2553 Commits (7f5b9f40cb9a888447b7097cc5b834417d9fad66)

Author	SHA1	Message	Date
sdong	2800aa069a	Remove compressed block cache (#11117) ... Summary: Compressed block cache is replaced by compressed secondary cache. Remove the feature. Pull Request resolved: https://github.com/facebook/rocksdb/pull/11117 Test Plan: See CI passes Reviewed By: pdillinger Differential Revision: D42700164 fbshipit-source-id: 6cbb24e460da29311150865f60ecb98637f9f67d	2 years ago
Peter Dillinger	4a9185340d	A better contract for best_efforts_recovery (#11085) ... Summary: Capture more of the original intent at a high level, without getting bogged down in low-level details. The old text made some weak promises about handling of LOCK files. There should be no specific concern for LOCK files, because we already rely on LockFile() to create the file if it's not present already. And the lock file is generally size 0, so don't have to worry about truncation. Added a unit test. Pull Request resolved: https://github.com/facebook/rocksdb/pull/11085 Test Plan: existing tests, and a new one. Reviewed By: siying Differential Revision: D42713233 Pulled By: pdillinger fbshipit-source-id: 2fce7c974d35fac065037c9c4c7326a59c9fe340	2 years ago
Changyu Bi	e0ea0dc6bd	Improve documentation for `allow_ingest_behind` (#11119) ... Summary: update documentation to mention that only universal compaction is supported. Pull Request resolved: https://github.com/facebook/rocksdb/pull/11119 Reviewed By: ajkr Differential Revision: D42715986 Pulled By: cbi42 fbshipit-source-id: 91b145d3318334cb92857c5c0ffc0efed6fa4363	2 years ago
Hui Xiao	7e7548477c	Update HISTORY.md/version.h/format compatiblity test for 7.10 release (#11114) ... Summary: Pull Request resolved: https://github.com/facebook/rocksdb/pull/11114 Reviewed By: ajkr Differential Revision: D42685234 Pulled By: hx235 fbshipit-source-id: 79908a66ab9052a2552f080049065462ebf2f94c	2 years ago
Andrew Kryczka	b7fbcefda8	Add API to limit blast radius of merge operator failure (#11092) ... Summary: Prior to this PR, `FullMergeV2()` can only return `false` to indicate failure, which causes any operation invoking it to fail. During a compaction, such a failure causes the compaction to fail and causes the DB to irreversibly enter read-only mode. Some users asked for a way to allow the merge operator to fail without such widespread damage. To limit the blast radius of merge operator failures, this PR introduces the `MergeOperationOutput::op_failure_scope` API. When unpopulated (`kDefault`) or set to `kTryMerge`, the merge operator failure handling is the same as before. When set to `kMustMerge`, merge operator failure still causes failure to operations that must merge (`Get()`, iterator, `MultiGet()`, etc.). However, under `kMustMerge`, flushes/compactions can survive merge operator failures by outputting the unmerged input operands. Pull Request resolved: https://github.com/facebook/rocksdb/pull/11092 Reviewed By: siying Differential Revision: D42525673 Pulled By: ajkr fbshipit-source-id: 951dc3bf190f86347dccf3381be967565cda52ee	2 years ago
Peter Dillinger	9f7801c5f1	Major Cache refactoring, CPU efficiency improvement (#10975) ... Summary: This is several refactorings bundled into one to avoid having to incrementally re-modify uses of Cache several times. Overall, there are breaking changes to Cache class, and it becomes more of low-level interface for implementing caches, especially block cache. New internal APIs make using Cache cleaner than before, and more insulated from block cache evolution. Hopefully, this is the last really big block cache refactoring, because of rather effectively decoupling the implementations from the uses. This change also removes the EXPERIMENTAL designation on the SecondaryCache support in Cache. It seems reasonably mature at this point but still subject to change/evolution (as I warn in the API docs for Cache). The high-level motivation for this refactoring is to minimize code duplication / compounding complexity in adding SecondaryCache support to HyperClockCache (in a later PR). Other benefits listed below. * static_cast lines of code +29 -35 (net removed 6) * reinterpret_cast lines of code +6 -32 (net removed 26) ## cache.h and secondary_cache.h * Always use CacheItemHelper with entries instead of just a Deleter. There are several motivations / justifications: * Simpler for implementations to deal with just one Insert and one Lookup. * Simpler and more efficient implementation because we don't have to track which entries are using helpers and which are using deleters * Gets rid of hack to classify cache entries by their deleter. Instead, the CacheItemHelper includes a CacheEntryRole. This simplifies a lot of code (cache_entry_roles.h almost eliminated). Fixes https://github.com/facebook/rocksdb/issues/9428. * Makes it trivial to adjust SecondaryCache behavior based on kind of block (e.g. don't re-compress filter blocks). * It is arguably less convenient for many direct users of Cache, but direct users of Cache are now rare with introduction of typed_cache.h (below). * I considered and rejected an alternative approach in which we reduce customizability by assuming each secondary cache compatible value starts with a Slice referencing the uncompressed block contents (already true or mostly true), but we apparently intend to stack secondary caches. Saving an entry from a compressed secondary to a lower tier requires custom handling offered by SaveToCallback, etc. * Make CreateCallback part of the helper and introduce CreateContext to work with it (alternative to https://github.com/facebook/rocksdb/issues/10562). This cleans up the interface while still allowing context to be provided for loading/parsing values into primary cache. This model works for async lookup in BlockBasedTable reader (reader owns a CreateContext) under the assumption that it always waits on secondary cache operations to finish. (Otherwise, the CreateContext could be destroyed while async operation depending on it continues.) This likely contributes most to the observed performance improvement because it saves an std::function backed by a heap allocation. * Use char* for serialized data, e.g. in SaveToCallback, where void* was confusingly used. (We use `char*` for serialized byte data all over RocksDB, with many advantages over `void*`. `memcpy` etc. are legacy APIs that should not be mimicked.) * Add a type alias Cache::ObjectPtr = void*, so that we can better indicate the intent of the void* when it is to be the object associated with a Cache entry. Related: started (but did not complete) a refactoring to move away from "value" of a cache entry toward "object" or "obj". (It is confusing to call Cache a key-value store (like DB) when it is really storing arbitrary in-memory objects, not byte strings.) * Remove unnecessary key param from DeleterFn. This is good for efficiency in HyperClockCache, which does not directly store the cache key in memory. (Alternative to https://github.com/facebook/rocksdb/issues/10774) * Add allocator to Cache DeleterFn. This is a kind of future-proofing change in case we get more serious about using the Cache allocator for memory tracked by the Cache. Right now, only the uncompressed block contents are allocated using the allocator, and a pointer to that allocator is saved as part of the cached object so that the deleter can use it. (See CacheAllocationPtr.) If in the future we are able to "flatten out" our Cache objects some more, it would be good not to have to track the allocator as part of each object. * Removes legacy `ApplyToAllCacheEntries` and changes `ApplyToAllEntries` signature for Deleter->CacheItemHelper change. ## typed_cache.h Adds various "typed" interfaces to the Cache as internal APIs, so that most uses of Cache can use simple type safe code without casting and without explicit deleters, etc. Almost all of the non-test, non-glue code uses of Cache have been migrated. (Follow-up work: CompressedSecondaryCache deserves deeper attention to migrate.) This change expands RocksDB's internal usage of metaprogramming and SFINAE (https://en.cppreference.com/w/cpp/language/sfinae). The existing usages of Cache are divided up at a high level into these new interfaces. See updated existing uses of Cache for examples of how these are used. * PlaceholderCacheInterface - Used for making cache reservations, with entries that have a charge but no value. * BasicTypedCacheInterface<TValue> - Used for primary cache storage of objects of type TValue, which can be cleaned up with std::default_delete<TValue>. The role is provided by TValue::kCacheEntryRole or given in an optional template parameter. * FullTypedCacheInterface<TValue, TCreateContext> - Used for secondary cache compatible storage of objects of type TValue. In addition to BasicTypedCacheInterface constraints, we require TValue::ContentSlice() to return persistable data. This simplifies usage for the normal case of simple secondary cache compatibility (can give you a Slice to the data already in memory). In addition to TCreateContext performing the role of Cache::CreateContext, it is also expected to provide a factory function for creating TValue. * For each of these, there's a "Shared" version (e.g. FullTypedSharedCacheInterface) that holds a shared_ptr to the Cache, rather than assuming external ownership by holding only a raw `Cache*`. These interfaces introduce specific handle types for each interface instantiation, so that it's easy to see what kind of object is controlled by a handle. (Ultimately, this might not be worth the extra complexity, but it seems OK so far.) Note: I attempted to make the cache 'charge' automatically inferred from the cache object type, such as by expecting an ApproximateMemoryUsage() function, but this is not so clean because there are cases where we need to compute the charge ahead of time and don't want to re-compute it. ## block_cache.h This header is essentially the replacement for the old block_like_traits.h. It includes various things to support block cache access with typed_cache.h for block-based table. ## block_based_table_reader.cc Before this change, accessing the block cache here was an awkward mix of static polymorphism (template TBlocklike) and switch-case on a dynamic BlockType value. This change mostly unifies on static polymorphism, relying on minor hacks in block_cache.h to distinguish variants of Block. We still check BlockType in some places (especially for stats, which could be improved in follow-up work) but at least the BlockType is a static constant from the template parameter. (No more awkward partial redundancy between static and dynamic info.) This likely contributes to the overall performance improvement, but hasn't been tested in isolation. The other key source of simplification here is a more unified system of creating block cache objects: for directly populating from primary cache and for promotion from secondary cache. Both use BlockCreateContext, for context and for factory functions. ## block_based_table_builder.cc, cache_dump_load_impl.cc Before this change, warming caches was super ugly code. Both of these source files had switch statements to basically transition from the dynamic BlockType world to the static TBlocklike world. None of that mess is needed anymore as there's a new, untyped WarmInCache function that handles all the details just as promotion from SecondaryCache would. (Fixes `TODO akanksha: Dedup below code` in block_based_table_builder.cc.) ## Everything else Mostly just updating Cache users to use new typed APIs when reasonably possible, or changed Cache APIs when not. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10975 Test Plan: tests updated Performance test setup similar to https://github.com/facebook/rocksdb/issues/10626 (by cache size, LRUCache when not "hyper" for HyperClockCache): 34MB 1thread base.hyper -> kops/s: 0.745 io_bytes/op: 2.52504e+06 miss_ratio: 0.140906 max_rss_mb: 76.4844 34MB 1thread new.hyper -> kops/s: 0.751 io_bytes/op: 2.5123e+06 miss_ratio: 0.140161 max_rss_mb: 79.3594 34MB 1thread base -> kops/s: 0.254 io_bytes/op: 1.36073e+07 miss_ratio: 0.918818 max_rss_mb: 45.9297 34MB 1thread new -> kops/s: 0.252 io_bytes/op: 1.36157e+07 miss_ratio: 0.918999 max_rss_mb: 44.1523 34MB 32thread base.hyper -> kops/s: 7.272 io_bytes/op: 2.88323e+06 miss_ratio: 0.162532 max_rss_mb: 516.602 34MB 32thread new.hyper -> kops/s: 7.214 io_bytes/op: 2.99046e+06 miss_ratio: 0.168818 max_rss_mb: 518.293 34MB 32thread base -> kops/s: 3.528 io_bytes/op: 1.35722e+07 miss_ratio: 0.914691 max_rss_mb: 264.926 34MB 32thread new -> kops/s: 3.604 io_bytes/op: 1.35744e+07 miss_ratio: 0.915054 max_rss_mb: 264.488 233MB 1thread base.hyper -> kops/s: 53.909 io_bytes/op: 2552.35 miss_ratio: 0.0440566 max_rss_mb: 241.984 233MB 1thread new.hyper -> kops/s: 62.792 io_bytes/op: 2549.79 miss_ratio: 0.044043 max_rss_mb: 241.922 233MB 1thread base -> kops/s: 1.197 io_bytes/op: 2.75173e+06 miss_ratio: 0.103093 max_rss_mb: 241.559 233MB 1thread new -> kops/s: 1.199 io_bytes/op: 2.73723e+06 miss_ratio: 0.10305 max_rss_mb: 240.93 233MB 32thread base.hyper -> kops/s: 1298.69 io_bytes/op: 2539.12 miss_ratio: 0.0440307 max_rss_mb: 371.418 233MB 32thread new.hyper -> kops/s: 1421.35 io_bytes/op: 2538.75 miss_ratio: 0.0440307 max_rss_mb: 347.273 233MB 32thread base -> kops/s: 9.693 io_bytes/op: 2.77304e+06 miss_ratio: 0.103745 max_rss_mb: 569.691 233MB 32thread new -> kops/s: 9.75 io_bytes/op: 2.77559e+06 miss_ratio: 0.103798 max_rss_mb: 552.82 1597MB 1thread base.hyper -> kops/s: 58.607 io_bytes/op: 1449.14 miss_ratio: 0.0249324 max_rss_mb: 1583.55 1597MB 1thread new.hyper -> kops/s: 69.6 io_bytes/op: 1434.89 miss_ratio: 0.0247167 max_rss_mb: 1584.02 1597MB 1thread base -> kops/s: 60.478 io_bytes/op: 1421.28 miss_ratio: 0.024452 max_rss_mb: 1589.45 1597MB 1thread new -> kops/s: 63.973 io_bytes/op: 1416.07 miss_ratio: 0.0243766 max_rss_mb: 1589.24 1597MB 32thread base.hyper -> kops/s: 1436.2 io_bytes/op: 1357.93 miss_ratio: 0.0235353 max_rss_mb: 1692.92 1597MB 32thread new.hyper -> kops/s: 1605.03 io_bytes/op: 1358.04 miss_ratio: 0.023538 max_rss_mb: 1702.78 1597MB 32thread base -> kops/s: 280.059 io_bytes/op: 1350.34 miss_ratio: 0.023289 max_rss_mb: 1675.36 1597MB 32thread new -> kops/s: 283.125 io_bytes/op: 1351.05 miss_ratio: 0.0232797 max_rss_mb: 1703.83 Almost uniformly improving over base revision, especially for hot paths with HyperClockCache, up to 12% higher throughput seen (1597MB, 32thread, hyper). The improvement for that is likely coming from much simplified code for providing context for secondary cache promotion (CreateCallback/CreateContext), and possibly from less branching in block_based_table_reader. And likely a small improvement from not reconstituting key for DeleterFn. Reviewed By: anand1976 Differential Revision: D42417818 Pulled By: pdillinger fbshipit-source-id: f86bfdd584dce27c028b151ba56818ad14f7a432	2 years ago
Niklas Fiekas	ff04fb154b	Add C API for ReadOptions::async_io (#11062) ... Summary: Pull Request resolved: https://github.com/facebook/rocksdb/pull/11062 Reviewed By: hx235 Differential Revision: D42297489 Pulled By: ajkr fbshipit-source-id: 03fe1477c1ae1f8af73dc77a6986fdc7025edf4f	2 years ago
HuangYi	33aca893c2	add c-api for setting option optimize_filters_for_memory (#11044) ... Summary: Pull Request resolved: https://github.com/facebook/rocksdb/pull/11044 Reviewed By: cbi42 Differential Revision: D42152851 Pulled By: ajkr fbshipit-source-id: 81710d9503ba4f23f112c72ebf16a48112e27158	2 years ago
Hui Xiao	9502856edd	Add missing range conflict check between file ingestion and RefitLevel() (#10988) ... Summary: **Context:** File ingestion never checks whether the key range it acts on overlaps with an ongoing RefitLevel() (used in `CompactRange()` with `change_level=true`). That's because RefitLevel() doesn't register and make its key range known to file ingestion. Though it checks overlapping with other compactions by https://github.com/facebook/rocksdb/blob/7.8.fb/db/external_sst_file_ingestion_job.cc#L998. RefitLevel() (used in `CompactRange()` with `change_level=true`) doesn't check whether the key range it acts on overlaps with an ongoing file ingestion. That's because file ingestion does not register and make its key range known to other compactions. - Note that non-refitlevel-compaction (e.g, manual compaction w/o RefitLevel() or general compaction) also does not check key range overlap with ongoing file ingestion for the same reason. - But it's fine. Credited to cbi42's discovery, `WaitForIngestFile` was called by background and foreground compactions. They were introduced in 0f88160f67, 5c64fb67d2 and 87dfc1d23e. - Regardless, this PR registers file ingestion like a compaction is a general approach that will also add range conflict check between file ingestion and non-refitlevel-compaction, though it has not been the issue motivated this PR. Above are bugs resulting in two bad consequences: - If file ingestion and RefitLevel() creates files in the same level, then range-overlapped files will be created at that level and caught as corruption by `force_consistency_checks=true` - If file ingestion and RefitLevel() creates file in different levels, then with one further compaction on the ingested file, it can result in two same keys both with seqno 0 in two different levels. Then with iterator's [optimization](c62f322169/db/db_iter.cc (L342-L343)) that assumes no two same keys both with seqno 0, it will either break this assertion in debug build or, even worst, return value of this same key for the key after it, which is the wrong value to return, in release build. Therefore we decide to introduce range conflict check for file ingestion and RefitLevel() inspired from the existing range conflict check among compactions. **Summary:** - Treat file ingestion job and RefitLevel() as `Compaction` of new compaction reasons: `CompactionReason::kExternalSstIngestion` and `CompactionReason::kRefitLevel` and register/unregister them. File ingestion is treated as compaction from L0 to different levels and RefitLevel() as compaction from source level to target level. - Check for `RangeOverlapWithCompaction` with other ongoing compactions, `RegisterCompaction()` on this "compaction" before changing the LSM state in `VersionStorageInfo`, and `UnregisterCompaction()` after changing. - Replace scattered fixes (0f88160f67, 5c64fb67d2 and 87dfc1d23e.) that prevents overlapping between file ingestion and non-refit-level compaction with this fix cuz those practices are easy to overlook. - Misc: logic cleanup, see PR comments Pull Request resolved: https://github.com/facebook/rocksdb/pull/10988 Test Plan: - New unit test `DBCompactionTestWithOngoingFileIngestionParam*` that failed pre-fix and passed afterwards. - Made compatible with existing tests, see PR comments - make check - [Ongoing] Stress test rehearsal with normal value and aggressive CI value https://github.com/facebook/rocksdb/pull/10761 Reviewed By: cbi42 Differential Revision: D41535685 Pulled By: hx235 fbshipit-source-id: 549833a577ba1496d20a870583d4caa737da1258	2 years ago
Peter Dillinger	02f2b20864	Add BackupEngine feature to exclude files (#11030) ... Summary: We have a request for RocksDB to essentially support disconnected incremental backup. In other words, if there is limited or no connectivity to the primary backup dir, we should still be able to take an incremental backup relative to that primary backup dir, assuming some metadata about that primary dir is available (and obviously anticipating primary backup dir will be fully available if restore is needed). To support that, this feature allows the API user to "exclude" DB files from backup. This only applies to files that can be shared between backups (sst and blob files), and excluded files are tracked in the backup metadata sufficiently to ensure they are restored at restore time. At restore time, the user provides a set of alternate backup directories (as open BackupEngines, which can be read-only), and excluded files must be found in one of the backup directories ("included" in some backup). This feature depends on backup schema version 2 features, though schema version 2.0 support is not sufficient to read / restore a backup with exclusions. This change updates the schema version to 2.1 because of this feature, so that it's easy to recognize whether a RocksDB release supports this feature, while backups not using the feature are fully compatible with 2.0. Also in this PR: * Stacked on https://github.com/facebook/rocksdb/pull/11029 * Allow progress_callback to be empty, not just no-op function, and recover from exceptions thrown by BackupEngine callbacks. * The internal-only `AsBackupEngine()` function is working around the diamond hierarchy of `BackupEngineImplThreadSafe` to get to the internals, without using confusing features like virtual inheritance. Pull Request resolved: https://github.com/facebook/rocksdb/pull/11030 Test Plan: unit tests added / updated Reviewed By: ajkr Differential Revision: D42004388 Pulled By: pdillinger fbshipit-source-id: 31b6e533d308a5462e528d9012d650482d974077	2 years ago
Yanqin Jin	c93ba7db5d	Revise LockWAL/UnlockWAL implementation (#11020) ... Summary: RocksDB has two public APIs: `DB::LockWAL()`/`DB::UnlockWAL()`. The current implementation acquires and releases the internal `DBImpl::log_write_mutex_`. According to the comment on `DBImpl::log_write_mutex_`: https://github.com/facebook/rocksdb/blob/7.8.fb/db/db_impl/db_impl.h#L2287:L2288 > Note: to avoid dealock, if needed to acquire both log_write_mutex_ and mutex_, the order should be first mutex_ and then log_write_mutex_. This puts limitations on how applications can use the `LockWAL()` API. After `LockWAL()` returns ok, then application should not perform any operation that acquires `mutex_`. Currently, the use case of `LockWAL()` is MyRocks implementing the MySQL storage engine handlerton `lock_hton_log` interface. The operation that MyRocks performs after `LockWAL()` is `GetSortedWalFiless()` which not only acquires mutex_, but also `log_write_mutex_`. There are two issues: 1. Applications using these two APIs may hang if one thread calls `GetSortedWalFiles()` after calling `LockWAL()` because log_write_mutex is not recursive. 2. Two threads may dead lock due to lock order inversion. To fix these issues, we can modify the implementation of LockWAL so that it does not keep `log_write_mutex_` held until UnlockWAL. To achieve the goal of locking the WAL, we can instead manually inject a write stall so that all future writes will be stopped. Pull Request resolved: https://github.com/facebook/rocksdb/pull/11020 Test Plan: make check Reviewed By: ajkr Differential Revision: D41785203 Pulled By: riversand963 fbshipit-source-id: 5ccb7a9c6eb9a2c3fa80fd2c399cc2568b8f89ce	2 years ago
Hui Xiao	98d5db5c2e	Sort L0 files by newly introduced epoch_num (#10922) ... Summary: **Context:** Sorting L0 files by `largest_seqno` has at least two inconvenience: - File ingestion and compaction involving ingested files can create files of overlapping seqno range with the existing files. `force_consistency_check=true` will catch such overlap seqno range even those harmless overlap. - For example, consider the following sequence of events ("key@n" indicates key at seqno "n") - insert k1@1 to memtable m1 - ingest file s1 with k2@2, ingest file s2 with k3@3 - insert k4@4 to m1 - compact files s1, s2 and result in new file s3 of seqno range [2, 3] - flush m1 and result in new file s4 of seqno range [1, 4]. And `force_consistency_check=true` will think s4 and s3 has file reordering corruption that might cause retuning an old value of k1 - However such caught corruption is a false positive since s1, s2 will not have overlapped keys with k1 or whatever inserted into m1 before ingest file s1 by the requirement of file ingestion (otherwise the m1 will be flushed first before any of the file ingestion completes). Therefore there in fact isn't any file reordering corruption. - Single delete can decrease a file's largest seqno and ordering by `largest_seqno` can introduce a wrong ordering hence file reordering corruption - For example, consider the following sequence of events ("key@n" indicates key at seqno "n", Credit to ajkr for this example) - an existing SST s1 contains only k1@1 - insert k1@2 to memtable m1 - ingest file s2 with k3@3, ingest file s3 with k4@4 - insert single delete k5@5 in m1 - flush m1 and result in new file s4 of seqno range [2, 5] - compact s1, s2, s3 and result in new file s5 of seqno range [1, 4] - compact s4 and result in new file s6 of seqno range [2] due to single delete - By the last step, we have file ordering by largest seqno (">" means "newer") : s5 > s6 while s6 contains a newer version of the k1's value (i.e, k1@2) than s5, which is a real reordering corruption. While this can be caught by `force_consistency_check=true`, there isn't a good way to prevent this from happening if ordering by `largest_seqno` Therefore, we are redesigning the sorting criteria of L0 files and avoid above inconvenience. Credit to ajkr , we now introduce `epoch_num` which describes the order of a file being flushed or ingested/imported (compaction output file will has the minimum `epoch_num` among input files'). This will avoid the above inconvenience in the following ways: - In the first case above, there will no longer be overlap seqno range check in `force_consistency_check=true` but `epoch_number` ordering check. This will result in file ordering s1 < s2 < s4 (pre-compaction) and s3 < s4 (post-compaction) which won't trigger false positive corruption. See test class `DBCompactionTestL0FilesMisorderCorruption*` for more. - In the second case above, this will result in file ordering s1 < s2 < s3 < s4 (pre-compacting s1, s2, s3), s5 < s4 (post-compacting s1, s2, s3), s5 < s6 (post-compacting s4), which are correct file ordering without causing any corruption. **Summary:** - Introduce `epoch_number` stored per `ColumnFamilyData` and sort CF's L0 files by their assigned `epoch_number` instead of `largest_seqno`. - `epoch_number` is increased and assigned upon `VersionEdit::AddFile()` for flush (or similarly for WriteLevel0TableForRecovery) and file ingestion (except for allow_behind_true, which will always get assigned as the `kReservedEpochNumberForFileIngestedBehind`) - Compaction output file is assigned with the minimum `epoch_number` among input files' - Refit level: reuse refitted file's epoch_number - Other paths needing `epoch_number` treatment: - Import column families: reuse file's epoch_number if exists. If not, assign one based on `NewestFirstBySeqNo` - Repair: reuse file's epoch_number if exists. If not, assign one based on `NewestFirstBySeqNo`. - Assigning new epoch_number to a file and adding this file to LSM tree should be atomic. This is guaranteed by us assigning epoch_number right upon `VersionEdit::AddFile()` where this version edit will be apply to LSM tree shape right after by holding the db mutex (e.g, flush, file ingestion, import column family) or by there is only 1 ongoing edit per CF (e.g, WriteLevel0TableForRecovery, Repair). - Assigning the minimum input epoch number to compaction output file won't misorder L0 files (even through later `Refit(target_level=0)`). It's due to for every key "k" in the input range, a legit compaction will cover a continuous epoch number range of that key. As long as we assign the key "k" the minimum input epoch number, it won't become newer or older than the versions of this key that aren't included in this compaction hence no misorder. - Persist `epoch_number` of each file in manifest and recover `epoch_number` on db recovery - Backward compatibility with old db without `epoch_number` support is guaranteed by assigning `epoch_number` to recovered files by `NewestFirstBySeqno` order. See `VersionStorageInfo::RecoverEpochNumbers()` for more - Forward compatibility with manifest is guaranteed by flexibility of `NewFileCustomTag` - Replace `force_consistent_check` on L0 with `epoch_number` and remove false positive check like case 1 with `largest_seqno` above - Due to backward compatibility issue, we might encounter files with missing epoch number at the beginning of db recovery. We will still use old L0 sorting mechanism (`NewestFirstBySeqno`) to check/sort them till we infer their epoch number. See usages of `EpochNumberRequirement`. - Remove fix https://github.com/facebook/rocksdb/pull/5958#issue-511150930 and their outdated tests to file reordering corruption because such fix can be replaced by this PR. - Misc: - update existing tests with `epoch_number` so make check will pass - update https://github.com/facebook/rocksdb/pull/5958#issue-511150930 tests to verify corruption is fixed using `epoch_number` and cover universal/fifo compaction/CompactRange/CompactFile cases - assert db_mutex is held for a few places before calling ColumnFamilyData::NewEpochNumber() Pull Request resolved: https://github.com/facebook/rocksdb/pull/10922 Test Plan: - `make check` - New unit tests under `db/db_compaction_test.cc`, `db/db_test2.cc`, `db/version_builder_test.cc`, `db/repair_test.cc` - Updated tests (i.e, `DBCompactionTestL0FilesMisorderCorruption*`) under https://github.com/facebook/rocksdb/pull/5958#issue-511150930 - [Ongoing] Compatibility test: manually run 36a5686ec0 (with file ingestion off for running the `.orig` binary to prevent this bug affecting upgrade/downgrade formality checking) for 1 hour on `simple black/white box`, `cf_consistency/txn/enable_ts with whitebox + test_best_efforts_recovery with blackbox` - [Ongoing] normal db stress test - [Ongoing] db stress test with aggressive value https://github.com/facebook/rocksdb/pull/10761 Reviewed By: ajkr Differential Revision: D41063187 Pulled By: hx235 fbshipit-source-id: 826cb23455de7beaabe2d16c57682a82733a32a9	2 years ago
Peter Dillinger	433d7e4594	Improve error messages for SST footer and size errors (#11009) ... Summary: Previously, you could get a format_version error if SST file size was too small in manifest, or a weird "too short" error if too big in manifest. Now we ensure: * Magic number error is reported first if we attempt to open an SST file and the footer is completely bad. * Footer errors are reported with affected file. * If manifest file size doesn't match actual, then the error includes expected and actual sizes (if an error is reported; in some cases we allow the file to be too big) Pull Request resolved: https://github.com/facebook/rocksdb/pull/11009 Test Plan: unit tests added, some manual Previously, the code for "file too short" in footer processing was only covered by some tests attempting to verify SST checksums on non-SST files (fixed). Reviewed By: siying Differential Revision: D41656272 Pulled By: pdillinger fbshipit-source-id: 3da32702eb5aaedbea0e5e74742ad57edd7ad3df	2 years ago
Changyu Bi	534fb06dd3	Prevent iterating over range tombstones beyond `iterate_upper_bound` (#10966) ... Summary: Currently, `iterate_upper_bound` is not checked for range tombstone keys in MergingIterator. This may impact performance when there is a large number of range tombstones right after `iterate_upper_bound`. This PR fixes this issue by checking `iterate_upper_bound` in MergingIterator for range tombstone keys. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10966 Test Plan: - added unit test - stress test: `python3 tools/db_crashtest.py whitebox --simple --verify_iterator_with_expected_state_one_in=5 --delrangepercent=5 --prefixpercent=18 --writepercent=48 --readpercen=15 --duration=36000 --range_deletion_width=100` - ran different stress tests over sandcastle - Falcon team ran some test traffic and saw reduced CPU usage on processing range tombstones. Reviewed By: ajkr Differential Revision: D41414172 Pulled By: cbi42 fbshipit-source-id: 9b2c29eb3abb99327c6a649bdc412e70d863f981	2 years ago
Yanqin Jin	3d0d6b8140	Make best-efforts recovery verify SST unique ID before Version construction (#10962) ... Summary: The check for SST unique IDs added to best-efforts recovery (`Options::best_efforts_recovery` is true). With best_efforts_recovery being true, RocksDB will recover to the latest point in MANIFEST such that all valid SST files included up to this point pass unique ID checks as well. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10962 Test Plan: make check Reviewed By: pdillinger Differential Revision: D41378241 Pulled By: riversand963 fbshipit-source-id: a036064e2c17dec13d080a24ef2a9f85d607b16c	2 years ago
anand76	f4cfcfe824	Post 7.9.0 release branch cut updates (#10974) ... Summary: Update HISTORY.md, version.h, and check_format_compatible.sh Pull Request resolved: https://github.com/facebook/rocksdb/pull/10974 Reviewed By: akankshamahajan15 Differential Revision: D41455289 Pulled By: anand1976 fbshipit-source-id: 99888ebcb9109e5ced80584a66b20123f8783c0b	2 years ago
Peter Dillinger	e079d562af	Add a SecondaryCache::InsertSaved() API, use in CacheDumper impl (#10945) ... Summary: Can simplify some ugly code in cache_dump_load_impl.cc by having an API in SecondaryCache that can directly consume persisted data. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10945 Test Plan: existing tests for CacheDumper, added basic unit test Reviewed By: anand1976 Differential Revision: D41231497 Pulled By: pdillinger fbshipit-source-id: b8ec993ef7d3e7efd68aae8602fd3f858da58068	2 years ago
Peter Dillinger	3182beeffc	Observe and warn about misconfigured HyperClockCache (#10965) ... Summary: Background. One of the core risks of chosing HyperClockCache is ending up with degraded performance if estimated_entry_charge is very significantly wrong. Too low leads to under-utilized hash table, which wastes a bit of (tracked) memory and likely increases access times due to larger working set size (more TLB misses). Too high leads to fully populated hash table (at some limit with reasonable lookup performance) and not being able to cache as many objects as the memory limit would allow. In either case, performance degradation is graceful/continuous but can be quite significant. For example, cutting block size in half without updating estimated_entry_charge could lead to a large portion of configured block cache memory (up to roughly 1/3) going unused. Fix. This change adds a mechanism through which the DB periodically probes the block cache(s) for "problems" to report, and adds diagnostics to the HyperClockCache for bad estimated_entry_charge. The periodic probing is currently done with DumpStats / stats_dump_period_sec, and diagnostics reported to info_log (normally LOG file). Pull Request resolved: https://github.com/facebook/rocksdb/pull/10965 Test Plan: unit test included. Doesn't cover all the implemented subtleties of reporting, but ensures basics of when to report or not. Also manual testing with db_bench. Create db with ``` ./db_bench --benchmarks=fillrandom,flush --num=3000000 --disable_wal=1 ``` Use and check LOG file for HyperClockCache for various block sizes (used as estimated_entry_charge) ``` ./db_bench --use_existing_db --benchmarks=readrandom --num=3000000 --duration=20 --stats_dump_period_sec=8 --cache_type=hyper_clock_cache -block_size=XXXX ``` Seeing warnings / errors or not as expected. Reviewed By: anand1976 Differential Revision: D41406932 Pulled By: pdillinger fbshipit-source-id: 4ca56162b73017e4b9cec2cad74466f49c27a0a7	2 years ago
Peter Dillinger	8c0f5b1fcf	Mark HyperClockCache as production-ready (#10963) ... Summary: After a couple minor bug fixes and successful productions roll-outs in a few places, I think we can mark this as production-ready. It has a clear value proposition for many workloads, even if we don't have clear advice for every workload yet. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10963 Test Plan: existing tests, comment changes only Reviewed By: siying Differential Revision: D41384083 Pulled By: pdillinger fbshipit-source-id: 56359f01a57bb28de8697666b342382fac72ce6d	2 years ago
Peter Dillinger	b55e70357c	Re-arrange cache.h to prepare for refactoring (#10942) ... Summary: No material changes to code or comments, just re-arranging things to prepare for a big refactoring, making it easier to what changed. Some specifics: * This groups things together in Cache in anticipation of secondary cache features being marked production-ready (vs. experimental). * CacheEntryRole will be needed in definition of class Cache, so that has been moved above it. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10942 Test Plan: existing tests Reviewed By: anand1976 Differential Revision: D41205509 Pulled By: pdillinger fbshipit-source-id: 3f2559ab1651c758918dc97056951fa2b5eb0348	2 years ago
anand76	ecba6a320e	Add some async read stats (#10947) ... Summary: Add stats for time spent in the ReadAsync call, and async read errors. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10947 Test Plan: Run db_bench and look at stats Reviewed By: akankshamahajan15 Differential Revision: D41236637 Pulled By: anand1976 fbshipit-source-id: 70539b69a28491d57acead449436a761f7108acf	2 years ago
Peter Dillinger	f321e8fc98	Don't attempt to use SecondaryCache on block_cache_compressed (#10944) ... Summary: Compressed block cache depends on reading the block compression marker beyond the payload block size. Only the payload bytes were being saved and loaded from SecondaryCache -> boom! This removes some unnecessary code attempting to combine these two competing features. Note that BlockContents was previously used for block-based filter in block cache, but that support has been removed. Also marking block_cache_compressed as deprecated in this commit as we expect it to be replaced with SecondaryCache. This problem was discovered during refactoring but didn't want to combine bug fix with that refactoring. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10944 Test Plan: test added that fails on base revision (at least with ASAN) Reviewed By: akankshamahajan15 Differential Revision: D41205578 Pulled By: pdillinger fbshipit-source-id: 1b29d36c7a6552355ac6511fcdc67038ef4af29f	2 years ago
Yanqin Jin	7d26e4c5a3	Basic Support for Merge with user-defined timestamp (#10819) ... Summary: This PR implements the originally disabled `Merge()` APIs when user-defined timestamp is enabled. Simplest usage: ```cpp // assume string append merge op is used with '.' as delimiter. // ts1 < ts2 db->Put(WriteOptions(), "key", ts1, "v0"); db->Merge(WriteOptions(), "key", ts2, "1"); ReadOptions ro; ro.timestamp = &ts2; db->Get(ro, "key", &value); ASSERT_EQ("v0.1", value); ``` Some code comments are added for clarity. Note: support for timestamp in `DB::GetMergeOperands()` will be done in a follow-up PR. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10819 Test Plan: make check Reviewed By: ltamasi Differential Revision: D40603195 Pulled By: riversand963 fbshipit-source-id: f96d6f183258f3392d80377025529f7660503013	2 years ago
akankshamahajan	0ed1a800ed	Fix override error in system_clock.h (#10858) ... Summary: Fix error ``` rocksdb/system_clock.h:30:11: error: '~SystemClock' overrides a destructor but is not marked 'override' [-Werror,-Wsuggest-destructor-override] virtual ~SystemClock() {} ``` Pull Request resolved: https://github.com/facebook/rocksdb/pull/10858 Test Plan: Ran internally Reviewed By: siying Differential Revision: D40652374 Pulled By: akankshamahajan15 fbshipit-source-id: 5dda8ca03ea57d709442c87e23e5fe097d7db672	2 years ago
akankshamahajan	966cd42c7d	Update header file to include right copyright (#10854) ... Summary: Pull Request resolved: https://github.com/facebook/rocksdb/pull/10854 Reviewed By: siying Differential Revision: D40651483 Pulled By: akankshamahajan15 fbshipit-source-id: 95ce53297e9699a34cc80439bc7553f6cc3ac957	2 years ago
sdong	b0d9776b70	clang format files under include/ (#10850) ... Summary: Run clang-format against files under include/ Pull Request resolved: https://github.com/facebook/rocksdb/pull/10850 Test Plan: Watch existing CI to pass. Reviewed By: ajkr Differential Revision: D40646158 fbshipit-source-id: 8ce04b107c837630f4000a478d0c871577090263	2 years ago
Changyu Bi	deb6a24be2	Remove range tombstone test code from sst_file_reader (#10847) ... Summary: `#include "db/range_tombstone_fragmenter.h"` seems to break some internal test for 7.8 release. I'm removing it from sst_file_reader.h for now to unblock release. This should be fine as it is only used in a unit test for DeleteRange with timestamp. In addition, it does not seem to be useful to support delete range for sst file writer, since the range tombstone won't cover any key (its sequence number is 0). So maybe we can remove it in the future. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10847 Test Plan: CI. Reviewed By: akankshamahajan15 Differential Revision: D40620865 Pulled By: cbi42 fbshipit-source-id: be44b2f31e062bff87ed1b8d94482c3f7eaa370c	2 years ago
akankshamahajan	daceb85c51	Update version.h, HISTORY.md and add branches to compatibility check (#10846) ... Summary: Pull Request resolved: https://github.com/facebook/rocksdb/pull/10846 Reviewed By: ajkr Differential Revision: D40617997 Pulled By: akankshamahajan15 fbshipit-source-id: 4b2d6e85dbca7e73b930c4165869b693d3e4e137	2 years ago
Peter Dillinger	27c9705ac4	Use kXXH3 as default checksum (CPU efficiency) (#10778) ... Summary: Since this has been supported for about a year, I think it's time to make it the default. This should improve CPU efficiency slightly on most hardware. A current DB performance comparison using buck+clang build: ``` TEST_TMPDIR=/dev/shm ./db_bench -checksum_type={1,4} -benchmarks=fillseq[-X1000] -num=3000000 -disable_wal ``` kXXH3 (+0.2% DB write throughput): `fillseq [AVG 1000 runs] : 822149 (± 1004) ops/sec; 91.0 (± 0.1) MB/sec` kCRC32c: `fillseq [AVG 1000 runs] : 820484 (± 1203) ops/sec; 90.8 (± 0.1) MB/sec` Micro benchmark comparison: ``` ./db_bench --benchmarks=xxh3[-X20],crc32c[-X20] ``` Machine 1, buck+clang build: `xxh3 [AVG 20 runs] : 3358616 (± 19091) ops/sec; 13119.6 (± 74.6) MB/sec` `crc32c [AVG 20 runs] : 2578725 (± 7742) ops/sec; 10073.1 (± 30.2) MB/sec` Machine 2, make+gcc build, DEBUG_LEVEL=0 PORTABLE=0: `xxh3 [AVG 20 runs] : 6182084 (± 137223) ops/sec; 24148.8 (± 536.0) MB/sec` `crc32c [AVG 20 runs] : 5032465 (± 42454) ops/sec; 19658.1 (± 165.8) MB/sec` Pull Request resolved: https://github.com/facebook/rocksdb/pull/10778 Test Plan: make check, unit tests updated Reviewed By: ajkr Differential Revision: D40112510 Pulled By: pdillinger fbshipit-source-id: e59a8d50a60346137732f8668ba7cfac93be2b37	2 years ago
akankshamahajan	0e7b27bfcf	Refactor block cache tracing APIs (#10811) ... Summary: Refactor the classes, APIs and data structures for block cache tracing to allow a user provided trace writer to be used. Currently, only a TraceWriter is supported, with a default built-in implementation of FileTraceWriter. The TraceWriter, however, takes a flat trace record and is thus only suitable for file tracing. This PR introduces an abstract BlockCacheTraceWriter class that takes a structured BlockCacheTraceRecord. The BlockCacheTraceWriter implementation can then format and log the record in whatever way it sees fit. The default BlockCacheTraceWriterImpl does file tracing using a user provided TraceWriter. `DB::StartBlockTrace` will internally redirect to changed `BlockCacheTrace::StartBlockCacheTrace`. New API `DB::StartBlockTrace` is also added that directly takes `BlockCacheTraceWriter` pointer. This same philosophy can be applied to KV and IO tracing as well. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10811 Test Plan: existing unit tests Old API DB::StartBlockTrace checked with db_bench tool create database ``` ./db_bench --benchmarks="fillseq" \ --key_size=20 --prefix_size=20 --keys_per_prefix=0 --value_size=100 \ --cache_index_and_filter_blocks --cache_size=1048576 \ --disable_auto_compactions=1 --disable_wal=1 --compression_type=none \ --min_level_to_compress=-1 --compression_ratio=1 --num=10000000 ``` To trace block cache accesses when running readrandom benchmark: ``` ./db_bench --benchmarks="readrandom" --use_existing_db --duration=60 \ --key_size=20 --prefix_size=20 --keys_per_prefix=0 --value_size=100 \ --cache_index_and_filter_blocks --cache_size=1048576 \ --disable_auto_compactions=1 --disable_wal=1 --compression_type=none \ --min_level_to_compress=-1 --compression_ratio=1 --num=10000000 \ --threads=16 \ -block_cache_trace_file="/tmp/binary_trace_test_example" \ -block_cache_trace_max_trace_file_size_in_bytes=1073741824 \ -block_cache_trace_sampling_frequency=1 ``` Reviewed By: anand1976 Differential Revision: D40435289 Pulled By: akankshamahajan15 fbshipit-source-id: fa2755f4788185e19f4605e731641cfd21ab3282	2 years ago
Changyu Bi	333abe9c55	Ignore max_compaction_bytes for compaction input that are within output key-range (#10835) ... Summary: When picking compaction input files, we sometimes stop picking a file that is fully included in the output key-range due to hitting max_compaction_bytes. Including these input files can potentially reduce WA at the expense of larger compactions. Larger compaction should be fine as files from input level are usually 10X smaller than files from output level. This PR adds a mutable CF option `ignore_max_compaction_bytes_for_input` that is enabled by default. We can remove this option once we are sure it is safe. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10835 Test Plan: - CI, a unit test on max_compaction_bytes fails before turning this flag off. - Benchmark does not show much difference in WA: `./db_bench --benchmarks=fillrandom,waitforcompaction,stats,levelstats -max_background_jobs=12 -num=2000000000 -target_file_size_base=33554432 --write_buffer_size=33554432` ``` main: ** Compaction Stats [default] ** Level Files Size Score Read(GB) Rn(GB) Rnp1(GB) Write(GB) Wnew(GB) Moved(GB) W-Amp Rd(MB/s) Wr(MB/s) Comp(sec) CompMergeCPU(sec) Comp(cnt) Avg(sec) KeyIn KeyDrop Rblob(GB) Wblob(GB) ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ L0 3/0 91.59 MB 0.8 70.9 0.0 70.9 200.8 129.9 0.0 1.5 25.2 71.2 2886.55 2463.45 9725 0.297 1093M 254K 0.0 0.0 L1 9/0 248.03 MB 1.0 392.0 129.8 262.2 391.7 129.5 0.0 3.0 69.0 68.9 5821.71 5536.90 804 7.241 6029M 5814K 0.0 0.0 L2 87/0 2.50 GB 1.0 537.0 128.5 408.5 533.8 125.2 0.7 4.2 69.5 69.1 7912.24 7323.70 4417 1.791 8299M 36M 0.0 0.0 L3 836/0 24.99 GB 1.0 616.9 118.3 498.7 594.5 95.8 5.2 5.0 66.9 64.5 9442.38 8490.28 4204 2.246 9749M 306M 0.0 0.0 L4 2355/0 62.95 GB 0.3 67.3 37.1 30.2 54.2 24.0 38.9 1.5 72.2 58.2 954.37 821.18 917 1.041 1076M 173M 0.0 0.0 Sum 3290/0 90.77 GB 0.0 1684.2 413.7 1270.5 1775.0 504.5 44.9 13.7 63.8 67.3 27017.25 24635.52 20067 1.346 26G 522M 0.0 0.0 Cumulative compaction: 1774.96 GB write, 154.29 MB/s write, 1684.19 GB read, 146.40 MB/s read, 27017.3 seconds This PR: ** Compaction Stats [default] ** Level Files Size Score Read(GB) Rn(GB) Rnp1(GB) Write(GB) Wnew(GB) Moved(GB) W-Amp Rd(MB/s) Wr(MB/s) Comp(sec) CompMergeCPU(sec) Comp(cnt) Avg(sec) KeyIn KeyDrop Rblob(GB) Wblob(GB) ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ L0 3/0 45.71 MB 0.8 72.9 0.0 72.9 202.8 129.9 0.0 1.6 25.4 70.7 2938.16 2510.36 9741 0.302 1124M 265K 0.0 0.0 L1 8/0 234.54 MB 0.9 384.5 129.8 254.7 384.2 129.6 0.0 3.0 69.0 68.9 5708.08 5424.43 791 7.216 5913M 5753K 0.0 0.0 L2 84/0 2.47 GB 1.0 543.1 128.6 414.5 539.9 125.4 0.7 4.2 69.6 69.2 7989.31 7403.13 4418 1.808 8393M 36M 0.0 0.0 L3 839/0 24.96 GB 1.0 615.6 118.4 497.2 593.2 96.0 5.1 5.0 66.6 64.1 9471.23 8489.31 4193 2.259 9726M 306M 0.0 0.0 L4 2360/0 63.04 GB 0.3 67.6 37.3 30.3 54.4 24.1 38.9 1.5 71.5 57.6 967.30 827.99 907 1.066 1080M 173M 0.0 0.0 Sum 3294/0 90.75 GB 0.0 1683.8 414.2 1269.6 1774.5 504.9 44.8 13.7 63.7 67.1 27074.08 24655.22 20050 1.350 26G 522M 0.0 0.0 Cumulative compaction: 1774.52 GB write, 157.09 MB/s write, 1683.77 GB read, 149.06 MB/s read, 27074.1 seconds ``` Reviewed By: ajkr Differential Revision: D40518319 Pulled By: cbi42 fbshipit-source-id: f4ea614bc0ebefe007ffaf05bb9aec9a8ca25b60	2 years ago
Andrew Kryczka	33ceea9b76	Add DB property for fast block cache stats collection (#10832) ... Summary: This new property allows users to trigger the background block cache stats collection mode through the `GetProperty()` and `GetMapProperty()` APIs. The background mode has much lower overhead at the expense of returning stale values in more cases. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10832 Test Plan: updated unit test Reviewed By: pdillinger Differential Revision: D40497883 Pulled By: ajkr fbshipit-source-id: bdcc93402f426463abb2153756aad9e295447343	2 years ago
Jay Zhuang	c401f285c3	Add option `preserve_internal_time_seconds` to preserve the time info (#10747) ... Summary: Add option `preserve_internal_time_seconds` to preserve the internal time information. It's mostly for the migration of the existing data to tiered storage ( `preclude_last_level_data_seconds`). When the tiering feature is just enabled, the existing data won't have the time information to decide if it's hot or cold. Enabling this feature will start collect and preserve the time information for the new data. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10747 Reviewed By: siying Differential Revision: D39910141 Pulled By: siying fbshipit-source-id: 25c21638e37b1a7c44006f636b7d714fe7242138	2 years ago
Levi Tamasi	d6d8c007ff	Verify columns in NonBatchedOpsStressTest::VerifyDb (#10783) ... Summary: As the first step of covering the wide-column functionality of iterators in our stress tests, the patch adds verification logic to `NonBatchedOpsStressTest::VerifyDb` that checks whether the iterator's value and columns are in sync. Note: I plan to update the other types of stress tests and add similar verification for prefix scans etc. in separate PRs. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10783 Test Plan: Ran some simple blackbox crash tests. Reviewed By: riversand963 Differential Revision: D40152370 Pulled By: riversand963 fbshipit-source-id: 8f9d17d7af5da58ccf1bd2057cab53cc9645ac35	2 years ago
Yanqin Jin	4d82b94896	Sanitize min_write_buffer_number_to_merge to 1 with atomic_flush (#10773) ... Summary: With current implementation, within the same RocksDB instance, all column families with non-empty memtables will be scheduled for flush if RocksDB determines that any column family needs to be flushed, e.g. memtable full, write buffer manager, etc., if atomic flush is enabled. Not doing so can lead to data loss and inconsistency when WAL is disabled, which is a common setting when atomic flush is enabled. Therefore, setting a per-column-family knob, min_write_buffer_number_to_merge to a value greater than 1 is not compatible with atomic flush, and should be sanitized during column family creation and db open. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10773 Test Plan: Reproduce: D39993203 has detailed steps. Run the test with and without the fix. Reviewed By: cbi42 Differential Revision: D40077955 Pulled By: cbi42 fbshipit-source-id: 451a9179eb531ac42eaccf40b451b9dec4085240	2 years ago
Jay Zhuang	f007ad8b4f	RoundRobin TTL compaction (#10725) ... Summary: For RoundRobin compaction, the data should be mostly sorted per level and within level. Use normal compaction picker for RR until all expired data is compacted. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10725 Reviewed By: ajkr Differential Revision: D39771069 Pulled By: jay-zhuang fbshipit-source-id: 7ccf88d7c093fad5673bda73a7b08cc4757780cd	2 years ago
Peter Dillinger	5f4391dda2	Some clean-up of secondary cache (#10730) ... Summary: This is intended as a step toward possibly separating secondary cache integration from the Cache implementation as much as possible, to (hopefully) minimize code duplication in adding secondary cache support to HyperClockCache. * Major clarifications to API docs of secondary cache compatible parts of Cache. For example, previously the docs seemed to suggest that Wait() was not needed if IsReady()==true. And it wasn't clear what operations were actually supported on pending handles. * Add some assertions related to these requirements, such as that we don't Release() before Wait() (which would leak a secondary cache handle). * Fix a leaky abstraction with dummy handles, which are supposed to be internal to the Cache. Previously, these just used value=nullptr to indicate dummy handle, which meant that they could be confused with legitimate value=nullptr cases like cache reservations. Also fixed blob_source_test which was relying on this leaky abstraction. * Drop "incomplete" terminology, which was another name for "pending". * Split handle flags into "mutable" ones requiring mutex and "immutable" ones which do not. Because of single-threaded access to pending handles, the "Is Pending" flag can be in the "immutable" set. This allows removal of a TSAN work-around and removing a mutex acquire-release in IsReady(). * Remove some unnecessary handling of charges on handles of failed lookups. Keeping total_charge=0 means no special handling needed. (Removed one unnecessary mutex acquire/release.) * Simplify handling of dummy handle in Lookup(). There is no need to explicitly Ref & Release w/Erase if we generally overwrite the dummy anyway. (Removed one mutex acquire/release, a call to Release().) Intended follow-up: * Clarify APIs in secondary_cache.h * Doesn't SecondaryCacheResultHandle transfer ownership of the Value() on success (implementations should not release the value in destructor)? * Does Wait() need to be called if IsReady() == true? (This would be different from Cache.) * Do Value() and Size() have undefined behavior if IsReady() == false? * Why have a custom API for what is essentially a std::future<std::pair<void*, size_t>>? * Improve unit testing of standalone handle case * Apparent null `e` bug in `free_standalone_handle` case * Clean up secondary cache testing in lru_cache_test * Why does TestSecondaryCacheResultHandle hold on to a Cache::Handle? * Why does TestSecondaryCacheResultHandle::Wait() do nothing? Shouldn't it establish the post-condition IsReady() == true? * (Assuming that is sorted out...) Shouldn't TestSecondaryCache::WaitAll simply wait on each handle in order (no casting required)? How about making that the default implementation? * Why does TestSecondaryCacheResultHandle::Size() check Value() first? If the API is intended to be returning 0 before IsReady(), then that is weird but should at least be documented. Otherwise, if it's intended to be undefined behavior, we should assert IsReady(). * Consider replacing "standalone" and "dummy" entries with a single kind of "weak" entry that deletes its value when it reaches zero refs. Suppose you are using compressed secondary cache and have two iterators at similar places. It will probably common for one iterator to have standalone results pinned (out of cache) when the second iterator needs those same blocks and has to re-load them from secondary cache and duplicate the memory. Combining the dummy and the standalone should fix this. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10730 Test Plan: existing tests (minor update), and crash test with sanitizers and secondary cache Performance test for any regressions in LRUCache (primary only): Create DB with ``` TEST_TMPDIR=/dev/shm ./db_bench -benchmarks=fillrandom -num=30000000 -disable_wal=1 -bloom_bits=16 ``` Test before & after (run at same time) with ``` TEST_TMPDIR=/dev/shm ./db_bench -benchmarks=readrandom[-X100] -readonly -num=30000000 -bloom_bits=16 -cache_index_and_filter_blocks=1 -cache_size=233000000 -duration 30 -threads=16 ``` Before: readrandom [AVG 100 runs] : 22234 (± 63) ops/sec; 1.6 (± 0.0) MB/sec After: readrandom [AVG 100 runs] : 22197 (± 64) ops/sec; 1.6 (± 0.0) MB/sec That's within 0.2%, which is not significant by the confidence intervals. Reviewed By: anand1976 Differential Revision: D39826010 Pulled By: anand1976 fbshipit-source-id: 3202b4a91f673231c97648ae070e502ae16b0f44	2 years ago
akankshamahajan	ae0f9c3339	Add new property in IOOptions to skip recursing through directories and list only files during GetChildren. (#10668) ... Summary: Add new property "do_not_recurse" in IOOptions for underlying file system to skip iteration of directories during DB::Open if there are no sub directories and list only files. By default this property is set to false. This property is set true currently in the code where RocksDB is sure only files are needed during DB::Open. Provided support in PosixFileSystem to use "do_not_recurse". TestPlan: - Existing tests Pull Request resolved: https://github.com/facebook/rocksdb/pull/10668 Reviewed By: anand1976 Differential Revision: D39471683 Pulled By: akankshamahajan15 fbshipit-source-id: 90e32f0b86d5346d53bc2714d3a0e7002590527f	2 years ago
Changyu Bi	9f2363f4c4	User-defined timestamp support for `DeleteRange()` (#10661) ... Summary: Add user-defined timestamp support for range deletion. The new API is `DeleteRange(opt, cf, begin_key, end_key, ts)`. Most of the change is to update the comparator to compare without timestamp. Other than that, major changes are - internal range tombstone data structures (`FragmentedRangeTombstoneList`, `RangeTombstone`, etc.) to store timestamps. - Garbage collection of range tombstones and range tombstone covered keys during compaction. - Get()/MultiGet() to return the timestamp of a range tombstone when needed. - Get/Iterator with range tombstones bounded by readoptions.timestamp. - timestamp crash test now issues DeleteRange by default. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10661 Test Plan: - Added unit test: `make check` - Stress test: `python3 tools/db_crashtest.py --enable_ts whitebox --readpercent=57 --prefixpercent=4 --writepercent=25 -delpercent=5 --iterpercent=5 --delrangepercent=4` - Ran `db_bench` to measure regression when timestamp is not enabled. The tests are for write (with some range deletion) and iterate with DB fitting in memory: `./db_bench--benchmarks=fillrandom,seekrandom --writes_per_range_tombstone=200 --max_write_buffer_number=100 --min_write_buffer_number_to_merge=100 --writes=500000 --reads=500000 --seek_nexts=10 --disable_auto_compactions -disable_wal=true --max_num_range_tombstones=1000`. Did not see consistent regression in no timestamp case. | micros/op | fillrandom | seekrandom | | --- | --- | --- | |main| 2.58 |10.96| |PR 10661| 2.68 |10.63| Reviewed By: riversand963 Differential Revision: D39441192 Pulled By: cbi42 fbshipit-source-id: f05aca3c41605caf110daf0ff405919f300ddec2	2 years ago
Changyu Bi	fd71a82f4f	Use actual file size when checking max_compaction_size (#10728) ... Summary: currently, there are places in compaction_picker where we add up `compensated_file_size` of files being compacted and limit the sum to be under `max_compaction_bytes`. `compensated_file_size` contains booster for point tombstones and should be used only for determining file's compaction priority. This PR replaces `compensated_file_size` with actual file size in such places. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10728 Test Plan: CI Reviewed By: ajkr Differential Revision: D39789427 Pulled By: cbi42 fbshipit-source-id: 1f89fb6c0159c53bf01d8dc783f465959f442c81	2 years ago
Jay Zhuang	f3cc66632b	Align compaction output file boundaries to the next level ones (#10655) ... Summary: Try to align the compaction output file boundaries to the next level ones (grandparent level), to reduce the level compaction write-amplification. In level compaction, there are "wasted" data at the beginning and end of the output level files. Align the file boundary can avoid such "wasted" compaction. With this PR, it tries to align the non-bottommost level file boundaries to its next level ones. It may cut file when the file size is large enough (at least 50% of target_file_size) and not too large (2x target_file_size). db_bench shows about 12.56% compaction reduction: ``` TEST_TMPDIR=/data/dbbench2 ./db_bench --benchmarks=fillrandom,readrandom -max_background_jobs=12 -num=400000000 -target_file_size_base=33554432 # baseline: Flush(GB): cumulative 25.882, interval 7.216 Cumulative compaction: 285.90 GB write, 162.36 MB/s write, 269.68 GB read, 153.15 MB/s read, 2926.7 seconds # with this change: Flush(GB): cumulative 25.882, interval 7.753 Cumulative compaction: 249.97 GB write, 141.96 MB/s write, 233.74 GB read, 132.74 MB/s read, 2534.9 seconds ``` The compaction simulator shows a similar result (14% with 100G random data). As a side effect, with this PR, the SST file size can exceed the target_file_size, but is capped at 2x target_file_size. And there will be smaller files. Here are file size statistics when loading 100GB with the target file size 32MB: ``` baseline this_PR count 1.656000e+03 1.705000e+03 mean 3.116062e+07 3.028076e+07 std 7.145242e+06 8.046139e+06 ``` The feature is enabled by default, to revert to the old behavior disable it with `AdvancedColumnFamilyOptions.level_compaction_dynamic_file_size = false` Also includes https://github.com/facebook/rocksdb/issues/1963 to cut file before skippable grandparent file. Which is for use case like user adding 2 or more non-overlapping data range at the same time, it can reduce the overlapping of 2 datasets in the lower levels. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10655 Reviewed By: cbi42 Differential Revision: D39552321 Pulled By: jay-zhuang fbshipit-source-id: 640d15f159ab0cd973f2426cfc3af266fc8bdde2	2 years ago
gitbw95	47b57a3731	add SetCapacity and GetCapacity for secondary cache (#10712) ... Summary: To support tuning secondary cache dynamically, add `SetCapacity()` and `GetCapacity()` for CompressedSecondaryCache. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10712 Test Plan: Unit Tests Reviewed By: anand1976 Differential Revision: D39685212 Pulled By: gitbw95 fbshipit-source-id: 19573c67237011927320207732b5de083cb87240	2 years ago
Levi Tamasi	6d2a9832d9	Clarify API comments for blob_cache/prepopulate_blob_cache (#10723) ... Summary: Pull Request resolved: https://github.com/facebook/rocksdb/pull/10723 Reviewed By: riversand963 Differential Revision: D39749277 Pulled By: ltamasi fbshipit-source-id: 4bda94b4620a0db1fcd4309c7ad03fc23e8718cb	2 years ago
walter	1b351fd9fe	Add C API to set avoid_unnecessary_blocking_io option (#10693) ... Summary: Pull Request resolved: https://github.com/facebook/rocksdb/pull/10693 Reviewed By: riversand963 Differential Revision: D39668399 Pulled By: ajkr fbshipit-source-id: 66d46e5104c49d235a14e8df8eec3af285ab9752	2 years ago
Peter Dillinger	ef443cead4	Refactor to avoid confusing "raw block" (#10408) ... Summary: We have a lot of confusing code because of mixed, sometimes completely opposite uses of of the term "raw block" or "raw contents", sometimes within the same source file. For example, in `BlockBasedTableBuilder`, `raw_block_contents` and `raw_size` generally referred to uncompressed block contents and size, while `WriteRawBlock` referred to writing a block that is already compressed if it is going to be. Meanwhile, in `BlockBasedTable`, `raw_block_contents` either referred to a (maybe compressed) block with trailer, or a maybe compressed block maybe without trailer. (Note: left as follow-up work to use C++ typing to better sort out the various kinds of BlockContents.) This change primarily tries to apply some consistent terminology around the kinds of block representations, avoiding the unclear "raw". (Any meaning of "raw" assumes some bias toward the storage layer or toward the logical data layer.) Preferred terminology: * **Serialized block** - bytes that go into storage. For block-based table (usually the case) this includes the block trailer. WART: block `size` may or may not include the trailer; need to be clear about whether it does or not. * **Maybe compressed block** - like a serialized block, but without the trailer (or no promise of including a trailer). Must be accompanied by a CompressionType. * **Uncompressed block** - "payload" bytes that are either stored with no compression, used as input to compression function, or result of decompression function. * **Parsed block** - an in-memory form of a block in block cache, as it is used by the table reader. Different C++ types are used depending on the block type (see block_like_traits.h). Other refactorings: * Misc corrections/improvements of internal API comments * Remove a few misleading / unhelpful / redundant comments. * Use move semantics in some places to simplify contracts * Use better parameter names to indicate which parameters are used for outputs * Remove some extraneous `extern` * Various clean-ups to `CacheDumperImpl` (mostly unnecessary code) Pull Request resolved: https://github.com/facebook/rocksdb/pull/10408 Test Plan: existing tests Reviewed By: akankshamahajan15 Differential Revision: D38172617 Pulled By: pdillinger fbshipit-source-id: ccb99299f324ac5ca46996d34c5089621a4f260c	2 years ago
Levi Tamasi	12f5a1e35c	Clarify comments for cache priorities and pool options (#10718) ... Summary: Pull Request resolved: https://github.com/facebook/rocksdb/pull/10718 Reviewed By: riversand963 Differential Revision: D39707115 Pulled By: ltamasi fbshipit-source-id: 59aec8c732482f063d0abaad4d9200ba57ebf437	2 years ago
Yanqin Jin	2b6f3510c2	Update version number and HISTORY in main branch (#10694) ... Summary: This PR bumps up version number from 7.7 to 7.8 in main branch, indicating that next release will be 7.8. We are going to release 7.7 soon. Since 7.7.fb branch has been created, we can land this to main. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10694 Reviewed By: gitbw95 Differential Revision: D39581577 Pulled By: riversand963 fbshipit-source-id: 84f3fecf25fd9ac96e46b4cd6d50ddb6edc89427	2 years ago
gitbw95	2cc5b39560	Add enable_split_merge option for CompressedSecondaryCache (#10690) ... Summary: `enable_custom_split_merge` is added for enabling the custom split and merge feature, which split the compressed value into chunks so that they may better fit jemalloc bins. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10690 Test Plan: Unit Tests Stress Tests Reviewed By: anand1976 Differential Revision: D39567604 Pulled By: gitbw95 fbshipit-source-id: f6d1d46200f365220055f793514601dcb0edc4b7	2 years ago
Peter Dillinger	0f91c72adc	Call experimental new clock cache HyperClockCache (#10684) ... Summary: This change establishes a distinctive name for the experimental new lock-free clock cache (originally developed by guidotag and revamped in PR https://github.com/facebook/rocksdb/issues/10626). A few reasons: * We want to make it clear that this is a fundamentally different implementation vs. the old clock cache, to avoid people saying "I already tried clock cache." * We want to highlight the key feature: it's fast (especially under parallel load) * Because it requires an estimated charge per entry, it is not drop-in API compatible with old clock cache. This estimate might always be required for highest performance, and giving it a distinct name should reduce confusion about the distinct API requirements. * We might develop a variant requiring the same estimate parameter but with LRU eviction. In that case, using the name HyperLRUCache should make things more clear. (FastLRUCache is just a prototype that might soon be removed.) Some API detail: * To reduce copy-pasting parameter lists, etc. as in LRUCache construction, I have a `MakeSharedCache()` function on `HyperClockCacheOptions` instead of `NewHyperClockCache()`. * Changes -cache_type=clock_cache to -cache_type=hyper_clock_cache for applicable tools. I think this is more consistent / sustainable for reasons already stated. For performance tests see https://github.com/facebook/rocksdb/pull/10626 Pull Request resolved: https://github.com/facebook/rocksdb/pull/10684 Test Plan: no interesting functional changes; tests updated Reviewed By: anand1976 Differential Revision: D39547800 Pulled By: pdillinger fbshipit-source-id: 5c0fe1b5cf3cb680ab369b928c8569682b9795bf	2 years ago
Peter Dillinger	5724348689	Revamp, optimize new experimental clock cache (#10626) ... Summary: * Consolidates most metadata into a single word per slot so that more can be accomplished with a single atomic update. In the common case, Lookup was previously about 4 atomic updates, now just 1 atomic update. Common case Release was previously 1 atomic read + 1 atomic update, now just 1 atomic update. * Eliminate spins / waits / yields, which likely threaten some "lock free" benefits. Compare-exchange loops are only used in explicit Erase, and strict_capacity_limit=true Insert. Eviction uses opportunistic compare- exchange. * Relaxes some aggressiveness and guarantees. For example, * Duplicate Inserts will sometimes go undetected and the shadow duplicate will age out with eviction. * In many cases, the older Inserted value for a given cache key will be kept (i.e. Insert does not support overwrite). * Entries explicitly erased (rather than evicted) might not be freed immediately in some rare cases. * With strict_capacity_limit=false, capacity limit is not tracked/enforced as precisely as LRUCache, but is self-correcting and should only deviate by a very small number of extra or fewer entries. * Use smaller "computed default" number of cache shards in many cases, because benefits to larger usage tracking / eviction pools outweigh the small cost of more lock-free atomic contention. The improvement in CPU and I/O is dramatic in some limit-memory cases. * Even without the sharding change, the eviction algorithm is likely more effective than LRU overall because it's more stateful, even though the "hot path" state tracking for it is essentially free with ref counting. It is like a generalized CLOCK with aging (see code comments). I don't have performance numbers showing a specific improvement, but in theory, for a Poisson access pattern to each block, keeping some state allows better estimation of time to next access (Poisson interval) than strict LRU. The bounded randomness in CLOCK can also reduce "cliff" effect for repeated range scans approaching and exceeding cache size. ## Hot path algorithm comparison Rough descriptions, focusing on number and kind of atomic operations: * Old `Lookup()` (2-5 atomic updates per probe): ``` Loop: Increment internal ref count at slot If possible hit: Check flags atomic (and non-atomic fields) If cache hit: Three distinct updates to 'flags' atomic Increment refs for internal-to-external Return Decrement internal ref count while atomic read 'displacements' > 0 ``` * New `Lookup()` (1-2 atomic updates per probe): ``` Loop: Increment acquire counter in meta word (optimistic) If visible entry (already read meta word): If match (read non-atomic fields): Return Else: Decrement acquire counter in meta word Else if invisible entry (rare, already read meta word): Decrement acquire counter in meta word while atomic read 'displacements' > 0 ``` * Old `Release()` (1 atomic update, conditional on atomic read, rarely more): ``` Read atomic ref count If last reference and invisible (rare): Use CAS etc. to remove Return Else: Decrement ref count ``` * New `Release()` (1 unconditional atomic update, rarely more): ``` Increment release counter in meta word If last reference and invisible (rare): Use CAS etc. to remove Return ``` ## Performance test setup Build DB with ``` TEST_TMPDIR=/dev/shm ./db_bench -benchmarks=fillrandom -num=30000000 -disable_wal=1 -bloom_bits=16 ``` Test with ``` TEST_TMPDIR=/dev/shm ./db_bench -benchmarks=readrandom -readonly -num=30000000 -bloom_bits=16 -cache_index_and_filter_blocks=1 -cache_size=${CACHE_MB}000000 -duration 60 -threads=$THREADS -statistics ``` Numbers on a single socket Skylake Xeon system with 48 hardware threads, DEBUG_LEVEL=0 PORTABLE=0. Very similar story on a dual socket system with 80 hardware threads. Using (every 2nd) Fibonacci MB cache sizes to sample the territory between powers of two. Configurations: base: LRUCache before this change, but with db_bench change to default cache_numshardbits=-1 (instead of fixed at 6) folly: LRUCache before this change, with folly enabled (distributed mutex) but on an old compiler (sorry) gt_clock: experimental ClockCache before this change new_clock: experimental ClockCache with this change ## Performance test results First test "hot path" read performance, with block cache large enough for whole DB: 4181MB 1thread base -> kops/s: 47.761 4181MB 1thread folly -> kops/s: 45.877 4181MB 1thread gt_clock -> kops/s: 51.092 4181MB 1thread new_clock -> kops/s: 53.944 4181MB 16thread base -> kops/s: 284.567 4181MB 16thread folly -> kops/s: 249.015 4181MB 16thread gt_clock -> kops/s: 743.762 4181MB 16thread new_clock -> kops/s: 861.821 4181MB 24thread base -> kops/s: 303.415 4181MB 24thread folly -> kops/s: 266.548 4181MB 24thread gt_clock -> kops/s: 975.706 4181MB 24thread new_clock -> kops/s: 1205.64 (~= 24 * 53.944) 4181MB 32thread base -> kops/s: 311.251 4181MB 32thread folly -> kops/s: 274.952 4181MB 32thread gt_clock -> kops/s: 1045.98 4181MB 32thread new_clock -> kops/s: 1370.38 4181MB 48thread base -> kops/s: 310.504 4181MB 48thread folly -> kops/s: 268.322 4181MB 48thread gt_clock -> kops/s: 1195.65 4181MB 48thread new_clock -> kops/s: 1604.85 (~= 24 * 1.25 * 53.944) 4181MB 64thread base -> kops/s: 307.839 4181MB 64thread folly -> kops/s: 272.172 4181MB 64thread gt_clock -> kops/s: 1204.47 4181MB 64thread new_clock -> kops/s: 1615.37 4181MB 128thread base -> kops/s: 310.934 4181MB 128thread folly -> kops/s: 267.468 4181MB 128thread gt_clock -> kops/s: 1188.75 4181MB 128thread new_clock -> kops/s: 1595.46 Whether we have just one thread on a quiet system or an overload of threads, the new version wins every time in thousand-ops per second, sometimes dramatically so. Mutex-based implementation quickly becomes contention-limited. New clock cache shows essentially perfect scaling up to number of physical cores (24), and then each hyperthreaded core adding about 1/4 the throughput of an additional physical core (see 48 thread case). Block cache miss rates (omitted above) are negligible across the board. With partitioned instead of full filters, the maximum speed-up vs. base is more like 2.5x rather than 5x. Now test a large block cache with low miss ratio, but some eviction is required: 1597MB 1thread base -> kops/s: 46.603 io_bytes/op: 1584.63 miss_ratio: 0.0201066 max_rss_mb: 1589.23 1597MB 1thread folly -> kops/s: 45.079 io_bytes/op: 1530.03 miss_ratio: 0.019872 max_rss_mb: 1550.43 1597MB 1thread gt_clock -> kops/s: 48.711 io_bytes/op: 1566.63 miss_ratio: 0.0198923 max_rss_mb: 1691.4 1597MB 1thread new_clock -> kops/s: 51.531 io_bytes/op: 1589.07 miss_ratio: 0.0201969 max_rss_mb: 1583.56 1597MB 32thread base -> kops/s: 301.174 io_bytes/op: 1439.52 miss_ratio: 0.0184218 max_rss_mb: 1656.59 1597MB 32thread folly -> kops/s: 273.09 io_bytes/op: 1375.12 miss_ratio: 0.0180002 max_rss_mb: 1586.8 1597MB 32thread gt_clock -> kops/s: 904.497 io_bytes/op: 1411.29 miss_ratio: 0.0179934 max_rss_mb: 1775.89 1597MB 32thread new_clock -> kops/s: 1182.59 io_bytes/op: 1440.77 miss_ratio: 0.0185449 max_rss_mb: 1636.45 1597MB 128thread base -> kops/s: 309.91 io_bytes/op: 1438.25 miss_ratio: 0.018399 max_rss_mb: 1689.98 1597MB 128thread folly -> kops/s: 267.605 io_bytes/op: 1394.16 miss_ratio: 0.0180286 max_rss_mb: 1631.91 1597MB 128thread gt_clock -> kops/s: 691.518 io_bytes/op: 9056.73 miss_ratio: 0.0186572 max_rss_mb: 1982.26 1597MB 128thread new_clock -> kops/s: 1406.12 io_bytes/op: 1440.82 miss_ratio: 0.0185463 max_rss_mb: 1685.63 610MB 1thread base -> kops/s: 45.511 io_bytes/op: 2279.61 miss_ratio: 0.0290528 max_rss_mb: 615.137 610MB 1thread folly -> kops/s: 43.386 io_bytes/op: 2217.29 miss_ratio: 0.0289282 max_rss_mb: 600.996 610MB 1thread gt_clock -> kops/s: 46.207 io_bytes/op: 2275.51 miss_ratio: 0.0290057 max_rss_mb: 637.934 610MB 1thread new_clock -> kops/s: 48.879 io_bytes/op: 2283.1 miss_ratio: 0.0291253 max_rss_mb: 613.5 610MB 32thread base -> kops/s: 306.59 io_bytes/op: 2250 miss_ratio: 0.0288721 max_rss_mb: 683.402 610MB 32thread folly -> kops/s: 269.176 io_bytes/op: 2187.86 miss_ratio: 0.0286938 max_rss_mb: 628.742 610MB 32thread gt_clock -> kops/s: 855.097 io_bytes/op: 2279.26 miss_ratio: 0.0288009 max_rss_mb: 733.062 610MB 32thread new_clock -> kops/s: 1121.47 io_bytes/op: 2244.29 miss_ratio: 0.0289046 max_rss_mb: 666.453 610MB 128thread base -> kops/s: 305.079 io_bytes/op: 2252.43 miss_ratio: 0.0288884 max_rss_mb: 723.457 610MB 128thread folly -> kops/s: 269.583 io_bytes/op: 2204.58 miss_ratio: 0.0287001 max_rss_mb: 676.426 610MB 128thread gt_clock -> kops/s: 53.298 io_bytes/op: 8128.98 miss_ratio: 0.0292452 max_rss_mb: 956.273 610MB 128thread new_clock -> kops/s: 1301.09 io_bytes/op: 2246.04 miss_ratio: 0.0289171 max_rss_mb: 788.812 The new version is still winning every time, sometimes dramatically so, and we can tell from the maximum resident memory numbers (which contain some noise, by the way) that the new cache is not cheating on memory usage. IMPORTANT: The previous generation experimental clock cache appears to hit a serious bottleneck in the higher thread count configurations, presumably due to some of its waiting functionality. (The same bottleneck is not seen with partitioned index+filters.) Now we consider even smaller cache sizes, with higher miss ratios, eviction work, etc. 233MB 1thread base -> kops/s: 10.557 io_bytes/op: 227040 miss_ratio: 0.0403105 max_rss_mb: 247.371 233MB 1thread folly -> kops/s: 15.348 io_bytes/op: 112007 miss_ratio: 0.0372238 max_rss_mb: 245.293 233MB 1thread gt_clock -> kops/s: 6.365 io_bytes/op: 244854 miss_ratio: 0.0413873 max_rss_mb: 259.844 233MB 1thread new_clock -> kops/s: 47.501 io_bytes/op: 2591.93 miss_ratio: 0.0330989 max_rss_mb: 242.461 233MB 32thread base -> kops/s: 96.498 io_bytes/op: 363379 miss_ratio: 0.0459966 max_rss_mb: 479.227 233MB 32thread folly -> kops/s: 109.95 io_bytes/op: 314799 miss_ratio: 0.0450032 max_rss_mb: 400.738 233MB 32thread gt_clock -> kops/s: 2.353 io_bytes/op: 385397 miss_ratio: 0.048445 max_rss_mb: 500.688 233MB 32thread new_clock -> kops/s: 1088.95 io_bytes/op: 2567.02 miss_ratio: 0.0330593 max_rss_mb: 303.402 233MB 128thread base -> kops/s: 84.302 io_bytes/op: 378020 miss_ratio: 0.0466558 max_rss_mb: 1051.84 233MB 128thread folly -> kops/s: 89.921 io_bytes/op: 338242 miss_ratio: 0.0460309 max_rss_mb: 812.785 233MB 128thread gt_clock -> kops/s: 2.588 io_bytes/op: 462833 miss_ratio: 0.0509158 max_rss_mb: 1109.94 233MB 128thread new_clock -> kops/s: 1299.26 io_bytes/op: 2565.94 miss_ratio: 0.0330531 max_rss_mb: 361.016 89MB 1thread base -> kops/s: 0.574 io_bytes/op: 5.35977e+06 miss_ratio: 0.274427 max_rss_mb: 91.3086 89MB 1thread folly -> kops/s: 0.578 io_bytes/op: 5.16549e+06 miss_ratio: 0.27276 max_rss_mb: 96.8984 89MB 1thread gt_clock -> kops/s: 0.512 io_bytes/op: 4.13111e+06 miss_ratio: 0.242817 max_rss_mb: 119.441 89MB 1thread new_clock -> kops/s: 48.172 io_bytes/op: 2709.76 miss_ratio: 0.0346162 max_rss_mb: 100.754 89MB 32thread base -> kops/s: 5.779 io_bytes/op: 6.14192e+06 miss_ratio: 0.320399 max_rss_mb: 311.812 89MB 32thread folly -> kops/s: 5.601 io_bytes/op: 5.83838e+06 miss_ratio: 0.313123 max_rss_mb: 252.418 89MB 32thread gt_clock -> kops/s: 0.77 io_bytes/op: 3.99236e+06 miss_ratio: 0.236296 max_rss_mb: 396.422 89MB 32thread new_clock -> kops/s: 1064.97 io_bytes/op: 2687.23 miss_ratio: 0.0346134 max_rss_mb: 155.293 89MB 128thread base -> kops/s: 4.959 io_bytes/op: 6.20297e+06 miss_ratio: 0.323945 max_rss_mb: 823.43 89MB 128thread folly -> kops/s: 4.962 io_bytes/op: 5.9601e+06 miss_ratio: 0.319857 max_rss_mb: 626.824 89MB 128thread gt_clock -> kops/s: 1.009 io_bytes/op: 4.1083e+06 miss_ratio: 0.242512 max_rss_mb: 1095.32 89MB 128thread new_clock -> kops/s: 1224.39 io_bytes/op: 2688.2 miss_ratio: 0.0346207 max_rss_mb: 218.223 ^ Now something interesting has happened: the new clock cache has gained a dramatic lead in the single-threaded case, and this is because the cache is so small, and full filters are so big, that dividing the cache into 64 shards leads to significant (random) imbalances in cache shards and excessive churn in imbalanced shards. This new clock cache only uses two shards for this configuration, and that helps to ensure that entries are part of a sufficiently big pool that their eviction order resembles the single-shard order. (This effect is not seen with partitioned index+filters.) Even smaller cache size: 34MB 1thread base -> kops/s: 0.198 io_bytes/op: 1.65342e+07 miss_ratio: 0.939466 max_rss_mb: 48.6914 34MB 1thread folly -> kops/s: 0.201 io_bytes/op: 1.63416e+07 miss_ratio: 0.939081 max_rss_mb: 45.3281 34MB 1thread gt_clock -> kops/s: 0.448 io_bytes/op: 4.43957e+06 miss_ratio: 0.266749 max_rss_mb: 100.523 34MB 1thread new_clock -> kops/s: 1.055 io_bytes/op: 1.85439e+06 miss_ratio: 0.107512 max_rss_mb: 75.3125 34MB 32thread base -> kops/s: 3.346 io_bytes/op: 1.64852e+07 miss_ratio: 0.93596 max_rss_mb: 180.48 34MB 32thread folly -> kops/s: 3.431 io_bytes/op: 1.62857e+07 miss_ratio: 0.935693 max_rss_mb: 137.531 34MB 32thread gt_clock -> kops/s: 1.47 io_bytes/op: 4.89704e+06 miss_ratio: 0.295081 max_rss_mb: 392.465 34MB 32thread new_clock -> kops/s: 8.19 io_bytes/op: 3.70456e+06 miss_ratio: 0.20826 max_rss_mb: 519.793 34MB 128thread base -> kops/s: 2.293 io_bytes/op: 1.64351e+07 miss_ratio: 0.931866 max_rss_mb: 449.484 34MB 128thread folly -> kops/s: 2.34 io_bytes/op: 1.6219e+07 miss_ratio: 0.932023 max_rss_mb: 396.457 34MB 128thread gt_clock -> kops/s: 1.798 io_bytes/op: 5.4241e+06 miss_ratio: 0.324881 max_rss_mb: 1104.41 34MB 128thread new_clock -> kops/s: 10.519 io_bytes/op: 2.39354e+06 miss_ratio: 0.136147 max_rss_mb: 1050.52 As the miss ratio gets higher (say, above 10%), the CPU time spent in eviction starts to erode the advantage of using fewer shards (13% miss rate much lower than 94%). LRU's O(1) eviction time can eventually pay off when there's enough block cache churn: 13MB 1thread base -> kops/s: 0.195 io_bytes/op: 1.65732e+07 miss_ratio: 0.946604 max_rss_mb: 45.6328 13MB 1thread folly -> kops/s: 0.197 io_bytes/op: 1.63793e+07 miss_ratio: 0.94661 max_rss_mb: 33.8633 13MB 1thread gt_clock -> kops/s: 0.519 io_bytes/op: 4.43316e+06 miss_ratio: 0.269379 max_rss_mb: 100.684 13MB 1thread new_clock -> kops/s: 0.176 io_bytes/op: 1.54148e+07 miss_ratio: 0.91545 max_rss_mb: 66.2383 13MB 32thread base -> kops/s: 3.266 io_bytes/op: 1.65544e+07 miss_ratio: 0.943386 max_rss_mb: 132.492 13MB 32thread folly -> kops/s: 3.396 io_bytes/op: 1.63142e+07 miss_ratio: 0.943243 max_rss_mb: 101.863 13MB 32thread gt_clock -> kops/s: 2.758 io_bytes/op: 5.13714e+06 miss_ratio: 0.310652 max_rss_mb: 396.121 13MB 32thread new_clock -> kops/s: 3.11 io_bytes/op: 1.23419e+07 miss_ratio: 0.708425 max_rss_mb: 321.758 13MB 128thread base -> kops/s: 2.31 io_bytes/op: 1.64823e+07 miss_ratio: 0.939543 max_rss_mb: 425.539 13MB 128thread folly -> kops/s: 2.339 io_bytes/op: 1.6242e+07 miss_ratio: 0.939966 max_rss_mb: 346.098 13MB 128thread gt_clock -> kops/s: 3.223 io_bytes/op: 5.76928e+06 miss_ratio: 0.345899 max_rss_mb: 1087.77 13MB 128thread new_clock -> kops/s: 2.984 io_bytes/op: 1.05341e+07 miss_ratio: 0.606198 max_rss_mb: 898.27 gt_clock is clearly blowing way past its memory budget for lower miss rates and best throughput. new_clock also seems to be exceeding budgets, and this warrants more investigation but is not the use case we are targeting with the new cache. With partitioned index+filter, the miss ratio is much better, and although still high enough that the eviction CPU time is definitely offsetting mutex contention: 13MB 1thread base -> kops/s: 16.326 io_bytes/op: 23743.9 miss_ratio: 0.205362 max_rss_mb: 65.2852 13MB 1thread folly -> kops/s: 15.574 io_bytes/op: 19415 miss_ratio: 0.184157 max_rss_mb: 56.3516 13MB 1thread gt_clock -> kops/s: 14.459 io_bytes/op: 22873 miss_ratio: 0.198355 max_rss_mb: 63.9688 13MB 1thread new_clock -> kops/s: 16.34 io_bytes/op: 24386.5 miss_ratio: 0.210512 max_rss_mb: 61.707 13MB 128thread base -> kops/s: 289.786 io_bytes/op: 23710.9 miss_ratio: 0.205056 max_rss_mb: 103.57 13MB 128thread folly -> kops/s: 185.282 io_bytes/op: 19433.1 miss_ratio: 0.184275 max_rss_mb: 116.219 13MB 128thread gt_clock -> kops/s: 354.451 io_bytes/op: 23150.6 miss_ratio: 0.200495 max_rss_mb: 102.871 13MB 128thread new_clock -> kops/s: 295.359 io_bytes/op: 24626.4 miss_ratio: 0.212452 max_rss_mb: 121.109 Pull Request resolved: https://github.com/facebook/rocksdb/pull/10626 Test Plan: updated unit tests, stress/crash test runs including with TSAN, ASAN, UBSAN Reviewed By: anand1976 Differential Revision: D39368406 Pulled By: pdillinger fbshipit-source-id: 5afc44da4c656f8f751b44552bbf27bd3ca6fef9	2 years ago