecatedAddFile({file1}); | |
1e59de90 | 335 | ASSERT_OK(s) << s.ToString(); |
7c673cae FG |
336 | ASSERT_EQ(db_->GetLatestSequenceNumber(), 0U); |
337 | for (int k = 0; k < 100; k++) { | |
338 | ASSERT_EQ(Get(Key(k)), Key(k) + "_val"); | |
339 | } | |
340 | ||
341 | DestroyAndRecreateExternalSSTFilesDir(); | |
342 | } | |
343 | ||
20effc67 TL |
344 | TEST_F(ExternalSSTFileBasicTest, IngestFileWithFileChecksum) { |
345 | Options old_options = CurrentOptions(); | |
346 | Options options = CurrentOptions(); | |
347 | options.file_checksum_gen_factory = GetFileChecksumGenCrc32cFactory(); | |
348 | const ImmutableCFOptions ioptions(options); | |
349 | ChecksumVerifyHelper checksum_helper(options); | |
350 | ||
351 | SstFileWriter sst_file_writer(EnvOptions(), options); | |
352 | ||
353 | // file01.sst (1000 => 1099) | |
354 | std::string file1 = sst_files_dir_ + "file01.sst"; | |
355 | ASSERT_OK(sst_file_writer.Open(file1)); | |
356 | for (int k = 1000; k < 1100; k++) { | |
357 | ASSERT_OK(sst_file_writer.Put(Key(k), Key(k) + "_val")); | |
358 | } | |
359 | ExternalSstFileInfo file1_info; | |
360 | Status s = sst_file_writer.Finish(&file1_info); | |
1e59de90 | 361 | ASSERT_OK(s) << s.ToString(); |
20effc67 TL |
362 | ASSERT_EQ(file1_info.file_path, file1); |
363 | ASSERT_EQ(file1_info.num_entries, 100); | |
364 | ASSERT_EQ(file1_info.smallest_key, Key(1000)); | |
365 | ASSERT_EQ(file1_info.largest_key, Key(1099)); | |
366 | std::string file_checksum1, file_checksum_func_name1; | |
367 | ASSERT_OK(checksum_helper.GetSingleFileChecksumAndFuncName( | |
368 | file1, &file_checksum1, &file_checksum_func_name1)); | |
369 | ASSERT_EQ(file1_info.file_checksum, file_checksum1); | |
370 | ASSERT_EQ(file1_info.file_checksum_func_name, file_checksum_func_name1); | |
371 | ||
372 | // file02.sst (1100 => 1299) | |
373 | std::string file2 = sst_files_dir_ + "file02.sst"; | |
374 | ASSERT_OK(sst_file_writer.Open(file2)); | |
375 | for (int k = 1100; k < 1300; k++) { | |
376 | ASSERT_OK(sst_file_writer.Put(Key(k), Key(k) + "_val")); | |
377 | } | |
378 | ExternalSstFileInfo file2_info; | |
379 | s = sst_file_writer.Finish(&file2_info); | |
1e59de90 | 380 | ASSERT_OK(s) << s.ToString(); |
20effc67 TL |
381 | ASSERT_EQ(file2_info.file_path, file2); |
382 | ASSERT_EQ(file2_info.num_entries, 200); | |
383 | ASSERT_EQ(file2_info.smallest_key, Key(1100)); | |
384 | ASSERT_EQ(file2_info.largest_key, Key(1299)); | |
385 | std::string file_checksum2, file_checksum_func_name2; | |
386 | ASSERT_OK(checksum_helper.GetSingleFileChecksumAndFuncName( | |
387 | file2, &file_checksum2, &file_checksum_func_name2)); | |
388 | ASSERT_EQ(file2_info.file_checksum, file_checksum2); | |
389 | ASSERT_EQ(file2_info.file_checksum_func_name, file_checksum_func_name2); | |
390 | ||
391 | // file03.sst (1300 => 1499) | |
392 | std::string file3 = sst_files_dir_ + "file03.sst"; | |
393 | ASSERT_OK(sst_file_writer.Open(file3)); | |
394 | for (int k = 1300; k < 1500; k++) { | |
395 | ASSERT_OK(sst_file_writer.Put(Key(k), Key(k) + "_val_overlap")); | |
396 | } | |
397 | ExternalSstFileInfo file3_info; | |
398 | s = sst_file_writer.Finish(&file3_info); | |
1e59de90 | 399 | ASSERT_OK(s) << s.ToString(); |
20effc67 TL |
400 | ASSERT_EQ(file3_info.file_path, file3); |
401 | ASSERT_EQ(file3_info.num_entries, 200); | |
402 | ASSERT_EQ(file3_info.smallest_key, Key(1300)); | |
403 | ASSERT_EQ(file3_info.largest_key, Key(1499)); | |
404 | std::string file_checksum3, file_checksum_func_name3; | |
405 | ASSERT_OK(checksum_helper.GetSingleFileChecksumAndFuncName( | |
406 | file3, &file_checksum3, &file_checksum_func_name3)); | |
407 | ASSERT_EQ(file3_info.file_checksum, file_checksum3); | |
408 | ASSERT_EQ(file3_info.file_checksum_func_name, file_checksum_func_name3); | |
409 | ||
410 | // file04.sst (1500 => 1799) | |
411 | std::string file4 = sst_files_dir_ + "file04.sst"; | |
412 | ASSERT_OK(sst_file_writer.Open(file4)); | |
413 | for (int k = 1500; k < 1800; k++) { | |
414 | ASSERT_OK(sst_file_writer.Put(Key(k), Key(k) + "_val_overlap")); | |
415 | } | |
416 | ExternalSstFileInfo file4_info; | |
417 | s = sst_file_writer.Finish(&file4_info); | |
1e59de90 | 418 | ASSERT_OK(s) << s.ToString(); |
20effc67 TL |
419 | ASSERT_EQ(file4_info.file_path, file4); |
420 | ASSERT_EQ(file4_info.num_entries, 300); | |
421 | ASSERT_EQ(file4_info.smallest_key, Key(1500)); | |
422 | ASSERT_EQ(file4_info.largest_key, Key(1799)); | |
423 | std::string file_checksum4, file_checksum_func_name4; | |
424 | ASSERT_OK(checksum_helper.GetSingleFileChecksumAndFuncName( | |
425 | file4, &file_checksum4, &file_checksum_func_name4)); | |
426 | ASSERT_EQ(file4_info.file_checksum, file_checksum4); | |
427 | ASSERT_EQ(file4_info.file_checksum_func_name, file_checksum_func_name4); | |
428 | ||
429 | // file05.sst (1800 => 1899) | |
430 | std::string file5 = sst_files_dir_ + "file05.sst"; | |
431 | ASSERT_OK(sst_file_writer.Open(file5)); | |
432 | for (int k = 1800; k < 2000; k++) { | |
433 | ASSERT_OK(sst_file_writer.Put(Key(k), Key(k) + "_val_overlap")); | |
434 | } | |
435 | ExternalSstFileInfo file5_info; | |
436 | s = sst_file_writer.Finish(&file5_info); | |
1e59de90 | 437 | ASSERT_OK(s) << s.ToString(); |
20effc67 TL |
438 | ASSERT_EQ(file5_info.file_path, file5); |
439 | ASSERT_EQ(file5_info.num_entries, 200); | |
440 | ASSERT_EQ(file5_info.smallest_key, Key(1800)); | |
441 | ASSERT_EQ(file5_info.largest_key, Key(1999)); | |
442 | std::string file_checksum5, file_checksum_func_name5; | |
443 | ASSERT_OK(checksum_helper.GetSingleFileChecksumAndFuncName( | |
444 | file5, &file_checksum5, &file_checksum_func_name5)); | |
445 | ASSERT_EQ(file5_info.file_checksum, file_checksum5); | |
446 | ASSERT_EQ(file5_info.file_checksum_func_name, file_checksum_func_name5); | |
447 | ||
448 | // file06.sst (2000 => 2199) | |
449 | std::string file6 = sst_files_dir_ + "file06.sst"; | |
450 | ASSERT_OK(sst_file_writer.Open(file6)); | |
451 | for (int k = 2000; k < 2200; k++) { | |
452 | ASSERT_OK(sst_file_writer.Put(Key(k), Key(k) + "_val_overlap")); | |
453 | } | |
454 | ExternalSstFileInfo file6_info; | |
455 | s = sst_file_writer.Finish(&file6_info); | |
1e59de90 | 456 | ASSERT_OK(s) << s.ToString(); |
20effc67 TL |
457 | ASSERT_EQ(file6_info.file_path, file6); |
458 | ASSERT_EQ(file6_info.num_entries, 200); | |
459 | ASSERT_EQ(file6_info.smallest_key, Key(2000)); | |
460 | ASSERT_EQ(file6_info.largest_key, Key(2199)); | |
461 | std::string file_checksum6, file_checksum_func_name6; | |
462 | ASSERT_OK(checksum_helper.GetSingleFileChecksumAndFuncName( | |
463 | file6, &file_checksum6, &file_checksum_func_name6)); | |
464 | ASSERT_EQ(file6_info.file_checksum, file_checksum6); | |
465 | ASSERT_EQ(file6_info.file_checksum_func_name, file_checksum_func_name6); | |
466 | ||
467 | s = AddFileWithFileChecksum({file1}, {file_checksum1, "xyz"}, | |
468 | {file_checksum1}, true, false, false, false); | |
469 | // does not care the checksum input since db does not enable file checksum | |
1e59de90 | 470 | ASSERT_OK(s) << s.ToString(); |
20effc67 TL |
471 | ASSERT_OK(env_->FileExists(file1)); |
472 | std::vector<LiveFileMetaData> live_files; | |
473 | dbfull()->GetLiveFilesMetaData(&live_files); | |
474 | std::set<std::string> set1; | |
475 | for (auto f : live_files) { | |
476 | set1.insert(f.name); | |
477 | ASSERT_EQ(f.file_checksum, kUnknownFileChecksum); | |
478 | ASSERT_EQ(f.file_checksum_func_name, kUnknownFileChecksumFuncName); | |
479 | } | |
480 | ||
1e59de90 TL |
481 | // check the temperature of the file being ingested |
482 | ColumnFamilyMetaData metadata; | |
483 | db_->GetColumnFamilyMetaData(&metadata); | |
484 | ASSERT_EQ(1, metadata.file_count); | |
485 | ASSERT_EQ(Temperature::kUnknown, metadata.levels[6].files[0].temperature); | |
486 | auto size = GetSstSizeHelper(Temperature::kUnknown); | |
487 | ASSERT_GT(size, 0); | |
488 | size = GetSstSizeHelper(Temperature::kWarm); | |
489 | ASSERT_EQ(size, 0); | |
490 | size = GetSstSizeHelper(Temperature::kHot); | |
491 | ASSERT_EQ(size, 0); | |
492 | size = GetSstSizeHelper(Temperature::kCold); | |
493 | ASSERT_EQ(size, 0); | |
494 | ||
20effc67 TL |
495 | // Reopen Db with checksum enabled |
496 | Reopen(options); | |
497 | // Enable verify_file_checksum option | |
498 | // The checksum vector does not match, fail the ingestion | |
499 | s = AddFileWithFileChecksum({file2}, {file_checksum2, "xyz"}, | |
500 | {file_checksum_func_name2}, true, false, false, | |
501 | false); | |
1e59de90 | 502 | ASSERT_NOK(s) << s.ToString(); |
20effc67 TL |
503 | |
504 | // Enable verify_file_checksum option | |
505 | // The checksum name does not match, fail the ingestion | |
506 | s = AddFileWithFileChecksum({file2}, {file_checksum2}, {"xyz"}, true, false, | |
507 | false, false); | |
1e59de90 | 508 | ASSERT_NOK(s) << s.ToString(); |
20effc67 TL |
509 | |
510 | // Enable verify_file_checksum option | |
511 | // The checksum itself does not match, fail the ingestion | |
512 | s = AddFileWithFileChecksum({file2}, {"xyz"}, {file_checksum_func_name2}, | |
513 | true, false, false, false); | |
1e59de90 | 514 | ASSERT_NOK(s) << s.ToString(); |
20effc67 TL |
515 | |
516 | // Enable verify_file_checksum option | |
517 | // All matches, ingestion is successful | |
518 | s = AddFileWithFileChecksum({file2}, {file_checksum2}, | |
519 | {file_checksum_func_name2}, true, false, false, | |
520 | false); | |
1e59de90 | 521 | ASSERT_OK(s) << s.ToString(); |
20effc67 TL |
522 | std::vector<LiveFileMetaData> live_files1; |
523 | dbfull()->GetLiveFilesMetaData(&live_files1); | |
524 | for (auto f : live_files1) { | |
525 | if (set1.find(f.name) == set1.end()) { | |
526 | ASSERT_EQ(f.file_checksum, file_checksum2); | |
527 | ASSERT_EQ(f.file_checksum_func_name, file_checksum_func_name2); | |
528 | set1.insert(f.name); | |
529 | } | |
530 | } | |
531 | ASSERT_OK(env_->FileExists(file2)); | |
532 | ||
533 | // Enable verify_file_checksum option | |
534 | // No checksum information is provided, generate it when ingesting | |
535 | std::vector<std::string> checksum, checksum_func; | |
536 | s = AddFileWithFileChecksum({file3}, checksum, checksum_func, true, false, | |
537 | false, false); | |
1e59de90 | 538 | ASSERT_OK(s) << s.ToString(); |
20effc67 TL |
539 | std::vector<LiveFileMetaData> live_files2; |
540 | dbfull()->GetLiveFilesMetaData(&live_files2); | |
541 | for (auto f : live_files2) { | |
542 | if (set1.find(f.name) == set1.end()) { | |
543 | ASSERT_EQ(f.file_checksum, file_checksum3); | |
544 | ASSERT_EQ(f.file_checksum_func_name, file_checksum_func_name3); | |
545 | set1.insert(f.name); | |
546 | } | |
547 | } | |
1e59de90 | 548 | ASSERT_OK(s) << s.ToString(); |
20effc67 TL |
549 | ASSERT_OK(env_->FileExists(file3)); |
550 | ||
551 | // Does not enable verify_file_checksum options | |
552 | // The checksum name does not match, fail the ingestion | |
553 | s = AddFileWithFileChecksum({file4}, {file_checksum4}, {"xyz"}, false, false, | |
554 | false, false); | |
1e59de90 | 555 | ASSERT_NOK(s) << s.ToString(); |
20effc67 TL |
556 | |
557 | // Does not enable verify_file_checksum options | |
558 | // Checksum function name matches, store the checksum being ingested. | |
559 | s = AddFileWithFileChecksum({file4}, {"asd"}, {file_checksum_func_name4}, | |
560 | false, false, false, false); | |
1e59de90 | 561 | ASSERT_OK(s) << s.ToString(); |
20effc67 TL |
562 | std::vector<LiveFileMetaData> live_files3; |
563 | dbfull()->GetLiveFilesMetaData(&live_files3); | |
564 | for (auto f : live_files3) { | |
565 | if (set1.find(f.name) == set1.end()) { | |
566 | ASSERT_FALSE(f.file_checksum == file_checksum4); | |
567 | ASSERT_EQ(f.file_checksum, "asd"); | |
568 | ASSERT_EQ(f.file_checksum_func_name, file_checksum_func_name4); | |
569 | set1.insert(f.name); | |
570 | } | |
571 | } | |
1e59de90 | 572 | ASSERT_OK(s) << s.ToString(); |
20effc67 TL |
573 | ASSERT_OK(env_->FileExists(file4)); |
574 | ||
575 | // enable verify_file_checksum options, DB enable checksum, and enable | |
576 | // write_global_seq. So the checksum stored is different from the one | |
577 | // ingested due to the sequence number changes. | |
578 | s = AddFileWithFileChecksum({file5}, {file_checksum5}, | |
579 | {file_checksum_func_name5}, true, false, false, | |
580 | true); | |
1e59de90 | 581 | ASSERT_OK(s) << s.ToString(); |
20effc67 TL |
582 | std::vector<LiveFileMetaData> live_files4; |
583 | dbfull()->GetLiveFilesMetaData(&live_files4); | |
584 | for (auto f : live_files4) { | |
585 | if (set1.find(f.name) == set1.end()) { | |
586 | std::string cur_checksum5, cur_checksum_func_name5; | |
587 | ASSERT_OK(checksum_helper.GetSingleFileChecksumAndFuncName( | |
588 | dbname_ + f.name, &cur_checksum5, &cur_checksum_func_name5)); | |
589 | ASSERT_EQ(f.file_checksum, cur_checksum5); | |
590 | ASSERT_EQ(f.file_checksum_func_name, file_checksum_func_name5); | |
591 | set1.insert(f.name); | |
592 | } | |
593 | } | |
1e59de90 | 594 | ASSERT_OK(s) << s.ToString(); |
20effc67 TL |
595 | ASSERT_OK(env_->FileExists(file5)); |
596 | ||
597 | // Does not enable verify_file_checksum options and also the ingested file | |
598 | // checksum information is empty. DB will generate and store the checksum | |
599 | // in Manifest. | |
600 | std::vector<std::string> files_c6, files_name6; | |
601 | s = AddFileWithFileChecksum({file6}, files_c6, files_name6, false, false, | |
602 | false, false); | |
1e59de90 | 603 | ASSERT_OK(s) << s.ToString(); |
20effc67 TL |
604 | std::vector<LiveFileMetaData> live_files6; |
605 | dbfull()->GetLiveFilesMetaData(&live_files6); | |
606 | for (auto f : live_files6) { | |
607 | if (set1.find(f.name) == set1.end()) { | |
608 | ASSERT_EQ(f.file_checksum, file_checksum6); | |
609 | ASSERT_EQ(f.file_checksum_func_name, file_checksum_func_name6); | |
610 | set1.insert(f.name); | |
611 | } | |
612 | } | |
1e59de90 | 613 | ASSERT_OK(s) << s.ToString(); |
20effc67 | 614 | ASSERT_OK(env_->FileExists(file6)); |
1e59de90 TL |
615 | db_->GetColumnFamilyMetaData(&metadata); |
616 | size = GetSstSizeHelper(Temperature::kUnknown); | |
617 | ASSERT_GT(size, 0); | |
618 | size = GetSstSizeHelper(Temperature::kWarm); | |
619 | ASSERT_EQ(size, 0); | |
620 | size = GetSstSizeHelper(Temperature::kHot); | |
621 | ASSERT_EQ(size, 0); | |
622 | size = GetSstSizeHelper(Temperature::kCold); | |
623 | ASSERT_EQ(size, 0); | |
20effc67 TL |
624 | } |
625 | ||
7c673cae FG |
626 | TEST_F(ExternalSSTFileBasicTest, NoCopy) { |
627 | Options options = CurrentOptions(); | |
628 | const ImmutableCFOptions ioptions(options); | |
629 | ||
630 | SstFileWriter sst_file_writer(EnvOptions(), options); | |
631 | ||
632 | // file1.sst (0 => 99) | |
633 | std::string file1 = sst_files_dir_ + "file1.sst"; | |
634 | ASSERT_OK(sst_file_writer.Open(file1)); | |
635 | for (int k = 0; k < 100; k++) { | |
11fdf7f2 | 636 | ASSERT_OK(sst_file_writer.Put(Key(k), Key(k) + "_val")); |
7c673cae FG |
637 | } |
638 | ExternalSstFileInfo file1_info; | |
639 | Status s = sst_file_writer.Finish(&file1_info); | |
1e59de90 | 640 | ASSERT_OK(s) << s.ToString(); |
7c673cae FG |
641 | ASSERT_EQ(file1_info.file_path, file1); |
642 | ASSERT_EQ(file1_info.num_entries, 100); | |
643 | ASSERT_EQ(file1_info.smallest_key, Key(0)); | |
644 | ASSERT_EQ(file1_info.largest_key, Key(99)); | |
645 | ||
646 | // file2.sst (100 => 299) | |
647 | std::string file2 = sst_files_dir_ + "file2.sst"; | |
648 | ASSERT_OK(sst_file_writer.Open(file2)); | |
649 | for (int k = 100; k < 300; k++) { | |
11fdf7f2 | 650 | ASSERT_OK(sst_file_writer.Put(Key(k), Key(k) + "_val")); |
7c673cae FG |
651 | } |
652 | ExternalSstFileInfo file2_info; | |
653 | s = sst_file_writer.Finish(&file2_info); | |
1e59de90 | 654 | ASSERT_OK(s) << s.ToString(); |
7c673cae FG |
655 | ASSERT_EQ(file2_info.file_path, file2); |
656 | ASSERT_EQ(file2_info.num_entries, 200); | |
657 | ASSERT_EQ(file2_info.smallest_key, Key(100)); | |
658 | ASSERT_EQ(file2_info.largest_key, Key(299)); | |
659 | ||
660 | // file3.sst (110 => 124) .. overlap with file2.sst | |
661 | std::string file3 = sst_files_dir_ + "file3.sst"; | |
662 | ASSERT_OK(sst_file_writer.Open(file3)); | |
663 | for (int k = 110; k < 125; k++) { | |
11fdf7f2 | 664 | ASSERT_OK(sst_file_writer.Put(Key(k), Key(k) + "_val_overlap")); |
7c673cae FG |
665 | } |
666 | ExternalSstFileInfo file3_info; | |
667 | s = sst_file_writer.Finish(&file3_info); | |
1e59de90 | 668 | ASSERT_OK(s) << s.ToString(); |
7c673cae FG |
669 | ASSERT_EQ(file3_info.file_path, file3); |
670 | ASSERT_EQ(file3_info.num_entries, 15); | |
671 | ASSERT_EQ(file3_info.smallest_key, Key(110)); | |
672 | ASSERT_EQ(file3_info.largest_key, Key(124)); | |
11fdf7f2 | 673 | |
7c673cae | 674 | s = DeprecatedAddFile({file1}, true /* move file */); |
1e59de90 | 675 | ASSERT_OK(s) << s.ToString(); |
7c673cae FG |
676 | ASSERT_EQ(Status::NotFound(), env_->FileExists(file1)); |
677 | ||
678 | s = DeprecatedAddFile({file2}, false /* copy file */); | |
1e59de90 | 679 | ASSERT_OK(s) << s.ToString(); |
7c673cae FG |
680 | ASSERT_OK(env_->FileExists(file2)); |
681 | ||
11fdf7f2 TL |
682 | // This file has overlapping values with the existing data |
683 | s = DeprecatedAddFile({file3}, true /* move file */); | |
1e59de90 | 684 | ASSERT_NOK(s) << s.ToString(); |
7c673cae FG |
685 | ASSERT_OK(env_->FileExists(file3)); |
686 | ||
687 | for (int k = 0; k < 300; k++) { | |
688 | ASSERT_EQ(Get(Key(k)), Key(k) + "_val"); | |
689 | } | |
690 | } | |
691 | ||
494da23a TL |
692 | TEST_P(ExternalSSTFileBasicTest, IngestFileWithGlobalSeqnoPickedSeqno) { |
693 | bool write_global_seqno = std::get<0>(GetParam()); | |
694 | bool verify_checksums_before_ingest = std::get<1>(GetParam()); | |
7c673cae FG |
695 | do { |
696 | Options options = CurrentOptions(); | |
697 | DestroyAndReopen(options); | |
698 | std::map<std::string, std::string> true_data; | |
699 | ||
700 | int file_id = 1; | |
701 | ||
494da23a TL |
702 | ASSERT_OK(GenerateAndAddExternalFile( |
703 | options, {1, 2, 3, 4, 5, 6}, ValueType::kTypeValue, file_id++, | |
704 | write_global_seqno, verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 | 705 | // File doesn't overwrite any keys, no seqno needed |
7c673cae FG |
706 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 0); |
707 | ||
494da23a TL |
708 | ASSERT_OK(GenerateAndAddExternalFile( |
709 | options, {10, 11, 12, 13}, ValueType::kTypeValue, file_id++, | |
710 | write_global_seqno, verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 | 711 | // File doesn't overwrite any keys, no seqno needed |
7c673cae FG |
712 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 0); |
713 | ||
11fdf7f2 | 714 | ASSERT_OK(GenerateAndAddExternalFile( |
494da23a TL |
715 | options, {1, 4, 6}, ValueType::kTypeValue, file_id++, |
716 | write_global_seqno, verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 | 717 | // File overwrites some keys, a seqno will be assigned |
7c673cae FG |
718 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 1); |
719 | ||
11fdf7f2 | 720 | ASSERT_OK(GenerateAndAddExternalFile( |
494da23a TL |
721 | options, {11, 15, 19}, ValueType::kTypeValue, file_id++, |
722 | write_global_seqno, verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 | 723 | // File overwrites some keys, a seqno will be assigned |
7c673cae FG |
724 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 2); |
725 | ||
11fdf7f2 | 726 | ASSERT_OK(GenerateAndAddExternalFile( |
494da23a TL |
727 | options, {120, 130}, ValueType::kTypeValue, file_id++, |
728 | write_global_seqno, verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 | 729 | // File doesn't overwrite any keys, no seqno needed |
7c673cae FG |
730 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 2); |
731 | ||
11fdf7f2 | 732 | ASSERT_OK(GenerateAndAddExternalFile( |
494da23a TL |
733 | options, {1, 130}, ValueType::kTypeValue, file_id++, write_global_seqno, |
734 | verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 | 735 | // File overwrites some keys, a seqno will be assigned |
7c673cae FG |
736 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 3); |
737 | ||
738 | // Write some keys through normal write path | |
739 | for (int i = 0; i < 50; i++) { | |
740 | ASSERT_OK(Put(Key(i), "memtable")); | |
741 | true_data[Key(i)] = "memtable"; | |
742 | } | |
743 | SequenceNumber last_seqno = dbfull()->GetLatestSequenceNumber(); | |
744 | ||
11fdf7f2 | 745 | ASSERT_OK(GenerateAndAddExternalFile( |
494da23a TL |
746 | options, {60, 61, 62}, ValueType::kTypeValue, file_id++, |
747 | write_global_seqno, verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 | 748 | // File doesn't overwrite any keys, no seqno needed |
7c673cae FG |
749 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), last_seqno); |
750 | ||
11fdf7f2 | 751 | ASSERT_OK(GenerateAndAddExternalFile( |
494da23a TL |
752 | options, {40, 41, 42}, ValueType::kTypeValue, file_id++, |
753 | write_global_seqno, verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 TL |
754 | // File overwrites some keys, a seqno will be assigned |
755 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), last_seqno + 1); | |
756 | ||
757 | ASSERT_OK(GenerateAndAddExternalFile( | |
494da23a TL |
758 | options, {20, 30, 40}, ValueType::kTypeValue, file_id++, |
759 | write_global_seqno, verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 TL |
760 | // File overwrites some keys, a seqno will be assigned |
761 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), last_seqno + 2); | |
762 | ||
763 | const Snapshot* snapshot = db_->GetSnapshot(); | |
764 | ||
765 | // We will need a seqno for the file regardless if the file overwrite | |
766 | // keys in the DB or not because we have a snapshot | |
767 | ASSERT_OK(GenerateAndAddExternalFile( | |
494da23a TL |
768 | options, {1000, 1002}, ValueType::kTypeValue, file_id++, |
769 | write_global_seqno, verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 TL |
770 | // A global seqno will be assigned anyway because of the snapshot |
771 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), last_seqno + 3); | |
772 | ||
773 | ASSERT_OK(GenerateAndAddExternalFile( | |
494da23a TL |
774 | options, {2000, 3002}, ValueType::kTypeValue, file_id++, |
775 | write_global_seqno, verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 TL |
776 | // A global seqno will be assigned anyway because of the snapshot |
777 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), last_seqno + 4); | |
778 | ||
494da23a TL |
779 | ASSERT_OK(GenerateAndAddExternalFile( |
780 | options, {1, 20, 40, 100, 150}, ValueType::kTypeValue, file_id++, | |
781 | write_global_seqno, verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 TL |
782 | // A global seqno will be assigned anyway because of the snapshot |
783 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), last_seqno + 5); | |
784 | ||
785 | db_->ReleaseSnapshot(snapshot); | |
786 | ||
787 | ASSERT_OK(GenerateAndAddExternalFile( | |
494da23a TL |
788 | options, {5000, 5001}, ValueType::kTypeValue, file_id++, |
789 | write_global_seqno, verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 TL |
790 | // No snapshot anymore, no need to assign a seqno |
791 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), last_seqno + 5); | |
792 | ||
793 | size_t kcnt = 0; | |
794 | VerifyDBFromMap(true_data, &kcnt, false); | |
494da23a | 795 | } while (ChangeOptionsForFileIngestionTest()); |
11fdf7f2 TL |
796 | } |
797 | ||
494da23a TL |
798 | TEST_P(ExternalSSTFileBasicTest, IngestFileWithMultipleValueType) { |
799 | bool write_global_seqno = std::get<0>(GetParam()); | |
800 | bool verify_checksums_before_ingest = std::get<1>(GetParam()); | |
11fdf7f2 TL |
801 | do { |
802 | Options options = CurrentOptions(); | |
803 | options.merge_operator.reset(new TestPutOperator()); | |
804 | DestroyAndReopen(options); | |
805 | std::map<std::string, std::string> true_data; | |
806 | ||
807 | int file_id = 1; | |
808 | ||
494da23a TL |
809 | ASSERT_OK(GenerateAndAddExternalFile( |
810 | options, {1, 2, 3, 4, 5, 6}, ValueType::kTypeValue, file_id++, | |
811 | write_global_seqno, verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 TL |
812 | // File doesn't overwrite any keys, no seqno needed |
813 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 0); | |
814 | ||
494da23a TL |
815 | ASSERT_OK(GenerateAndAddExternalFile( |
816 | options, {10, 11, 12, 13}, ValueType::kTypeValue, file_id++, | |
817 | write_global_seqno, verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 TL |
818 | // File doesn't overwrite any keys, no seqno needed |
819 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 0); | |
820 | ||
821 | ASSERT_OK(GenerateAndAddExternalFile( | |
494da23a TL |
822 | options, {1, 4, 6}, ValueType::kTypeMerge, file_id++, |
823 | write_global_seqno, verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 TL |
824 | // File overwrites some keys, a seqno will be assigned |
825 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 1); | |
826 | ||
494da23a TL |
827 | ASSERT_OK(GenerateAndAddExternalFile( |
828 | options, {11, 15, 19}, ValueType::kTypeDeletion, file_id++, | |
829 | write_global_seqno, verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 TL |
830 | // File overwrites some keys, a seqno will be assigned |
831 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 2); | |
832 | ||
833 | ASSERT_OK(GenerateAndAddExternalFile( | |
494da23a TL |
834 | options, {120, 130}, ValueType::kTypeMerge, file_id++, |
835 | write_global_seqno, verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 TL |
836 | // File doesn't overwrite any keys, no seqno needed |
837 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 2); | |
838 | ||
839 | ASSERT_OK(GenerateAndAddExternalFile( | |
494da23a TL |
840 | options, {1, 130}, ValueType::kTypeDeletion, file_id++, |
841 | write_global_seqno, verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 TL |
842 | // File overwrites some keys, a seqno will be assigned |
843 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 3); | |
844 | ||
494da23a TL |
845 | ASSERT_OK(GenerateAndAddExternalFile( |
846 | options, {120}, {ValueType::kTypeValue}, {{120, 135}}, file_id++, | |
847 | write_global_seqno, verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 TL |
848 | // File overwrites some keys, a seqno will be assigned |
849 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 4); | |
850 | ||
494da23a TL |
851 | ASSERT_OK(GenerateAndAddExternalFile( |
852 | options, {}, {}, {{110, 120}}, file_id++, write_global_seqno, | |
853 | verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 TL |
854 | // The range deletion ends on a key, but it doesn't actually delete |
855 | // this key because the largest key in the range is exclusive. Still, | |
856 | // it counts as an overlap so a new seqno will be assigned. | |
857 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 5); | |
858 | ||
494da23a TL |
859 | ASSERT_OK(GenerateAndAddExternalFile( |
860 | options, {}, {}, {{100, 109}}, file_id++, write_global_seqno, | |
861 | verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 TL |
862 | // File doesn't overwrite any keys, no seqno needed |
863 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 5); | |
864 | ||
865 | // Write some keys through normal write path | |
866 | for (int i = 0; i < 50; i++) { | |
867 | ASSERT_OK(Put(Key(i), "memtable")); | |
868 | true_data[Key(i)] = "memtable"; | |
869 | } | |
870 | SequenceNumber last_seqno = dbfull()->GetLatestSequenceNumber(); | |
871 | ||
872 | ASSERT_OK(GenerateAndAddExternalFile( | |
494da23a TL |
873 | options, {60, 61, 62}, ValueType::kTypeValue, file_id++, |
874 | write_global_seqno, verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 TL |
875 | // File doesn't overwrite any keys, no seqno needed |
876 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), last_seqno); | |
877 | ||
878 | ASSERT_OK(GenerateAndAddExternalFile( | |
494da23a TL |
879 | options, {40, 41, 42}, ValueType::kTypeMerge, file_id++, |
880 | write_global_seqno, verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 | 881 | // File overwrites some keys, a seqno will be assigned |
7c673cae FG |
882 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), last_seqno + 1); |
883 | ||
494da23a TL |
884 | ASSERT_OK(GenerateAndAddExternalFile( |
885 | options, {20, 30, 40}, ValueType::kTypeDeletion, file_id++, | |
886 | write_global_seqno, verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 | 887 | // File overwrites some keys, a seqno will be assigned |
7c673cae FG |
888 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), last_seqno + 2); |
889 | ||
890 | const Snapshot* snapshot = db_->GetSnapshot(); | |
891 | ||
892 | // We will need a seqno for the file regardless if the file overwrite | |
893 | // keys in the DB or not because we have a snapshot | |
11fdf7f2 | 894 | ASSERT_OK(GenerateAndAddExternalFile( |
494da23a TL |
895 | options, {1000, 1002}, ValueType::kTypeMerge, file_id++, |
896 | write_global_seqno, verify_checksums_before_ingest, &true_data)); | |
7c673cae FG |
897 | // A global seqno will be assigned anyway because of the snapshot |
898 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), last_seqno + 3); | |
899 | ||
11fdf7f2 | 900 | ASSERT_OK(GenerateAndAddExternalFile( |
494da23a TL |
901 | options, {2000, 3002}, ValueType::kTypeMerge, file_id++, |
902 | write_global_seqno, verify_checksums_before_ingest, &true_data)); | |
7c673cae FG |
903 | // A global seqno will be assigned anyway because of the snapshot |
904 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), last_seqno + 4); | |
905 | ||
494da23a TL |
906 | ASSERT_OK(GenerateAndAddExternalFile( |
907 | options, {1, 20, 40, 100, 150}, ValueType::kTypeMerge, file_id++, | |
908 | write_global_seqno, verify_checksums_before_ingest, &true_data)); | |
7c673cae FG |
909 | // A global seqno will be assigned anyway because of the snapshot |
910 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), last_seqno + 5); | |
911 | ||
912 | db_->ReleaseSnapshot(snapshot); | |
913 | ||
11fdf7f2 | 914 | ASSERT_OK(GenerateAndAddExternalFile( |
494da23a TL |
915 | options, {5000, 5001}, ValueType::kTypeValue, file_id++, |
916 | write_global_seqno, verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 TL |
917 | // No snapshot anymore, no need to assign a seqno |
918 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), last_seqno + 5); | |
919 | ||
920 | size_t kcnt = 0; | |
921 | VerifyDBFromMap(true_data, &kcnt, false); | |
494da23a | 922 | } while (ChangeOptionsForFileIngestionTest()); |
11fdf7f2 TL |
923 | } |
924 | ||
494da23a TL |
925 | TEST_P(ExternalSSTFileBasicTest, IngestFileWithMixedValueType) { |
926 | bool write_global_seqno = std::get<0>(GetParam()); | |
927 | bool verify_checksums_before_ingest = std::get<1>(GetParam()); | |
11fdf7f2 TL |
928 | do { |
929 | Options options = CurrentOptions(); | |
930 | options.merge_operator.reset(new TestPutOperator()); | |
931 | DestroyAndReopen(options); | |
932 | std::map<std::string, std::string> true_data; | |
933 | ||
934 | int file_id = 1; | |
935 | ||
936 | ASSERT_OK(GenerateAndAddExternalFile( | |
937 | options, {1, 2, 3, 4, 5, 6}, | |
938 | {ValueType::kTypeValue, ValueType::kTypeMerge, ValueType::kTypeValue, | |
939 | ValueType::kTypeMerge, ValueType::kTypeValue, ValueType::kTypeMerge}, | |
494da23a TL |
940 | file_id++, write_global_seqno, verify_checksums_before_ingest, |
941 | &true_data)); | |
11fdf7f2 TL |
942 | // File doesn't overwrite any keys, no seqno needed |
943 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 0); | |
944 | ||
945 | ASSERT_OK(GenerateAndAddExternalFile( | |
946 | options, {10, 11, 12, 13}, | |
947 | {ValueType::kTypeValue, ValueType::kTypeMerge, ValueType::kTypeValue, | |
948 | ValueType::kTypeMerge}, | |
494da23a TL |
949 | file_id++, write_global_seqno, verify_checksums_before_ingest, |
950 | &true_data)); | |
11fdf7f2 TL |
951 | // File doesn't overwrite any keys, no seqno needed |
952 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 0); | |
953 | ||
954 | ASSERT_OK(GenerateAndAddExternalFile( | |
494da23a TL |
955 | options, {1, 4, 6}, |
956 | {ValueType::kTypeDeletion, ValueType::kTypeValue, | |
957 | ValueType::kTypeMerge}, | |
958 | file_id++, write_global_seqno, verify_checksums_before_ingest, | |
959 | &true_data)); | |
11fdf7f2 TL |
960 | // File overwrites some keys, a seqno will be assigned |
961 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 1); | |
962 | ||
963 | ASSERT_OK(GenerateAndAddExternalFile( | |
494da23a TL |
964 | options, {11, 15, 19}, |
965 | {ValueType::kTypeDeletion, ValueType::kTypeMerge, | |
966 | ValueType::kTypeValue}, | |
967 | file_id++, write_global_seqno, verify_checksums_before_ingest, | |
968 | &true_data)); | |
11fdf7f2 TL |
969 | // File overwrites some keys, a seqno will be assigned |
970 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 2); | |
971 | ||
972 | ASSERT_OK(GenerateAndAddExternalFile( | |
973 | options, {120, 130}, {ValueType::kTypeValue, ValueType::kTypeMerge}, | |
494da23a TL |
974 | file_id++, write_global_seqno, verify_checksums_before_ingest, |
975 | &true_data)); | |
11fdf7f2 TL |
976 | // File doesn't overwrite any keys, no seqno needed |
977 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 2); | |
978 | ||
979 | ASSERT_OK(GenerateAndAddExternalFile( | |
980 | options, {1, 130}, {ValueType::kTypeMerge, ValueType::kTypeDeletion}, | |
494da23a TL |
981 | file_id++, write_global_seqno, verify_checksums_before_ingest, |
982 | &true_data)); | |
11fdf7f2 TL |
983 | // File overwrites some keys, a seqno will be assigned |
984 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 3); | |
985 | ||
986 | ASSERT_OK(GenerateAndAddExternalFile( | |
987 | options, {150, 151, 152}, | |
988 | {ValueType::kTypeValue, ValueType::kTypeMerge, | |
989 | ValueType::kTypeDeletion}, | |
494da23a TL |
990 | {{150, 160}, {180, 190}}, file_id++, write_global_seqno, |
991 | verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 TL |
992 | // File doesn't overwrite any keys, no seqno needed |
993 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 3); | |
994 | ||
995 | ASSERT_OK(GenerateAndAddExternalFile( | |
996 | options, {150, 151, 152}, | |
997 | {ValueType::kTypeValue, ValueType::kTypeMerge, ValueType::kTypeValue}, | |
494da23a TL |
998 | {{200, 250}}, file_id++, write_global_seqno, |
999 | verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 TL |
1000 | // File overwrites some keys, a seqno will be assigned |
1001 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 4); | |
1002 | ||
1003 | ASSERT_OK(GenerateAndAddExternalFile( | |
1004 | options, {300, 301, 302}, | |
1005 | {ValueType::kTypeValue, ValueType::kTypeMerge, | |
1006 | ValueType::kTypeDeletion}, | |
494da23a TL |
1007 | {{1, 2}, {152, 154}}, file_id++, write_global_seqno, |
1008 | verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 TL |
1009 | // File overwrites some keys, a seqno will be assigned |
1010 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 5); | |
1011 | ||
1012 | // Write some keys through normal write path | |
1013 | for (int i = 0; i < 50; i++) { | |
1014 | ASSERT_OK(Put(Key(i), "memtable")); | |
1015 | true_data[Key(i)] = "memtable"; | |
1016 | } | |
1017 | SequenceNumber last_seqno = dbfull()->GetLatestSequenceNumber(); | |
1018 | ||
1019 | ASSERT_OK(GenerateAndAddExternalFile( | |
1020 | options, {60, 61, 62}, | |
1021 | {ValueType::kTypeValue, ValueType::kTypeMerge, ValueType::kTypeValue}, | |
494da23a TL |
1022 | file_id++, write_global_seqno, verify_checksums_before_ingest, |
1023 | &true_data)); | |
11fdf7f2 TL |
1024 | // File doesn't overwrite any keys, no seqno needed |
1025 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), last_seqno); | |
1026 | ||
1027 | ASSERT_OK(GenerateAndAddExternalFile( | |
1028 | options, {40, 41, 42}, | |
1029 | {ValueType::kTypeValue, ValueType::kTypeDeletion, | |
1030 | ValueType::kTypeDeletion}, | |
494da23a TL |
1031 | file_id++, write_global_seqno, verify_checksums_before_ingest, |
1032 | &true_data)); | |
11fdf7f2 TL |
1033 | // File overwrites some keys, a seqno will be assigned |
1034 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), last_seqno + 1); | |
1035 | ||
1036 | ASSERT_OK(GenerateAndAddExternalFile( | |
1037 | options, {20, 30, 40}, | |
1038 | {ValueType::kTypeDeletion, ValueType::kTypeDeletion, | |
1039 | ValueType::kTypeDeletion}, | |
494da23a TL |
1040 | file_id++, write_global_seqno, verify_checksums_before_ingest, |
1041 | &true_data)); | |
11fdf7f2 TL |
1042 | // File overwrites some keys, a seqno will be assigned |
1043 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), last_seqno + 2); | |
1044 | ||
1045 | const Snapshot* snapshot = db_->GetSnapshot(); | |
1046 | ||
1047 | // We will need a seqno for the file regardless if the file overwrite | |
1048 | // keys in the DB or not because we have a snapshot | |
1049 | ASSERT_OK(GenerateAndAddExternalFile( | |
1050 | options, {1000, 1002}, {ValueType::kTypeValue, ValueType::kTypeMerge}, | |
494da23a TL |
1051 | file_id++, write_global_seqno, verify_checksums_before_ingest, |
1052 | &true_data)); | |
11fdf7f2 TL |
1053 | // A global seqno will be assigned anyway because of the snapshot |
1054 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), last_seqno + 3); | |
1055 | ||
1056 | ASSERT_OK(GenerateAndAddExternalFile( | |
1057 | options, {2000, 3002}, {ValueType::kTypeValue, ValueType::kTypeMerge}, | |
494da23a TL |
1058 | file_id++, write_global_seqno, verify_checksums_before_ingest, |
1059 | &true_data)); | |
11fdf7f2 TL |
1060 | // A global seqno will be assigned anyway because of the snapshot |
1061 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), last_seqno + 4); | |
1062 | ||
1063 | ASSERT_OK(GenerateAndAddExternalFile( | |
1064 | options, {1, 20, 40, 100, 150}, | |
1065 | {ValueType::kTypeDeletion, ValueType::kTypeDeletion, | |
1066 | ValueType::kTypeValue, ValueType::kTypeMerge, ValueType::kTypeMerge}, | |
494da23a TL |
1067 | file_id++, write_global_seqno, verify_checksums_before_ingest, |
1068 | &true_data)); | |
11fdf7f2 TL |
1069 | // A global seqno will be assigned anyway because of the snapshot |
1070 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), last_seqno + 5); | |
1071 | ||
1072 | db_->ReleaseSnapshot(snapshot); | |
1073 | ||
1074 | ASSERT_OK(GenerateAndAddExternalFile( | |
1075 | options, {5000, 5001}, {ValueType::kTypeValue, ValueType::kTypeMerge}, | |
494da23a TL |
1076 | file_id++, write_global_seqno, verify_checksums_before_ingest, |
1077 | &true_data)); | |
7c673cae FG |
1078 | // No snapshot anymore, no need to assign a seqno |
1079 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), last_seqno + 5); | |
1080 | ||
1081 | size_t kcnt = 0; | |
1082 | VerifyDBFromMap(true_data, &kcnt, false); | |
494da23a | 1083 | } while (ChangeOptionsForFileIngestionTest()); |
7c673cae FG |
1084 | } |
1085 | ||
1086 | TEST_F(ExternalSSTFileBasicTest, FadviseTrigger) { | |
1087 | Options options = CurrentOptions(); | |
1088 | const int kNumKeys = 10000; | |
1089 | ||
1090 | size_t total_fadvised_bytes = 0; | |
f67539c2 | 1091 | ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack( |
11fdf7f2 | 1092 | "SstFileWriter::Rep::InvalidatePageCache", [&](void* arg) { |
7c673cae FG |
1093 | size_t fadvise_size = *(reinterpret_cast<size_t*>(arg)); |
1094 | total_fadvised_bytes += fadvise_size; | |
1095 | }); | |
f67539c2 | 1096 | ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing(); |
7c673cae FG |
1097 | |
1098 | std::unique_ptr<SstFileWriter> sst_file_writer; | |
1099 | ||
1100 | std::string sst_file_path = sst_files_dir_ + "file_fadvise_disable.sst"; | |
1101 | sst_file_writer.reset( | |
1102 | new SstFileWriter(EnvOptions(), options, nullptr, false)); | |
1103 | ASSERT_OK(sst_file_writer->Open(sst_file_path)); | |
1104 | for (int i = 0; i < kNumKeys; i++) { | |
11fdf7f2 | 1105 | ASSERT_OK(sst_file_writer->Put(Key(i), Key(i))); |
7c673cae FG |
1106 | } |
1107 | ASSERT_OK(sst_file_writer->Finish()); | |
1108 | // fadvise disabled | |
1109 | ASSERT_EQ(total_fadvised_bytes, 0); | |
1110 | ||
7c673cae FG |
1111 | sst_file_path = sst_files_dir_ + "file_fadvise_enable.sst"; |
1112 | sst_file_writer.reset( | |
1113 | new SstFileWriter(EnvOptions(), options, nullptr, true)); | |
1114 | ASSERT_OK(sst_file_writer->Open(sst_file_path)); | |
1115 | for (int i = 0; i < kNumKeys; i++) { | |
11fdf7f2 | 1116 | ASSERT_OK(sst_file_writer->Put(Key(i), Key(i))); |
7c673cae FG |
1117 | } |
1118 | ASSERT_OK(sst_file_writer->Finish()); | |
1119 | // fadvise enabled | |
1120 | ASSERT_EQ(total_fadvised_bytes, sst_file_writer->FileSize()); | |
1121 | ASSERT_GT(total_fadvised_bytes, 0); | |
1122 | ||
f67539c2 TL |
1123 | ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing(); |
1124 | } | |
1125 | ||
1126 | TEST_F(ExternalSSTFileBasicTest, SyncFailure) { | |
1127 | Options options; | |
1128 | options.create_if_missing = true; | |
1129 | options.env = fault_injection_test_env_.get(); | |
1130 | ||
1131 | std::vector<std::pair<std::string, std::string>> test_cases = { | |
1132 | {"ExternalSstFileIngestionJob::BeforeSyncIngestedFile", | |
1133 | "ExternalSstFileIngestionJob::AfterSyncIngestedFile"}, | |
1134 | {"ExternalSstFileIngestionJob::BeforeSyncDir", | |
1135 | "ExternalSstFileIngestionJob::AfterSyncDir"}, | |
1136 | {"ExternalSstFileIngestionJob::BeforeSyncGlobalSeqno", | |
1137 | "ExternalSstFileIngestionJob::AfterSyncGlobalSeqno"}}; | |
1138 | ||
1139 | for (size_t i = 0; i < test_cases.size(); i++) { | |
1e59de90 | 1140 | bool no_sync = false; |
f67539c2 TL |
1141 | SyncPoint::GetInstance()->SetCallBack(test_cases[i].first, [&](void*) { |
1142 | fault_injection_test_env_->SetFilesystemActive(false); | |
1143 | }); | |
1144 | SyncPoint::GetInstance()->SetCallBack(test_cases[i].second, [&](void*) { | |
1145 | fault_injection_test_env_->SetFilesystemActive(true); | |
1146 | }); | |
1e59de90 TL |
1147 | if (i == 0) { |
1148 | SyncPoint::GetInstance()->SetCallBack( | |
1149 | "ExternalSstFileIngestionJob::Prepare:Reopen", [&](void* s) { | |
1150 | Status* status = static_cast<Status*>(s); | |
1151 | if (status->IsNotSupported()) { | |
1152 | no_sync = true; | |
1153 | } | |
1154 | }); | |
1155 | } | |
1156 | if (i == 2) { | |
1157 | SyncPoint::GetInstance()->SetCallBack( | |
1158 | "ExternalSstFileIngestionJob::NewRandomRWFile", [&](void* s) { | |
1159 | Status* status = static_cast<Status*>(s); | |
1160 | if (status->IsNotSupported()) { | |
1161 | no_sync = true; | |
1162 | } | |
1163 | }); | |
1164 | } | |
f67539c2 TL |
1165 | SyncPoint::GetInstance()->EnableProcessing(); |
1166 | ||
1167 | DestroyAndReopen(options); | |
1168 | if (i == 2) { | |
1169 | ASSERT_OK(Put("foo", "v1")); | |
1170 | } | |
1171 | ||
1172 | Options sst_file_writer_options; | |
1e59de90 | 1173 | sst_file_writer_options.env = fault_injection_test_env_.get(); |
f67539c2 TL |
1174 | std::unique_ptr<SstFileWriter> sst_file_writer( |
1175 | new SstFileWriter(EnvOptions(), sst_file_writer_options)); | |
1176 | std::string file_name = | |
1e59de90 | 1177 | sst_files_dir_ + "sync_failure_test_" + std::to_string(i) + ".sst"; |
f67539c2 TL |
1178 | ASSERT_OK(sst_file_writer->Open(file_name)); |
1179 | ASSERT_OK(sst_file_writer->Put("bar", "v2")); | |
1180 | ASSERT_OK(sst_file_writer->Finish()); | |
1181 | ||
1182 | IngestExternalFileOptions ingest_opt; | |
1183 | if (i == 0) { | |
1184 | ingest_opt.move_files = true; | |
1185 | } | |
1186 | const Snapshot* snapshot = db_->GetSnapshot(); | |
1187 | if (i == 2) { | |
1188 | ingest_opt.write_global_seqno = true; | |
1189 | } | |
1e59de90 TL |
1190 | Status s = db_->IngestExternalFile({file_name}, ingest_opt); |
1191 | if (no_sync) { | |
1192 | ASSERT_OK(s); | |
1193 | } else { | |
1194 | ASSERT_NOK(s); | |
1195 | } | |
f67539c2 TL |
1196 | db_->ReleaseSnapshot(snapshot); |
1197 | ||
1198 | SyncPoint::GetInstance()->DisableProcessing(); | |
1199 | SyncPoint::GetInstance()->ClearAllCallBacks(); | |
1200 | Destroy(options); | |
1201 | } | |
1202 | } | |
1203 | ||
1e59de90 TL |
1204 | TEST_F(ExternalSSTFileBasicTest, ReopenNotSupported) { |
1205 | Options options; | |
1206 | options.create_if_missing = true; | |
1207 | options.env = env_; | |
1208 | ||
1209 | SyncPoint::GetInstance()->SetCallBack( | |
1210 | "ExternalSstFileIngestionJob::Prepare:Reopen", [&](void* arg) { | |
1211 | Status* s = static_cast<Status*>(arg); | |
1212 | *s = Status::NotSupported(); | |
1213 | }); | |
1214 | SyncPoint::GetInstance()->EnableProcessing(); | |
1215 | ||
1216 | DestroyAndReopen(options); | |
1217 | ||
1218 | Options sst_file_writer_options; | |
1219 | sst_file_writer_options.env = env_; | |
1220 | std::unique_ptr<SstFileWriter> sst_file_writer( | |
1221 | new SstFileWriter(EnvOptions(), sst_file_writer_options)); | |
1222 | std::string file_name = | |
1223 | sst_files_dir_ + "reopen_not_supported_test_" + ".sst"; | |
1224 | ASSERT_OK(sst_file_writer->Open(file_name)); | |
1225 | ASSERT_OK(sst_file_writer->Put("bar", "v2")); | |
1226 | ASSERT_OK(sst_file_writer->Finish()); | |
1227 | ||
1228 | IngestExternalFileOptions ingest_opt; | |
1229 | ingest_opt.move_files = true; | |
1230 | const Snapshot* snapshot = db_->GetSnapshot(); | |
1231 | ASSERT_OK(db_->IngestExternalFile({file_name}, ingest_opt)); | |
1232 | db_->ReleaseSnapshot(snapshot); | |
1233 | ||
1234 | SyncPoint::GetInstance()->DisableProcessing(); | |
1235 | SyncPoint::GetInstance()->ClearAllCallBacks(); | |
1236 | Destroy(options); | |
1237 | } | |
1238 | ||
f67539c2 TL |
1239 | TEST_F(ExternalSSTFileBasicTest, VerifyChecksumReadahead) { |
1240 | Options options; | |
1241 | options.create_if_missing = true; | |
20effc67 | 1242 | SpecialEnv senv(env_); |
f67539c2 TL |
1243 | options.env = &senv; |
1244 | DestroyAndReopen(options); | |
1245 | ||
1246 | Options sst_file_writer_options; | |
20effc67 | 1247 | sst_file_writer_options.env = env_; |
f67539c2 TL |
1248 | std::unique_ptr<SstFileWriter> sst_file_writer( |
1249 | new SstFileWriter(EnvOptions(), sst_file_writer_options)); | |
1250 | std::string file_name = sst_files_dir_ + "verify_checksum_readahead_test.sst"; | |
1251 | ASSERT_OK(sst_file_writer->Open(file_name)); | |
1252 | Random rnd(301); | |
20effc67 | 1253 | std::string value = rnd.RandomString(4000); |
f67539c2 TL |
1254 | for (int i = 0; i < 5000; i++) { |
1255 | ASSERT_OK(sst_file_writer->Put(DBTestBase::Key(i), value)); | |
1256 | } | |
1257 | ASSERT_OK(sst_file_writer->Finish()); | |
1258 | ||
1259 | // Ingest it once without verifying checksums to see the baseline | |
1260 | // preads. | |
1261 | IngestExternalFileOptions ingest_opt; | |
1262 | ingest_opt.move_files = false; | |
1263 | senv.count_random_reads_ = true; | |
1264 | senv.random_read_bytes_counter_ = 0; | |
1265 | ASSERT_OK(db_->IngestExternalFile({file_name}, ingest_opt)); | |
1266 | ||
1267 | auto base_num_reads = senv.random_read_counter_.Read(); | |
1268 | // Make sure the counter is enabled. | |
1269 | ASSERT_GT(base_num_reads, 0); | |
1270 | ||
1271 | // Ingest again and observe the reads made for for readahead. | |
1272 | ingest_opt.move_files = false; | |
1273 | ingest_opt.verify_checksums_before_ingest = true; | |
1274 | ingest_opt.verify_checksums_readahead_size = size_t{2 * 1024 * 1024}; | |
1275 | ||
1276 | senv.count_random_reads_ = true; | |
1277 | senv.random_read_bytes_counter_ = 0; | |
1278 | ASSERT_OK(db_->IngestExternalFile({file_name}, ingest_opt)); | |
1279 | ||
1280 | // Make sure the counter is enabled. | |
1281 | ASSERT_GT(senv.random_read_counter_.Read() - base_num_reads, 0); | |
1282 | ||
1283 | // The SST file is about 20MB. Readahead size is 2MB. | |
1284 | // Give a conservative 15 reads for metadata blocks, the number | |
1285 | // of random reads should be within 20 MB / 2MB + 15 = 25. | |
1286 | ASSERT_LE(senv.random_read_counter_.Read() - base_num_reads, 40); | |
1287 | ||
1288 | Destroy(options); | |
7c673cae FG |
1289 | } |
1290 | ||
20effc67 TL |
1291 | TEST_F(ExternalSSTFileBasicTest, IngestRangeDeletionTombstoneWithGlobalSeqno) { |
1292 | for (int i = 5; i < 25; i++) { | |
1293 | ASSERT_OK(db_->Put(WriteOptions(), db_->DefaultColumnFamily(), Key(i), | |
1294 | Key(i) + "_val")); | |
1295 | } | |
1296 | ||
1297 | Options options = CurrentOptions(); | |
1298 | options.disable_auto_compactions = true; | |
1299 | Reopen(options); | |
1300 | SstFileWriter sst_file_writer(EnvOptions(), options); | |
1301 | ||
1302 | // file.sst (delete 0 => 30) | |
1303 | std::string file = sst_files_dir_ + "file.sst"; | |
1304 | ASSERT_OK(sst_file_writer.Open(file)); | |
1305 | ASSERT_OK(sst_file_writer.DeleteRange(Key(0), Key(30))); | |
1306 | ExternalSstFileInfo file_info; | |
1307 | ASSERT_OK(sst_file_writer.Finish(&file_info)); | |
1308 | ASSERT_EQ(file_info.file_path, file); | |
1309 | ASSERT_EQ(file_info.num_entries, 0); | |
1310 | ASSERT_EQ(file_info.smallest_key, ""); | |
1311 | ASSERT_EQ(file_info.largest_key, ""); | |
1312 | ASSERT_EQ(file_info.num_range_del_entries, 1); | |
1313 | ASSERT_EQ(file_info.smallest_range_del_key, Key(0)); | |
1314 | ASSERT_EQ(file_info.largest_range_del_key, Key(30)); | |
1315 | ||
1316 | IngestExternalFileOptions ifo; | |
1317 | ifo.move_files = true; | |
1318 | ifo.snapshot_consistency = true; | |
1319 | ifo.allow_global_seqno = true; | |
1320 | ifo.write_global_seqno = true; | |
1321 | ifo.verify_checksums_before_ingest = false; | |
1322 | ASSERT_OK(db_->IngestExternalFile({file}, ifo)); | |
1323 | ||
1324 | for (int i = 5; i < 25; i++) { | |
1325 | std::string res; | |
1326 | ASSERT_TRUE(db_->Get(ReadOptions(), Key(i), &res).IsNotFound()); | |
1327 | } | |
1328 | } | |
1329 | ||
494da23a | 1330 | TEST_P(ExternalSSTFileBasicTest, IngestionWithRangeDeletions) { |
11fdf7f2 TL |
1331 | int kNumLevels = 7; |
1332 | Options options = CurrentOptions(); | |
1333 | options.disable_auto_compactions = true; | |
1334 | options.num_levels = kNumLevels; | |
1335 | Reopen(options); | |
1336 | ||
1337 | std::map<std::string, std::string> true_data; | |
1338 | int file_id = 1; | |
1339 | // prevent range deletions from being dropped due to becoming obsolete. | |
1340 | const Snapshot* snapshot = db_->GetSnapshot(); | |
1341 | ||
1342 | // range del [0, 50) in L6 file, [50, 100) in L0 file, [100, 150) in memtable | |
1343 | for (int i = 0; i < 3; i++) { | |
1344 | if (i != 0) { | |
1345 | db_->Flush(FlushOptions()); | |
1346 | if (i == 1) { | |
1347 | MoveFilesToLevel(kNumLevels - 1); | |
1348 | } | |
1349 | } | |
1350 | ASSERT_OK(db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), | |
1351 | Key(50 * i), Key(50 * (i + 1)))); | |
1352 | } | |
1353 | ASSERT_EQ(1, NumTableFilesAtLevel(0)); | |
1354 | ASSERT_EQ(0, NumTableFilesAtLevel(kNumLevels - 2)); | |
1355 | ASSERT_EQ(1, NumTableFilesAtLevel(kNumLevels - 1)); | |
1356 | ||
494da23a TL |
1357 | bool write_global_seqno = std::get<0>(GetParam()); |
1358 | bool verify_checksums_before_ingest = std::get<1>(GetParam()); | |
11fdf7f2 TL |
1359 | // overlaps with L0 file but not memtable, so flush is skipped and file is |
1360 | // ingested into L0 | |
1361 | SequenceNumber last_seqno = dbfull()->GetLatestSequenceNumber(); | |
1362 | ASSERT_OK(GenerateAndAddExternalFile( | |
1363 | options, {60, 90}, {ValueType::kTypeValue, ValueType::kTypeValue}, | |
494da23a TL |
1364 | {{65, 70}, {70, 85}}, file_id++, write_global_seqno, |
1365 | verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 TL |
1366 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), ++last_seqno); |
1367 | ASSERT_EQ(2, NumTableFilesAtLevel(0)); | |
1368 | ASSERT_EQ(0, NumTableFilesAtLevel(kNumLevels - 2)); | |
1369 | ASSERT_EQ(1, NumTableFilesAtLevel(options.num_levels - 1)); | |
1370 | ||
1371 | // overlaps with L6 file but not memtable or L0 file, so flush is skipped and | |
1372 | // file is ingested into L5 | |
1373 | ASSERT_OK(GenerateAndAddExternalFile( | |
1374 | options, {10, 40}, {ValueType::kTypeValue, ValueType::kTypeValue}, | |
494da23a TL |
1375 | file_id++, write_global_seqno, verify_checksums_before_ingest, |
1376 | &true_data)); | |
11fdf7f2 TL |
1377 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), ++last_seqno); |
1378 | ASSERT_EQ(2, NumTableFilesAtLevel(0)); | |
1379 | ASSERT_EQ(1, NumTableFilesAtLevel(kNumLevels - 2)); | |
1380 | ASSERT_EQ(1, NumTableFilesAtLevel(options.num_levels - 1)); | |
1381 | ||
1382 | // overlaps with L5 file but not memtable or L0 file, so flush is skipped and | |
1383 | // file is ingested into L4 | |
494da23a TL |
1384 | ASSERT_OK(GenerateAndAddExternalFile( |
1385 | options, {}, {}, {{5, 15}}, file_id++, write_global_seqno, | |
1386 | verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 TL |
1387 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), ++last_seqno); |
1388 | ASSERT_EQ(2, NumTableFilesAtLevel(0)); | |
1389 | ASSERT_EQ(1, NumTableFilesAtLevel(kNumLevels - 2)); | |
1390 | ASSERT_EQ(1, NumTableFilesAtLevel(options.num_levels - 2)); | |
1391 | ASSERT_EQ(1, NumTableFilesAtLevel(options.num_levels - 1)); | |
1392 | ||
1393 | // ingested file overlaps with memtable, so flush is triggered before the file | |
1394 | // is ingested such that the ingested data is considered newest. So L0 file | |
1395 | // count increases by two. | |
1396 | ASSERT_OK(GenerateAndAddExternalFile( | |
1397 | options, {100, 140}, {ValueType::kTypeValue, ValueType::kTypeValue}, | |
494da23a TL |
1398 | file_id++, write_global_seqno, verify_checksums_before_ingest, |
1399 | &true_data)); | |
11fdf7f2 TL |
1400 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), ++last_seqno); |
1401 | ASSERT_EQ(4, NumTableFilesAtLevel(0)); | |
1402 | ASSERT_EQ(1, NumTableFilesAtLevel(kNumLevels - 2)); | |
1403 | ASSERT_EQ(1, NumTableFilesAtLevel(options.num_levels - 1)); | |
1404 | ||
1405 | // snapshot unneeded now that all range deletions are persisted | |
1406 | db_->ReleaseSnapshot(snapshot); | |
1407 | ||
1408 | // overlaps with nothing, so places at bottom level and skips incrementing | |
1409 | // seqnum. | |
1410 | ASSERT_OK(GenerateAndAddExternalFile( | |
1411 | options, {151, 175}, {ValueType::kTypeValue, ValueType::kTypeValue}, | |
494da23a TL |
1412 | {{160, 200}}, file_id++, write_global_seqno, |
1413 | verify_checksums_before_ingest, &true_data)); | |
11fdf7f2 TL |
1414 | ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), last_seqno); |
1415 | ASSERT_EQ(4, NumTableFilesAtLevel(0)); | |
1416 | ASSERT_EQ(1, NumTableFilesAtLevel(kNumLevels - 2)); | |
1417 | ASSERT_EQ(2, NumTableFilesAtLevel(options.num_levels - 1)); | |
1418 | } | |
1419 | ||
f67539c2 TL |
1420 | TEST_F(ExternalSSTFileBasicTest, AdjacentRangeDeletionTombstones) { |
1421 | Options options = CurrentOptions(); | |
1422 | SstFileWriter sst_file_writer(EnvOptions(), options); | |
1423 | ||
1424 | // file8.sst (delete 300 => 400) | |
1425 | std::string file8 = sst_files_dir_ + "file8.sst"; | |
1426 | ASSERT_OK(sst_file_writer.Open(file8)); | |
1427 | ASSERT_OK(sst_file_writer.DeleteRange(Key(300), Key(400))); | |
1428 | ExternalSstFileInfo file8_info; | |
1429 | Status s = sst_file_writer.Finish(&file8_info); | |
1e59de90 | 1430 | ASSERT_OK(s) << s.ToString(); |
f67539c2 TL |
1431 | ASSERT_EQ(file8_info.file_path, file8); |
1432 | ASSERT_EQ(file8_info.num_entries, 0); | |
1433 | ASSERT_EQ(file8_info.smallest_key, ""); | |
1434 | ASSERT_EQ(file8_info.largest_key, ""); | |
1435 | ASSERT_EQ(file8_info.num_range_del_entries, 1); | |
1436 | ASSERT_EQ(file8_info.smallest_range_del_key, Key(300)); | |
1437 | ASSERT_EQ(file8_info.largest_range_del_key, Key(400)); | |
1438 | ||
1439 | // file9.sst (delete 400 => 500) | |
1440 | std::string file9 = sst_files_dir_ + "file9.sst"; | |
1441 | ASSERT_OK(sst_file_writer.Open(file9)); | |
1442 | ASSERT_OK(sst_file_writer.DeleteRange(Key(400), Key(500))); | |
1443 | ExternalSstFileInfo file9_info; | |
1444 | s = sst_file_writer.Finish(&file9_info); | |
1e59de90 | 1445 | ASSERT_OK(s) << s.ToString(); |
f67539c2 TL |
1446 | ASSERT_EQ(file9_info.file_path, file9); |
1447 | ASSERT_EQ(file9_info.num_entries, 0); | |
1448 | ASSERT_EQ(file9_info.smallest_key, ""); | |
1449 | ASSERT_EQ(file9_info.largest_key, ""); | |
1450 | ASSERT_EQ(file9_info.num_range_del_entries, 1); | |
1451 | ASSERT_EQ(file9_info.smallest_range_del_key, Key(400)); | |
1452 | ASSERT_EQ(file9_info.largest_range_del_key, Key(500)); | |
1453 | ||
1454 | // Range deletion tombstones are exclusive on their end key, so these SSTs | |
1455 | // should not be considered as overlapping. | |
1456 | s = DeprecatedAddFile({file8, file9}); | |
1e59de90 | 1457 | ASSERT_OK(s) << s.ToString(); |
f67539c2 TL |
1458 | ASSERT_EQ(db_->GetLatestSequenceNumber(), 0U); |
1459 | DestroyAndRecreateExternalSSTFilesDir(); | |
1460 | } | |
1461 | ||
494da23a TL |
1462 | TEST_P(ExternalSSTFileBasicTest, IngestFileWithBadBlockChecksum) { |
1463 | bool change_checksum_called = false; | |
1464 | const auto& change_checksum = [&](void* arg) { | |
1465 | if (!change_checksum_called) { | |
1466 | char* buf = reinterpret_cast<char*>(arg); | |
1467 | assert(nullptr != buf); | |
1468 | buf[0] ^= 0x1; | |
1469 | change_checksum_called = true; | |
1470 | } | |
1471 | }; | |
1472 | SyncPoint::GetInstance()->DisableProcessing(); | |
1473 | SyncPoint::GetInstance()->ClearAllCallBacks(); | |
1474 | SyncPoint::GetInstance()->SetCallBack( | |
1e59de90 | 1475 | "BlockBasedTableBuilder::WriteMaybeCompressedBlock:TamperWithChecksum", |
494da23a TL |
1476 | change_checksum); |
1477 | SyncPoint::GetInstance()->EnableProcessing(); | |
1478 | int file_id = 0; | |
1479 | bool write_global_seqno = std::get<0>(GetParam()); | |
1480 | bool verify_checksums_before_ingest = std::get<1>(GetParam()); | |
1481 | do { | |
1482 | Options options = CurrentOptions(); | |
1483 | DestroyAndReopen(options); | |
1484 | std::map<std::string, std::string> true_data; | |
1485 | Status s = GenerateAndAddExternalFile( | |
1486 | options, {1, 2, 3, 4, 5, 6}, ValueType::kTypeValue, file_id++, | |
1487 | write_global_seqno, verify_checksums_before_ingest, &true_data); | |
1488 | if (verify_checksums_before_ingest) { | |
1489 | ASSERT_NOK(s); | |
1490 | } else { | |
1491 | ASSERT_OK(s); | |
1492 | } | |
1493 | change_checksum_called = false; | |
1494 | } while (ChangeOptionsForFileIngestionTest()); | |
1495 | } | |
1496 | ||
1497 | TEST_P(ExternalSSTFileBasicTest, IngestFileWithFirstByteTampered) { | |
1e59de90 TL |
1498 | if (!random_rwfile_supported_) { |
1499 | ROCKSDB_GTEST_SKIP("Test requires NewRandomRWFile support"); | |
1500 | return; | |
1501 | } | |
494da23a TL |
1502 | SyncPoint::GetInstance()->DisableProcessing(); |
1503 | int file_id = 0; | |
1504 | EnvOptions env_options; | |
1505 | do { | |
1506 | Options options = CurrentOptions(); | |
1e59de90 | 1507 | std::string file_path = sst_files_dir_ + std::to_string(file_id++); |
494da23a TL |
1508 | SstFileWriter sst_file_writer(env_options, options); |
1509 | Status s = sst_file_writer.Open(file_path); | |
1510 | ASSERT_OK(s); | |
1511 | for (int i = 0; i != 100; ++i) { | |
1512 | std::string key = Key(i); | |
1e59de90 | 1513 | std::string value = Key(i) + std::to_string(0); |
494da23a TL |
1514 | ASSERT_OK(sst_file_writer.Put(key, value)); |
1515 | } | |
1516 | ASSERT_OK(sst_file_writer.Finish()); | |
1517 | { | |
1518 | // Get file size | |
1519 | uint64_t file_size = 0; | |
1520 | ASSERT_OK(env_->GetFileSize(file_path, &file_size)); | |
1521 | ASSERT_GT(file_size, 8); | |
1522 | std::unique_ptr<RandomRWFile> rwfile; | |
1523 | ASSERT_OK(env_->NewRandomRWFile(file_path, &rwfile, EnvOptions())); | |
1524 | // Manually corrupt the file | |
1525 | // We deterministically corrupt the first byte because we currently | |
1526 | // cannot choose a random offset. The reason for this limitation is that | |
1527 | // we do not checksum property block at present. | |
1528 | const uint64_t offset = 0; | |
1529 | char scratch[8] = {0}; | |
1530 | Slice buf; | |
1531 | ASSERT_OK(rwfile->Read(offset, sizeof(scratch), &buf, scratch)); | |
1532 | scratch[0] ^= 0xff; // flip one bit | |
1533 | ASSERT_OK(rwfile->Write(offset, buf)); | |
1534 | } | |
1535 | // Ingest file. | |
1536 | IngestExternalFileOptions ifo; | |
1537 | ifo.write_global_seqno = std::get<0>(GetParam()); | |
1538 | ifo.verify_checksums_before_ingest = std::get<1>(GetParam()); | |
1539 | s = db_->IngestExternalFile({file_path}, ifo); | |
1540 | if (ifo.verify_checksums_before_ingest) { | |
1541 | ASSERT_NOK(s); | |
1542 | } else { | |
1543 | ASSERT_OK(s); | |
1544 | } | |
1545 | } while (ChangeOptionsForFileIngestionTest()); | |
1546 | } | |
1547 | ||
1548 | TEST_P(ExternalSSTFileBasicTest, IngestExternalFileWithCorruptedPropsBlock) { | |
1549 | bool verify_checksums_before_ingest = std::get<1>(GetParam()); | |
1550 | if (!verify_checksums_before_ingest) { | |
1e59de90 TL |
1551 | ROCKSDB_GTEST_BYPASS("Bypassing test when !verify_checksums_before_ingest"); |
1552 | return; | |
1553 | } | |
1554 | if (!random_rwfile_supported_) { | |
1555 | ROCKSDB_GTEST_SKIP("Test requires NewRandomRWFile support"); | |
494da23a TL |
1556 | return; |
1557 | } | |
1558 | uint64_t props_block_offset = 0; | |
1559 | size_t props_block_size = 0; | |
1560 | const auto& get_props_block_offset = [&](void* arg) { | |
1561 | props_block_offset = *reinterpret_cast<uint64_t*>(arg); | |
1562 | }; | |
1563 | const auto& get_props_block_size = [&](void* arg) { | |
1564 | props_block_size = *reinterpret_cast<uint64_t*>(arg); | |
1565 | }; | |
1566 | SyncPoint::GetInstance()->DisableProcessing(); | |
1567 | SyncPoint::GetInstance()->ClearAllCallBacks(); | |
1568 | SyncPoint::GetInstance()->SetCallBack( | |
1569 | "BlockBasedTableBuilder::WritePropertiesBlock:GetPropsBlockOffset", | |
1570 | get_props_block_offset); | |
1571 | SyncPoint::GetInstance()->SetCallBack( | |
1572 | "BlockBasedTableBuilder::WritePropertiesBlock:GetPropsBlockSize", | |
1573 | get_props_block_size); | |
1574 | SyncPoint::GetInstance()->EnableProcessing(); | |
1575 | int file_id = 0; | |
1576 | Random64 rand(time(nullptr)); | |
1577 | do { | |
1e59de90 | 1578 | std::string file_path = sst_files_dir_ + std::to_string(file_id++); |
494da23a TL |
1579 | Options options = CurrentOptions(); |
1580 | SstFileWriter sst_file_writer(EnvOptions(), options); | |
1581 | Status s = sst_file_writer.Open(file_path); | |
1582 | ASSERT_OK(s); | |
1583 | for (int i = 0; i != 100; ++i) { | |
1584 | std::string key = Key(i); | |
1e59de90 | 1585 | std::string value = Key(i) + std::to_string(0); |
494da23a TL |
1586 | ASSERT_OK(sst_file_writer.Put(key, value)); |
1587 | } | |
1588 | ASSERT_OK(sst_file_writer.Finish()); | |
1589 | ||
1590 | { | |
1591 | std::unique_ptr<RandomRWFile> rwfile; | |
1592 | ASSERT_OK(env_->NewRandomRWFile(file_path, &rwfile, EnvOptions())); | |
1593 | // Manually corrupt the file | |
1594 | ASSERT_GT(props_block_size, 8); | |
1595 | uint64_t offset = | |
1596 | props_block_offset + rand.Next() % (props_block_size - 8); | |
1597 | char scratch[8] = {0}; | |
1598 | Slice buf; | |
1599 | ASSERT_OK(rwfile->Read(offset, sizeof(scratch), &buf, scratch)); | |
1600 | scratch[0] ^= 0xff; // flip one bit | |
1601 | ASSERT_OK(rwfile->Write(offset, buf)); | |
1602 | } | |
1603 | ||
1604 | // Ingest file. | |
1605 | IngestExternalFileOptions ifo; | |
1606 | ifo.write_global_seqno = std::get<0>(GetParam()); | |
1607 | ifo.verify_checksums_before_ingest = true; | |
1608 | s = db_->IngestExternalFile({file_path}, ifo); | |
1609 | ASSERT_NOK(s); | |
1610 | } while (ChangeOptionsForFileIngestionTest()); | |
1611 | } | |
1612 | ||
f67539c2 TL |
1613 | TEST_F(ExternalSSTFileBasicTest, OverlappingFiles) { |
1614 | Options options = CurrentOptions(); | |
1615 | ||
1616 | std::vector<std::string> files; | |
1617 | { | |
1618 | SstFileWriter sst_file_writer(EnvOptions(), options); | |
1619 | std::string file1 = sst_files_dir_ + "file1.sst"; | |
1620 | ASSERT_OK(sst_file_writer.Open(file1)); | |
1621 | ASSERT_OK(sst_file_writer.Put("a", "z")); | |
1622 | ASSERT_OK(sst_file_writer.Put("i", "m")); | |
1623 | ExternalSstFileInfo file1_info; | |
1624 | ASSERT_OK(sst_file_writer.Finish(&file1_info)); | |
1625 | files.push_back(std::move(file1)); | |
1626 | } | |
1627 | { | |
1628 | SstFileWriter sst_file_writer(EnvOptions(), options); | |
1629 | std::string file2 = sst_files_dir_ + "file2.sst"; | |
1630 | ASSERT_OK(sst_file_writer.Open(file2)); | |
1631 | ASSERT_OK(sst_file_writer.Put("i", "k")); | |
1632 | ExternalSstFileInfo file2_info; | |
1633 | ASSERT_OK(sst_file_writer.Finish(&file2_info)); | |
1634 | files.push_back(std::move(file2)); | |
1635 | } | |
1636 | ||
1637 | IngestExternalFileOptions ifo; | |
1638 | ASSERT_OK(db_->IngestExternalFile(files, ifo)); | |
1639 | ASSERT_EQ(Get("a"), "z"); | |
1640 | ASSERT_EQ(Get("i"), "k"); | |
1641 | ||
1642 | int total_keys = 0; | |
1643 | Iterator* iter = db_->NewIterator(ReadOptions()); | |
1644 | for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { | |
1645 | ASSERT_OK(iter->status()); | |
1646 | total_keys++; | |
1647 | } | |
1648 | delete iter; | |
1649 | ASSERT_EQ(total_keys, 2); | |
1650 | ||
1651 | ASSERT_EQ(2, NumTableFilesAtLevel(0)); | |
1652 | } | |
1653 | ||
1e59de90 TL |
1654 | TEST_F(ExternalSSTFileBasicTest, IngestFileAfterDBPut) { |
1655 | // Repro https://github.com/facebook/rocksdb/issues/6245. | |
1656 | // Flush three files to L0. Ingest one more file to trigger L0->L1 compaction | |
1657 | // via trivial move. The bug happened when L1 files were incorrectly sorted | |
1658 | // resulting in an old value for "k" returned by `Get()`. | |
1659 | Options options = CurrentOptions(); | |
1660 | ||
1661 | ASSERT_OK(Put("k", "a")); | |
1662 | Flush(); | |
1663 | ASSERT_OK(Put("k", "a")); | |
1664 | Flush(); | |
1665 | ASSERT_OK(Put("k", "a")); | |
1666 | Flush(); | |
1667 | SstFileWriter sst_file_writer(EnvOptions(), options); | |
1668 | ||
1669 | // Current file size should be 0 after sst_file_writer init and before open a | |
1670 | // file. | |
1671 | ASSERT_EQ(sst_file_writer.FileSize(), 0); | |
1672 | ||
1673 | std::string file1 = sst_files_dir_ + "file1.sst"; | |
1674 | ASSERT_OK(sst_file_writer.Open(file1)); | |
1675 | ASSERT_OK(sst_file_writer.Put("k", "b")); | |
1676 | ||
1677 | ExternalSstFileInfo file1_info; | |
1678 | Status s = sst_file_writer.Finish(&file1_info); | |
1679 | ASSERT_OK(s) << s.ToString(); | |
1680 | ||
1681 | // Current file size should be non-zero after success write. | |
1682 | ASSERT_GT(sst_file_writer.FileSize(), 0); | |
1683 | ||
1684 | IngestExternalFileOptions ifo; | |
1685 | s = db_->IngestExternalFile({file1}, ifo); | |
1686 | ASSERT_OK(s); | |
1687 | ASSERT_OK(dbfull()->TEST_WaitForCompact()); | |
1688 | ||
1689 | ASSERT_EQ(Get("k"), "b"); | |
1690 | } | |
1691 | ||
1692 | TEST_F(ExternalSSTFileBasicTest, IngestWithTemperature) { | |
1693 | Options options = CurrentOptions(); | |
1694 | const ImmutableCFOptions ioptions(options); | |
1695 | options.bottommost_temperature = Temperature::kWarm; | |
1696 | SstFileWriter sst_file_writer(EnvOptions(), options); | |
1697 | options.level0_file_num_compaction_trigger = 2; | |
1698 | Reopen(options); | |
1699 | ||
1700 | auto size = GetSstSizeHelper(Temperature::kUnknown); | |
1701 | ASSERT_EQ(size, 0); | |
1702 | size = GetSstSizeHelper(Temperature::kWarm); | |
1703 | ASSERT_EQ(size, 0); | |
1704 | size = GetSstSizeHelper(Temperature::kHot); | |
1705 | ASSERT_EQ(size, 0); | |
1706 | ||
1707 | // create file01.sst (1000 => 1099) and ingest it | |
1708 | std::string file1 = sst_files_dir_ + "file01.sst"; | |
1709 | ASSERT_OK(sst_file_writer.Open(file1)); | |
1710 | for (int k = 1000; k < 1100; k++) { | |
1711 | ASSERT_OK(sst_file_writer.Put(Key(k), Key(k) + "_val")); | |
1712 | } | |
1713 | ExternalSstFileInfo file1_info; | |
1714 | Status s = sst_file_writer.Finish(&file1_info); | |
1715 | ASSERT_OK(s); | |
1716 | ASSERT_EQ(file1_info.file_path, file1); | |
1717 | ASSERT_EQ(file1_info.num_entries, 100); | |
1718 | ASSERT_EQ(file1_info.smallest_key, Key(1000)); | |
1719 | ASSERT_EQ(file1_info.largest_key, Key(1099)); | |
1720 | ||
1721 | std::vector<std::string> files; | |
1722 | std::vector<std::string> files_checksums; | |
1723 | std::vector<std::string> files_checksum_func_names; | |
1724 | Temperature file_temperature = Temperature::kWarm; | |
1725 | ||
1726 | files.push_back(file1); | |
1727 | IngestExternalFileOptions in_opts; | |
1728 | in_opts.move_files = false; | |
1729 | in_opts.snapshot_consistency = true; | |
1730 | in_opts.allow_global_seqno = false; | |
1731 | in_opts.allow_blocking_flush = false; | |
1732 | in_opts.write_global_seqno = true; | |
1733 | in_opts.verify_file_checksum = false; | |
1734 | IngestExternalFileArg arg; | |
1735 | arg.column_family = db_->DefaultColumnFamily(); | |
1736 | arg.external_files = files; | |
1737 | arg.options = in_opts; | |
1738 | arg.files_checksums = files_checksums; | |
1739 | arg.files_checksum_func_names = files_checksum_func_names; | |
1740 | arg.file_temperature = file_temperature; | |
1741 | s = db_->IngestExternalFiles({arg}); | |
1742 | ASSERT_OK(s); | |
1743 | ||
1744 | // check the temperature of the file being ingested | |
1745 | ColumnFamilyMetaData metadata; | |
1746 | db_->GetColumnFamilyMetaData(&metadata); | |
1747 | ASSERT_EQ(1, metadata.file_count); | |
1748 | ASSERT_EQ(Temperature::kWarm, metadata.levels[6].files[0].temperature); | |
1749 | size = GetSstSizeHelper(Temperature::kUnknown); | |
1750 | ASSERT_EQ(size, 0); | |
1751 | size = GetSstSizeHelper(Temperature::kWarm); | |
1752 | ASSERT_GT(size, 1); | |
1753 | ||
1754 | // non-bottommost file still has unknown temperature | |
1755 | ASSERT_OK(Put("foo", "bar")); | |
1756 | ASSERT_OK(Put("bar", "bar")); | |
1757 | ASSERT_OK(Flush()); | |
1758 | db_->GetColumnFamilyMetaData(&metadata); | |
1759 | ASSERT_EQ(2, metadata.file_count); | |
1760 | ASSERT_EQ(Temperature::kUnknown, metadata.levels[0].files[0].temperature); | |
1761 | size = GetSstSizeHelper(Temperature::kUnknown); | |
1762 | ASSERT_GT(size, 0); | |
1763 | size = GetSstSizeHelper(Temperature::kWarm); | |
1764 | ASSERT_GT(size, 0); | |
1765 | ||
1766 | // reopen and check the information is persisted | |
1767 | Reopen(options); | |
1768 | db_->GetColumnFamilyMetaData(&metadata); | |
1769 | ASSERT_EQ(2, metadata.file_count); | |
1770 | ASSERT_EQ(Temperature::kUnknown, metadata.levels[0].files[0].temperature); | |
1771 | ASSERT_EQ(Temperature::kWarm, metadata.levels[6].files[0].temperature); | |
1772 | size = GetSstSizeHelper(Temperature::kUnknown); | |
1773 | ASSERT_GT(size, 0); | |
1774 | size = GetSstSizeHelper(Temperature::kWarm); | |
1775 | ASSERT_GT(size, 0); | |
1776 | ||
1777 | // check other non-exist temperatures | |
1778 | size = GetSstSizeHelper(Temperature::kHot); | |
1779 | ASSERT_EQ(size, 0); | |
1780 | size = GetSstSizeHelper(Temperature::kCold); | |
1781 | ASSERT_EQ(size, 0); | |
1782 | std::string prop; | |
1783 | ASSERT_TRUE(dbfull()->GetProperty( | |
1784 | DB::Properties::kLiveSstFilesSizeAtTemperature + std::to_string(22), | |
1785 | &prop)); | |
1786 | ASSERT_EQ(std::atoi(prop.c_str()), 0); | |
1787 | } | |
1788 | ||
1789 | TEST_F(ExternalSSTFileBasicTest, FailIfNotBottommostLevel) { | |
1790 | Options options = GetDefaultOptions(); | |
1791 | ||
1792 | std::string file_path = sst_files_dir_ + std::to_string(1); | |
1793 | SstFileWriter sfw(EnvOptions(), options); | |
1794 | ||
1795 | ASSERT_OK(sfw.Open(file_path)); | |
1796 | ASSERT_OK(sfw.Put("b", "dontcare")); | |
1797 | ASSERT_OK(sfw.Finish()); | |
1798 | ||
1799 | // Test universal compaction + ingest with snapshot consistency | |
1800 | options.create_if_missing = true; | |
1801 | options.compaction_style = CompactionStyle::kCompactionStyleUniversal; | |
1802 | DestroyAndReopen(options); | |
1803 | { | |
1804 | const Snapshot* snapshot = db_->GetSnapshot(); | |
1805 | ManagedSnapshot snapshot_guard(db_, snapshot); | |
1806 | IngestExternalFileOptions ifo; | |
1807 | ifo.fail_if_not_bottommost_level = true; | |
1808 | ifo.snapshot_consistency = true; | |
1809 | const Status s = db_->IngestExternalFile({file_path}, ifo); | |
1810 | ASSERT_TRUE(s.IsTryAgain()); | |
1811 | } | |
1812 | ||
1813 | // Test level compaction | |
1814 | options.compaction_style = CompactionStyle::kCompactionStyleLevel; | |
1815 | options.num_levels = 2; | |
1816 | DestroyAndReopen(options); | |
1817 | ASSERT_OK(db_->Put(WriteOptions(), "a", "dontcare")); | |
1818 | ASSERT_OK(db_->Put(WriteOptions(), "c", "dontcare")); | |
1819 | ASSERT_OK(db_->Flush(FlushOptions())); | |
1820 | ||
1821 | ASSERT_OK(db_->Put(WriteOptions(), "b", "dontcare")); | |
1822 | ASSERT_OK(db_->Put(WriteOptions(), "d", "dontcare")); | |
1823 | ASSERT_OK(db_->Flush(FlushOptions())); | |
1824 | ||
1825 | { | |
1826 | CompactRangeOptions cro; | |
1827 | cro.bottommost_level_compaction = BottommostLevelCompaction::kForce; | |
1828 | ASSERT_OK(db_->CompactRange(cro, nullptr, nullptr)); | |
1829 | ||
1830 | IngestExternalFileOptions ifo; | |
1831 | ifo.fail_if_not_bottommost_level = true; | |
1832 | const Status s = db_->IngestExternalFile({file_path}, ifo); | |
1833 | ASSERT_TRUE(s.IsTryAgain()); | |
1834 | } | |
1835 | } | |
1836 | ||
1837 | TEST_F(ExternalSSTFileBasicTest, VerifyChecksum) { | |
1838 | const std::string kPutVal = "put_val"; | |
1839 | const std::string kIngestedVal = "ingested_val"; | |
1840 | ||
1841 | ASSERT_OK(Put("k", kPutVal, WriteOptions())); | |
1842 | ASSERT_OK(Flush()); | |
1843 | ||
1844 | std::string external_file = sst_files_dir_ + "/file_to_ingest.sst"; | |
1845 | { | |
1846 | SstFileWriter sst_file_writer{EnvOptions(), CurrentOptions()}; | |
1847 | ||
1848 | ASSERT_OK(sst_file_writer.Open(external_file)); | |
1849 | ASSERT_OK(sst_file_writer.Put("k", kIngestedVal)); | |
1850 | ASSERT_OK(sst_file_writer.Finish()); | |
1851 | } | |
1852 | ||
1853 | ASSERT_OK(db_->IngestExternalFile(db_->DefaultColumnFamily(), {external_file}, | |
1854 | IngestExternalFileOptions())); | |
1855 | ||
1856 | ASSERT_OK(db_->VerifyChecksum()); | |
1857 | } | |
1858 | ||
1859 | TEST_F(ExternalSSTFileBasicTest, VerifySstUniqueId) { | |
1860 | const std::string kPutVal = "put_val"; | |
1861 | const std::string kIngestedVal = "ingested_val"; | |
1862 | ||
1863 | ASSERT_OK(Put("k", kPutVal, WriteOptions())); | |
1864 | ASSERT_OK(Flush()); | |
1865 | ||
1866 | std::string external_file = sst_files_dir_ + "/file_to_ingest.sst"; | |
1867 | { | |
1868 | SstFileWriter sst_file_writer{EnvOptions(), CurrentOptions()}; | |
1869 | ||
1870 | ASSERT_OK(sst_file_writer.Open(external_file)); | |
1871 | ASSERT_OK(sst_file_writer.Put("k", kIngestedVal)); | |
1872 | ASSERT_OK(sst_file_writer.Finish()); | |
1873 | } | |
1874 | ||
1875 | ASSERT_OK(db_->IngestExternalFile(db_->DefaultColumnFamily(), {external_file}, | |
1876 | IngestExternalFileOptions())); | |
1877 | ||
1878 | // Test ingest file without session_id and db_id (for example generated by an | |
1879 | // older version of sst_writer) | |
1880 | SyncPoint::GetInstance()->SetCallBack( | |
1881 | "PropertyBlockBuilder::AddTableProperty:Start", [&](void* props_vs) { | |
1882 | auto props = static_cast<TableProperties*>(props_vs); | |
1883 | // update table property session_id to a different one | |
1884 | props->db_session_id = ""; | |
1885 | props->db_id = ""; | |
1886 | }); | |
1887 | std::atomic_int skipped = 0, passed = 0; | |
1888 | SyncPoint::GetInstance()->SetCallBack( | |
1889 | "BlockBasedTable::Open::SkippedVerifyUniqueId", | |
1890 | [&](void* /*arg*/) { skipped++; }); | |
1891 | SyncPoint::GetInstance()->SetCallBack( | |
1892 | "BlockBasedTable::Open::PassedVerifyUniqueId", | |
1893 | [&](void* /*arg*/) { passed++; }); | |
1894 | SyncPoint::GetInstance()->EnableProcessing(); | |
1895 | ||
1896 | auto options = CurrentOptions(); | |
1897 | ASSERT_TRUE(options.verify_sst_unique_id_in_manifest); | |
1898 | Reopen(options); | |
1899 | ASSERT_EQ(skipped, 0); | |
1900 | ASSERT_EQ(passed, 2); // one flushed + one ingested | |
1901 | ||
1902 | external_file = sst_files_dir_ + "/file_to_ingest2.sst"; | |
1903 | { | |
1904 | SstFileWriter sst_file_writer{EnvOptions(), CurrentOptions()}; | |
1905 | ||
1906 | ASSERT_OK(sst_file_writer.Open(external_file)); | |
1907 | ASSERT_OK(sst_file_writer.Put("k", kIngestedVal)); | |
1908 | ASSERT_OK(sst_file_writer.Finish()); | |
1909 | } | |
1910 | ||
1911 | ASSERT_OK(db_->IngestExternalFile(db_->DefaultColumnFamily(), {external_file}, | |
1912 | IngestExternalFileOptions())); | |
1913 | ||
1914 | // Two table file opens skipping verification: | |
1915 | // * ExternalSstFileIngestionJob::GetIngestedFileInfo | |
1916 | // * TableCache::GetTableReader | |
1917 | ASSERT_EQ(skipped, 2); | |
1918 | ASSERT_EQ(passed, 2); | |
1919 | ||
1920 | // Check same after re-open (except no GetIngestedFileInfo) | |
1921 | skipped = 0; | |
1922 | passed = 0; | |
1923 | Reopen(options); | |
1924 | ASSERT_EQ(skipped, 1); | |
1925 | ASSERT_EQ(passed, 2); | |
1926 | } | |
1927 | ||
1928 | TEST_F(ExternalSSTFileBasicTest, StableSnapshotWhileLoggingToManifest) { | |
1929 | const std::string kPutVal = "put_val"; | |
1930 | const std::string kIngestedVal = "ingested_val"; | |
1931 | ||
1932 | ASSERT_OK(Put("k", kPutVal, WriteOptions())); | |
1933 | ASSERT_OK(Flush()); | |
1934 | ||
1935 | std::string external_file = sst_files_dir_ + "/file_to_ingest.sst"; | |
1936 | { | |
1937 | SstFileWriter sst_file_writer{EnvOptions(), CurrentOptions()}; | |
1938 | ASSERT_OK(sst_file_writer.Open(external_file)); | |
1939 | ASSERT_OK(sst_file_writer.Put("k", kIngestedVal)); | |
1940 | ASSERT_OK(sst_file_writer.Finish()); | |
1941 | } | |
1942 | ||
1943 | const Snapshot* snapshot = nullptr; | |
1944 | ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack( | |
1945 | "VersionSet::LogAndApply:WriteManifest", [&](void* /* arg */) { | |
1946 | // prevent background compaction job to call this callback | |
1947 | ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing(); | |
1948 | snapshot = db_->GetSnapshot(); | |
1949 | ReadOptions read_opts; | |
1950 | read_opts.snapshot = snapshot; | |
1951 | std::string value; | |
1952 | ASSERT_OK(db_->Get(read_opts, "k", &value)); | |
1953 | ASSERT_EQ(kPutVal, value); | |
1954 | }); | |
1955 | ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing(); | |
1956 | ||
1957 | ASSERT_OK(db_->IngestExternalFile(db_->DefaultColumnFamily(), {external_file}, | |
1958 | IngestExternalFileOptions())); | |
1959 | auto ingested_file_seqno = db_->GetLatestSequenceNumber(); | |
1960 | ASSERT_NE(nullptr, snapshot); | |
1961 | // snapshot is taken before SST ingestion is done | |
1962 | ASSERT_EQ(ingested_file_seqno, snapshot->GetSequenceNumber() + 1); | |
1963 | ||
1964 | ReadOptions read_opts; | |
1965 | read_opts.snapshot = snapshot; | |
1966 | std::string value; | |
1967 | ASSERT_OK(db_->Get(read_opts, "k", &value)); | |
1968 | ASSERT_EQ(kPutVal, value); | |
1969 | db_->ReleaseSnapshot(snapshot); | |
1970 | ||
1971 | // After reopen, sequence number should be up current such that | |
1972 | // ingested value is read | |
1973 | Reopen(CurrentOptions()); | |
1974 | ASSERT_OK(db_->Get(ReadOptions(), "k", &value)); | |
1975 | ASSERT_EQ(kIngestedVal, value); | |
1976 | ||
1977 | // New write should get higher seqno compared to ingested file | |
1978 | ASSERT_OK(Put("k", kPutVal, WriteOptions())); | |
1979 | ASSERT_EQ(db_->GetLatestSequenceNumber(), ingested_file_seqno + 1); | |
1980 | } | |
1981 | ||
494da23a TL |
1982 | INSTANTIATE_TEST_CASE_P(ExternalSSTFileBasicTest, ExternalSSTFileBasicTest, |
1983 | testing::Values(std::make_tuple(true, true), | |
1984 | std::make_tuple(true, false), | |
1985 | std::make_tuple(false, true), | |
1986 | std::make_tuple(false, false))); | |
1987 | ||
7c673cae FG |
1988 | #endif // ROCKSDB_LITE |
1989 | ||
f67539c2 | 1990 | } // namespace ROCKSDB_NAMESPACE |
7c673cae FG |
1991 | |
1992 | int main(int argc, char** argv) { | |
f67539c2 | 1993 | ROCKSDB_NAMESPACE::port::InstallStackTraceHandler(); |
7c673cae | 1994 | ::testing::InitGoogleTest(&argc, argv); |
1e59de90 | 1995 | RegisterCustomObjects(argc, argv); |
7c673cae FG |
1996 | return RUN_ALL_TESTS(); |
1997 | } |