Summary:
In `db_stress` profile the vast majority of CPU time is spent acquiring the `SyncPoint` mutex. I mistakenly assumed #3939 had fixed this mutex contention problem by disabling `SyncPoint` processing. But actually the lock was still being acquired just to check whether processing is enabled. We can avoid that overhead by using an atomic to track whether it's enabled.
Closes https://github.com/facebook/rocksdb/pull/3991
Differential Revision: D8393825
Pulled By: ajkr
fbshipit-source-id: 5bc4e3c722ee7304e7a9c2439998c456b05a6897
Summary:
Depending on the compression type, `CompressBlock` calls the compress method for each compression type. It calls ZSTD_Compress for both kZSTD and kZSTDNotFinalCompression (https://github.com/facebook/rocksdb/blob/master/table/block_based_table_builder.cc#L169).
However currently ZSTD_Compress only expects the type to be kZSTD and this is causing assert failures and crashes. The same also applies to ZSTD_Uncompress.
Closes https://github.com/facebook/rocksdb/pull/3964
Differential Revision: D8308715
Pulled By: miasantreble
fbshipit-source-id: e5125f53edb829c9c33733167bec74e4793d0782
Summary:
PR https://github.com/facebook/rocksdb/pull/3838 made some changes that triggers lint warnings.
Run `make format` to fix formatting as suggested by siying .
Also piggyback two changes:
1) fix singleton destruction order for windows and posix env
2) fix two clang warnings
Closes https://github.com/facebook/rocksdb/pull/3954
Differential Revision: D8272041
Pulled By: miasantreble
fbshipit-source-id: 7c4fd12bd17aac13534520de0c733328aa3c6c9f
Summary:
format_version=3 changes the format of SST index. This is however not being tested currently since tests only work with the default format_version which is currently 2. The patch extends the most related tests to also test for format_version=3.
Closes https://github.com/facebook/rocksdb/pull/3942
Differential Revision: D8238413
Pulled By: maysamyabandeh
fbshipit-source-id: 915725f55753dd8e9188e802bf471c23645ad035
Summary:
Ensure the PosixEnv singleton is destroyed first since its destructor waits for background threads to all complete. This ensures background threads cannot hit sync points after the SyncPoint singleton is destroyed, which was previously possible.
Closes https://github.com/facebook/rocksdb/pull/3951
Differential Revision: D8265295
Pulled By: ajkr
fbshipit-source-id: 7738dd458c5d993a78377dd0420e82badada81ab
Summary:
Windows does not have LD_PRELOAD mechanism to override all memory allocation functions and ZSTD makes use of C-tuntime calloc. During flushes and compactions default system allocator fragments and the system slows down considerably.
For builds with jemalloc we employ an advanced ZSTD context creation API that re-directs memory allocation to jemalloc. To reduce the cost of context creation on each block we cache ZSTD context within the block based table builder while a new SST file is being built, this will help all platform builds including those w/o jemalloc. This avoids system allocator fragmentation and improves the performance.
The change does not address random reads and currently on Windows reads with ZSTD regress as compared with SNAPPY compression.
Closes https://github.com/facebook/rocksdb/pull/3838
Differential Revision: D8229794
Pulled By: miasantreble
fbshipit-source-id: 719b622ab7bf4109819bc44f45ec66f0dd3ee80d
Summary:
Index blocks have the same format as data blocks. The keys therefore similarly to the keys in the data blocks are internal keys, which means that in addition to the user key it also has 8 bytes that encodes sequence number and value type. This extra 8 bytes however is not necessary in index blocks since the index keys act as an separator between two data blocks. The only exception is when the last key of a block and the first key of the next block share the same user key, in which the sequence number is required to act as a separator.
The patch excludes the sequence from index keys only if the above special case does not happen for any of the index keys. It then records that in the property block. The reader looks at the property block to see if it should expect sequence numbers in the keys of the index block.s
Closes https://github.com/facebook/rocksdb/pull/3894
Differential Revision: D8118775
Pulled By: maysamyabandeh
fbshipit-source-id: 915479f028b5799ca91671d67455ecdefbd873bd
Summary:
Previously we were using -1 as the default for every library, which was legacy from our zlib options. That worked for a while, but after zstd introduced a146ee04ae, it started giving poor compression ratios by default in zstd.
This PR adds a constant to RocksDB public API, `CompressionOptions::kDefaultCompressionLevel`, which will get translated to the default value specific to the compression library being used in "util/compression.h". The constant uses a number that appears to be larger than any library's maximum compression level.
Closes https://github.com/facebook/rocksdb/pull/3895
Differential Revision: D8125780
Pulled By: ajkr
fbshipit-source-id: 2db157a89118cd4f94577c2f4a0a5ff31c8391c6
Summary:
`RangeDelAggregator` holds the pointers returned by `BlockIter::key()` and `BlockIter::value()` so requires the data to which they point is pinned. `BlockIter::key()` points into block memory and is guaranteed to be pinned if and only if prefix encoding is disabled (or, equivalently, restart interval is set to one). I think `BlockIter::value()` is always pinned. Added an assert for these and removed the wrong TODO about increasing restart interval, which would enable key prefix encoding and break the assertion.
Closes https://github.com/facebook/rocksdb/pull/3875
Differential Revision: D8063667
Pulled By: ajkr
fbshipit-source-id: 60b5ebcc0cdd610dd6aad9e74a23378793672c41
Summary:
Right now ReverseBytewiseComparator::FindShortestSeparator() doesn't really shorten key, and ReverseBytewiseComparator::FindShortestSuccessor() seems to return wrong results. The code is confusing too as it uses BytewiseComparatorImpl::FindShortestSeparator() but the function actually won't do anything if the the first key is larger than the second.
Implement ReverseBytewiseComparator::FindShortestSeparator() and override ReverseBytewiseComparator::FindShortestSuccessor() to be empty.
Closes https://github.com/facebook/rocksdb/pull/3836
Differential Revision: D7959762
Pulled By: siying
fbshipit-source-id: 93acb621c16ce6f23e087ae4e19f7d84d1254683
Summary:
Currently manual_wal_flush if set in the options will be used only for the wal files created during wal switch. The configuration thus does not affect the first wal file. The patch fixes that and also update the related unit tests.
This PR is built on top of https://github.com/facebook/rocksdb/pull/3756
Closes https://github.com/facebook/rocksdb/pull/3824
Differential Revision: D7909153
Pulled By: maysamyabandeh
fbshipit-source-id: 024ed99d2555db06bf096c902b998e432bb7b9ce
Summary:
Previously `DBOptions::use_direct_io_for_flush_and_compaction=true` combined with `DBOptions::use_direct_reads=false` could cause RocksDB to simultaneously read from two file descriptors for the same file, where background reads used direct I/O and foreground reads used buffered I/O. Our measurements found this mixed-mode I/O negatively impacted foreground read perf, compared to when only buffered I/O was used.
This PR makes the mixed-mode I/O situation impossible by repurposing `DBOptions::use_direct_io_for_flush_and_compaction` to only apply to background writes, and `DBOptions::use_direct_reads` to apply to all reads. There is no risk of direct background direct writes happening simultaneously with buffered reads since we never read from and write to the same file simultaneously.
Closes https://github.com/facebook/rocksdb/pull/3829
Differential Revision: D7915443
Pulled By: ajkr
fbshipit-source-id: 78bcbf276449b7e7766ab6b0db246f789fb1b279
Summary:
`ReadaheadRandomAccessFile` had an unwritten assumption, which was that its wrapped file's `Read()` function always copies into the provided scratch buffer. Actually this was not true when the wrapped file was `PosixMmapReadableFile`, whose `Read()` implementation does no copying and instead returns a `Slice` pointing directly into the `mmap`'d memory region. This PR:
- prevents `ReadaheadRandomAccessFile` from ever wrapping mmap readable files
- adds an assert for the assumption `ReadaheadRandomAccessFile` makes about the wrapped file's use of scratch buffer
Closes https://github.com/facebook/rocksdb/pull/3813
Differential Revision: D7891513
Pulled By: ajkr
fbshipit-source-id: dc64a55222d6af280c39a1852ee39e9e9d7cde7d
Summary:
Rollback was disabled in stress test since there was a concurrency issue in WritePrepared rollback algorithm. The issue is fixed by caching the column family handles in WritePrepared to skip getting them from the db when needed for rollback.
Tested by running transaction stress test under tsan.
Closes https://github.com/facebook/rocksdb/pull/3785
Differential Revision: D7793727
Pulled By: maysamyabandeh
fbshipit-source-id: d81ab6fda0e53186ca69944cfe0712ce4869451e
Summary:
sync parent directory after deleting a file in delete scheduler. Otherwise, trim speed may not be as smooth as what we want.
Closes https://github.com/facebook/rocksdb/pull/3767
Differential Revision: D7760136
Pulled By: siying
fbshipit-source-id: ec131d53b61953f09c60d67e901e5eeb2716b05f
Summary:
WritePrepared rollback implementation is not ready to be invoked in the middle of workload. This is due the lack of synchronization to obtain the cf handle from db. Temporarily disabling this until the problem with rollback is fixed.
Closes https://github.com/facebook/rocksdb/pull/3772
Differential Revision: D7769041
Pulled By: maysamyabandeh
fbshipit-source-id: 0e3b0ce679bc2afba82e653a40afa3f045722754
Summary:
Background activities like compaction can negatively affect
latency of higher-priority tasks like request processing. To avoid this,
rocksdb already lowers the IO priority of background threads on Linux
systems. While this takes care of typical IO-bound systems, it does not
help much when CPU (temporarily) becomes the bottleneck. This is
especially likely when using more expensive compression settings.
This patch adds an API to allow for lowering the CPU priority of
background threads, modeled on the IO priority API. Benchmarks (see
below) show significant latency and throughput improvements when CPU
bound. As a result, workloads with some CPU usage bursts should benefit
from lower latencies at a given utilization, or should be able to push
utilization higher at a given request latency target.
A useful side effect is that compaction CPU usage is now easily visible
in common tools, allowing for an easier estimation of the contribution
of compaction vs. request processing threads.
As with IO priority, the implementation is limited to Linux, degrading
to a no-op on other systems.
Closes https://github.com/facebook/rocksdb/pull/3763
Differential Revision: D7740096
Pulled By: gwicke
fbshipit-source-id: e5d32373e8dc403a7b0c2227023f9ce4f22b413c
Summary:
Currently WritePrepared rolls back a transaction with prepare sequence number prepare_seq by i) write a single rollback batch with rollback_seq, ii) add <rollback_seq, rollback_seq> to commit cache, iii) remove prepare_seq from PrepareHeap.
This is correct assuming that there is no snapshot taken when a transaction is rolled back. This is the case the way MySQL does rollback which is after recovery. Otherwise if max_evicted_seq advances the prepare_seq, the live snapshot might assume data as committed since it does not find them in CommitCache.
The change is to simply add <prepare_seq. rollback_seq> to commit cache before removing prepare_seq from PrepareHeap. In this way if max_evicted_seq advances prpeare_seq, the existing mechanism that we have to check evicted entries against live snapshots will make sure that the live snapshot will not see the data of rolled back transaction.
Closes https://github.com/facebook/rocksdb/pull/3745
Differential Revision: D7696193
Pulled By: maysamyabandeh
fbshipit-source-id: c9a2d46341ddc03554dded1303520a1cab74ef9c
Summary:
Right now in `SyncClosedLogs`, `CopyFile`, and `AddRecord`, where `Sync` and `Append` are invoked in a loop, the error status are not checked. This could lead to potential corruption as later calls will overwrite the error status.
Closes https://github.com/facebook/rocksdb/pull/3740
Differential Revision: D7678848
Pulled By: miasantreble
fbshipit-source-id: 4b0b412975989dfe80348f73217b9c4122a4bd77
Summary:
Previously threads were named "rocksdb:bg\<index in thread pool\>", so the first thread in all thread pools would be named "rocksdb:bg0". Users want to be able to distinguish threads used for flush (high-pri) vs regular compaction (low-pri) vs compaction to bottom-level (bottom-pri). So I changed the thread naming convention to include the thread-pool priority.
Closes https://github.com/facebook/rocksdb/pull/3702
Differential Revision: D7581415
Pulled By: ajkr
fbshipit-source-id: ce04482b6acd956a401ef22dc168b84f76f7d7c1
Summary:
db_stress was already capable running transactions by setting use_txn. Running it under stress showed a couple of problems fixed in this patch.
- The uncommitted transaction must be either rolled back or commit after recovery.
- Current implementation of WritePrepared transaction cannot handle cf drop before crash. Clarified that in the comments and added safety checks. When running with use_txn, clear_column_family_one_in must be set to 0.
Closes https://github.com/facebook/rocksdb/pull/3733
Differential Revision: D7654419
Pulled By: maysamyabandeh
fbshipit-source-id: a024bad80a9dc99677398c00d29ff17d4436b7f3
Summary:
this PR fixes a few failed contbuild:
1. ASAN memory leak in Block::NewIterator (table/block.cc:429). the proper destruction of first_level_iter_ and second_level_iter_ of two_level_iterator.cc is missing from the code after the refactoring in https://github.com/facebook/rocksdb/pull/3406
2. various unused param errors introduced by https://github.com/facebook/rocksdb/pull/3662
3. updated comment for `ForceReleaseCachedEntry` to emphasize the use of `force_erase` flag.
Closes https://github.com/facebook/rocksdb/pull/3718
Reviewed By: maysamyabandeh
Differential Revision: D7621192
Pulled By: miasantreble
fbshipit-source-id: 476c94264083a0730ded957c29de7807e4f5b146
Summary:
This PR comments out the rest of the unused arguments which allow us to turn on the -Wunused-parameter flag. This is the second part of a codemod relating to https://github.com/facebook/rocksdb/pull/3557.
Closes https://github.com/facebook/rocksdb/pull/3662
Differential Revision: D7426121
Pulled By: Dayvedde
fbshipit-source-id: 223994923b42bd4953eb016a0129e47560f7e352
Summary:
In this change, an option to set different paths for different column families is added.
This option is set via cf_paths setting of ColumnFamilyOptions. This option will work in a similar fashion to db_paths setting. Cf_paths is a vector of Dbpath values which contains a pair of the absolute path and target size. Multiple levels in a Column family can go to different paths if cf_paths has more than one path.
To maintain backward compatibility, if cf_paths is not specified for a column family, db_paths setting will be used. Note that, if db_paths setting is also not specified, RocksDB already has code to use db_name as the only path.
Changes :
1) A new member "cf_paths" is added to ImmutableCfOptions. This is set, based on cf_paths setting of ColumnFamilyOptions and db_paths setting of ImmutableDbOptions. This member is used to identify the path information whenever files are accessed.
2) Validation checks are added for cf_paths setting based on existing checks for db_paths setting.
3) DestroyDB, PurgeObsoleteFiles etc. are edited to support multiple cf_paths.
4) Unit tests are added appropriately.
Closes https://github.com/facebook/rocksdb/pull/3102
Differential Revision: D6951697
Pulled By: ajkr
fbshipit-source-id: 60d2262862b0a8fd6605b09ccb0da32bb331787d
Summary:
Level Compaction with TTL.
As of today, a file could exist in the LSM tree without going through the compaction process for a really long time if there are no updates to the data in the file's key range. For example, in certain use cases, the keys are not actually "deleted"; instead they are just set to empty values. There might not be any more writes to this "deleted" key range, and if so, such data could remain in the LSM for a really long time resulting in wasted space.
Introducing a TTL could solve this problem. Files (and, in turn, data) older than TTL will be scheduled for compaction when there is no other background work. This will make the data go through the regular compaction process and get rid of old unwanted data.
This also has the (good) side-effect of all the data in the non-bottommost level being newer than ttl, and all data in the bottommost level older than ttl. It could lead to more writes while reducing space.
This functionality can be controlled by the newly introduced column family option -- ttl.
TODO for later:
- Make ttl mutable
- Extend TTL to Universal compaction as well? (TTL is already supported in FIFO)
- Maybe deprecate CompactionOptionsFIFO.ttl in favor of this new ttl option.
Closes https://github.com/facebook/rocksdb/pull/3591
Differential Revision: D7275442
Pulled By: sagar0
fbshipit-source-id: dcba484717341200d419b0953dafcdf9eb2f0267
Summary:
Manual compactions should be cancelled, just like scheduled compactions are cancelled, if sfm->EnoughRoomForCompaction is not true.
Closes https://github.com/facebook/rocksdb/pull/3670
Differential Revision: D7457683
Pulled By: amytai
fbshipit-source-id: 669b02fdb707f75db576d03d2c818fb98d1876f5
Summary:
This was failing the build on windows with zstd, warning treated as an error, 32-bit shift implicitly converted to 64-bit.
Closes https://github.com/facebook/rocksdb/pull/3624
Differential Revision: D7307883
Pulled By: gfosco
fbshipit-source-id: 68110e9b5b1b59b668dec6cf86b67556402574e7
Summary:
Provide a block_align option in BlockBasedTableOptions to allow
alignment of SST file data blocks. This will avoid higher
IOPS/throughput load due to < 4KB data blocks spanning 2 4KB pages.
When this option is set to true, the block alignment is set to lower of
block size and 4KB.
Closes https://github.com/facebook/rocksdb/pull/3502
Differential Revision: D7400897
Pulled By: anand1976
fbshipit-source-id: 04cc3bd144e88e3431a4f97604e63ad7a0f06d44
Summary:
Add `bytes_max_delete_chunk` in SstFileManager so that we can drop a large file in multiple batches.
Closes https://github.com/facebook/rocksdb/pull/3640
Differential Revision: D7358679
Pulled By: siying
fbshipit-source-id: ef17f0da2f5723dbece2669485a9b91b3edc0bb7
Summary:
Implemented PositionedAppend() and use_direct_io() for TestWritableFile.
With these changes, FaultInjectionTestEnv can be used with DirectIO enabled.
Closes https://github.com/facebook/rocksdb/pull/3586
Differential Revision: D7244305
Pulled By: yiwu-arbug
fbshipit-source-id: f6b7aece53daa0f9977bc684164a0693693e514c
Summary:
This diff handles cases where compaction causes an ENOSPC error.
This does not handle corner cases where another background job is started while compaction is running, and the other background job triggers ENOSPC, although we do allow the user to provision for these background jobs with SstFileManager::SetCompactionBufferSize.
It also does not handle the case where compaction has finished and some other background job independently triggers ENOSPC.
Usage: Functionality is inside SstFileManager. In particular, users should set SstFileManager::SetMaxAllowedSpaceUsage, which is the reference highwatermark for determining whether to cancel compactions.
Closes https://github.com/facebook/rocksdb/pull/3449
Differential Revision: D7016941
Pulled By: amytai
fbshipit-source-id: 8965ab8dd8b00972e771637a41b4e6c645450445
Summary:
Now that files scheduled for deletion are kept in the same directory, we don't need to constrain deletion scheduler to `db_paths[0]`. Previously this was done because there was a separate trash directory, and this constraint prevented files from being accidentally copied to another filesystem when they're scheduled for deletion.
Closes https://github.com/facebook/rocksdb/pull/3544
Differential Revision: D7093786
Pulled By: ajkr
fbshipit-source-id: 202f5c92d925eafebec1281fb95bb5828d33414f
Summary:
In attempting to build a static lib for use in iOS, I ran in to lots of type errors between uint64_t and size_t. This PR contains the changes I made to get `TARGET_OS=IOS make static_lib` to succeed while also getting Xcode to build successfully with the resulting `librocksdb.a` library imported.
This also compiles for me on macOS and tests fine, but I'm really not sure if I made the correct decisions about where to `static_cast` and where to change types.
Also up for discussion: is iOS worth supporting? Getting the static lib is just part one, we aren't providing any bridging headers or wrappers like the ObjectiveRocks project, it won't be a great experience.
Closes https://github.com/facebook/rocksdb/pull/3503
Differential Revision: D7106457
Pulled By: gfosco
fbshipit-source-id: 82ac2073de7e1f09b91f6b4faea91d18bd311f8e
Summary:
This patch addressed several issues.
Portability including db_test std::thread -> port::Thread Cc: @
and %z to ROCKSDB portable macro. Cc: maysamyabandeh
Implement Env::AreFilesSame
Make the implementation of file unique number more robust
Get rid of C-runtime and go directly to Windows API when dealing
with file primitives.
Implement GetSectorSize() and aling unbuffered read on the value if
available.
Adjust Windows Logger for the new interface, implement CloseImpl() Cc: anand1976
Fix test running script issue where $status var was of incorrect scope
so the failures were swallowed and not reported.
DestroyDB() creates a logger and opens a LOG file in the directory
being cleaned up. This holds a lock on the folder and the cleanup is
prevented. This fails one of the checkpoin tests. We observe the same in production.
We close the log file in this change.
Fix DBTest2.ReadAmpBitmapLiveInCacheAfterDBClose failure where the test
attempts to open a directory with NewRandomAccessFile which does not
work on Windows.
Fix DBTest.SoftLimit as it is dependent on thread timing. CC: yiwu-arbug
Closes https://github.com/facebook/rocksdb/pull/3552
Differential Revision: D7156304
Pulled By: siying
fbshipit-source-id: 43db0a757f1dfceffeb2b7988043156639173f5b
Summary:
Move DuplicateDetector and SetComparator to its own header file in util. It would also address a complaint in the unity test.
Closes https://github.com/facebook/rocksdb/pull/3567
Differential Revision: D7163268
Pulled By: maysamyabandeh
fbshipit-source-id: 6ddf82773473646dbbc1284ae601a78c4907c778
Summary:
Red diff to remove existing implementation of garbage collection. The current approach is reference counting kind of approach and require a lot of effort to get the size counter right on compaction and deletion. I'm going to go with a simple mark-sweep kind of approach and will send another PR for that.
CompactionEventListener was added solely for blob db and it adds complexity and overhead to compaction iterator. Removing it as well.
Closes https://github.com/facebook/rocksdb/pull/3551
Differential Revision: D7130190
Pulled By: yiwu-arbug
fbshipit-source-id: c3a375ad2639a3f6ed179df6eda602372cc5b8df
Summary:
The recent Logger::Close() and DBImpl::Close() implementation rely on
calling the CloseImpl() virtual function from the destructor, which will
not work. Refactor the implementation to have a private close helper
function in derived classes that can be called by both CloseImpl() and
the destructor.
Closes https://github.com/facebook/rocksdb/pull/3528
Reviewed By: gfosco
Differential Revision: D7049303
Pulled By: anand1976
fbshipit-source-id: 76a64cbf403209216dfe4864ecf96b5d7f3db9f4
Summary:
Deadlock: a memtable flush holds DB::mutex_ and calls ThreadLocalPtr::Scrape(), which locks ThreadLocalPtr mutex; meanwhile, a thread exit handler locks ThreadLocalPtr mutex and calls SuperVersionUnrefHandle, which tries to lock DB::mutex_.
This deadlock is hit all the time on our workload. It blocks our release.
In general, the problem is that ThreadLocalPtr takes an arbitrary callback and calls it while holding a lock on a global mutex. The same global mutex is (at least in some cases) locked by almost all ThreadLocalPtr methods, on any instance of ThreadLocalPtr. So, there'll be a deadlock if the callback tries to do anything to any instance of ThreadLocalPtr, or waits for another thread to do so.
So, probably the only safe way to use ThreadLocalPtr callbacks is to do only do simple and lock-free things in them.
This PR fixes the deadlock by making sure that local_sv_ never holds the last reference to a SuperVersion, and therefore SuperVersionUnrefHandle never has to do any nontrivial cleanup.
I also searched for other uses of ThreadLocalPtr to see if they may have similar bugs. There's only one other use, in transaction_lock_mgr.cc, and it looks fine.
Closes https://github.com/facebook/rocksdb/pull/3510
Reviewed By: sagar0
Differential Revision: D7005346
Pulled By: al13n321
fbshipit-source-id: 37575591b84f07a891d6659e87e784660fde815f
Summary:
Calling `std::vector::reserve()` causes memory to be reallocated and then data to be moved. It was called prior to adding every block. This reallocation could be done a huge amount of times, e.g., for users with large index blocks.
Instead, we can simply use `std::vector::emplace_back()` in such a way that preserves the no-memory-leak guarantee, while letting the vector decide when to reallocate space. Now I see reallocation/moving happen O(logN) times, rather than O(N) times, where N is the final size of vector.
Closes https://github.com/facebook/rocksdb/pull/3508
Differential Revision: D6994228
Pulled By: ajkr
fbshipit-source-id: ab7c11e13ff37c8c6c8249be7a79566a4068cd27
Summary:
- removed a few unneeded variables
- fused some variable declarations and their assignments
- fixed right-trimming code in string_util.cc to not underflow
- simplifed an assertion
- move non-nullptr check assertion before dereferencing of that pointer
- pass an std::string function parameter by const reference instead of by value (avoiding potential copy)
Closes https://github.com/facebook/rocksdb/pull/3507
Differential Revision: D7004679
Pulled By: sagar0
fbshipit-source-id: 52944952d9b56dfcac3bea3cd7878e315bb563c4
Summary:
We don't do fsync() after truncate in direct I/O writeable file (in fact we don't do any fsync ever). This can cause metadata not persistent to disk after the file is generated. We call it instead.
Closes https://github.com/facebook/rocksdb/pull/3500
Differential Revision: D6981482
Pulled By: siying
fbshipit-source-id: 7e2b591b7e5dd1b96fc0775515b8b9e6092980ef
Summary:
Now we suppress alignment UBSAN error as a whole. Suppressing 3-way CRC and murmurhash feels a better idea than turning off alignment check as a whole.
Closes https://github.com/facebook/rocksdb/pull/3495
Differential Revision: D6971273
Pulled By: siying
fbshipit-source-id: 080b59fed6df494b9f622ef7cb5d42d39e6a8cdf
Summary:
ForwardIterator::SVCleanup() sometimes didn't pin superversion when it was supposed to. See the added test for the scenario. Here's the ASAN output of the added test without the fix (using `COMPILE_WITH_ASAN=1 make`): https://pastebin.com/9rD0Ywws
Closes https://github.com/facebook/rocksdb/pull/3415
Differential Revision: D6817414
Pulled By: al13n321
fbshipit-source-id: bc80c44ea78a3a1fa885dfa448a26111f91afb24
Summary:
`ReadaheadRandomAccessFile` is used by iterators for file reads in several cases, like in compaction when `compaction_readahead_size > 0` or `use_direct_io_for_flush_and_compaction == true`, or in user iterator when `ReadOptions::readahead_size > 0`. `ReadaheadRandomAccessFile` maintains an internal buffer for readahead data. It assumes that, if the buffer's length is less than `ReadaheadRandomAccessFile::readahead_size_`, which is fixed in the constructor, then EOF has been reached so it doesn't try reading further.
Recently, d938226af4 started calling `RandomAccessFile::Prefetch` with various lengths: 8KB, 16KB, etc. When the `RandomAccessFile` is a `ReadaheadRandomAccessFile`, it triggers the above condition and incorrectly determines EOF. If a block is partially in the readahead buffer and EOF is incorrectly decided, the result is a truncated data block.
The problem is reproducible:
```
TEST_TMPDIR=/data/compaction_bench ./db_bench -benchmarks=fillrandom -write_buffer_size=1048576 -target_file_size_base=1048576 -block_size=18384 -use_direct_io_for_flush_and_compaction=true
...
put error: Corruption: truncated block read from /data/compaction_bench/dbbench/000014.sst offset 20245, expected 10143 bytes, got 8427
```
Closes https://github.com/facebook/rocksdb/pull/3454
Differential Revision: D6869405
Pulled By: ajkr
fbshipit-source-id: 87001c299e7600a37c0dcccbd0368e0954c929cf
Summary:
FilePrefetchBuffer::Prefetch is currently rounds the offset up which does not fit its new use cases in prefetching index/filter blocks, as it would skips over some the offsets that were requested to be prefetched. This patch rounds down instead.
Fixes#3180
Closes https://github.com/facebook/rocksdb/pull/3413
Differential Revision: D6816392
Pulled By: maysamyabandeh
fbshipit-source-id: 3aaeaf59c55d72b61dacfae6d4a8e65eccb3c553
Summary:
Currently, the only way to close an open DB is to destroy the DB
object. There is no way for the caller to know the status. In one
instance, the destructor encountered an error due to failure to
close a log file on HDFS. In order to prevent silent failures, we add
DB::Close() that calls CloseImpl() which must be implemented by its
descendants.
The main failure point in the destructor is closing the log file. This
patch also adds a Close() entry point to Logger in order to get status.
When DBOptions::info_log is allocated and owned by the DBImpl, it is
explicitly closed by DBImpl::CloseImpl().
Closes https://github.com/facebook/rocksdb/pull/3348
Differential Revision: D6698158
Pulled By: anand1976
fbshipit-source-id: 9468e2892553eb09c4c41b8723f590c0dbd8ab7d
Summary:
This fixes the following warnings when compiled with GCC7:
util/transaction_test_util.cc: In static member function ‘static rocksdb::Status rocksdb::RandomTransactionInserter::DBGet(rocksdb::DB*, rocksdb::Transaction*, rocksdb::ReadOptions&, uint16_t, uint64_t, bool, uint64_t*, std::__cxx11::string*, bool*)’:
util/transaction_test_util.cc:75:8: error: ‘snprintf’ output may be truncated before the last format character [-Werror=format-truncation=]
Status RandomTransactionInserter::DBGet(
^~~~~~~~~~~~~~~~~~~~~~~~~
util/transaction_test_util.cc:84:11: note: ‘snprintf’ output between 5 and 6 bytes into a destination of size 5
snprintf(prefix_buf, sizeof(prefix_buf), "%.4u", set_i + 1);
~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
util/transaction_test_util.cc: In static member function ‘static rocksdb::Status rocksdb::RandomTransactionInserter::Verify(rocksdb::DB*, uint16_t, uint64_t, bool, rocksdb::Random64*)’:
util/transaction_test_util.cc:245:8: error: ‘snprintf’ output may be truncated before the last format character [-Werror=format-truncation=]
Status RandomTransactionInserter::Verify(DB* db, uint16_t num_sets,
^~~~~~~~~~~~~~~~~~~~~~~~~
util/transaction_test_util.cc:268:13: note: ‘snprintf’ output between 5 and 6 bytes into a destination of size 5
snprintf(prefix_buf, sizeof(prefix_buf), "%.4u", set_i + 1);
~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Closes https://github.com/facebook/rocksdb/pull/3295
Differential Revision: D6609411
Pulled By: maysamyabandeh
fbshipit-source-id: 33f0add471056eb59db2f8bd4366e6dfbb1a187d
Summary:
**# Summary**
RocksDB uses SSE crc32 intrinsics to calculate the crc32 values but it does it in single way fashion (not pipelined on single CPU core). Intel's whitepaper () published an algorithm that uses 3-way pipelining for the crc32 intrinsics, then use pclmulqdq intrinsic to combine the values. Because pclmulqdq has overhead on its own, this algorithm will show perf gains on buffers larger than 216 bytes, which makes RocksDB a perfect user, since most of the buffers RocksDB call crc32c on is over 4KB. Initial db_bench show tremendous CPU gain.
This change uses the 3-way SSE algorithm by default. The old SSE algorithm is now behind a compiler tag NO_THREEWAY_CRC32C. If user compiles the code with NO_THREEWAY_CRC32C=1 then the old SSE Crc32c algorithm would be used. If the server does not have SSE4.2 at the run time the slow way (Non SSE) will be used.
**# Performance Test Results**
We ran the FillRandom and ReadRandom benchmarks in db_bench. ReadRandom is the point of interest here since it calculates the CRC32 for the in-mem buffers. We did 3 runs for each algorithm.
Before this change the CRC32 value computation takes about 11.5% of total CPU cost, and with the new 3-way algorithm it reduced to around 4.5%. The overall throughput also improved from 25.53MB/s to 27.63MB/s.
1) ReadRandom in db_bench overall metrics
PER RUN
Algorithm | run | micros/op | ops/sec |Throughput (MB/s)
3-way | 1 | 4.143 | 241387 | 26.7
3-way | 2 | 3.775 | 264872 | 29.3
3-way | 3 | 4.116 | 242929 | 26.9
FastCrc32c|1 | 4.037 | 247727 | 27.4
FastCrc32c|2 | 4.648 | 215166 | 23.8
FastCrc32c|3 | 4.352 | 229799 | 25.4
AVG
Algorithm | Average of micros/op | Average of ops/sec | Average of Throughput (MB/s)
3-way | 4.01 | 249,729 | 27.63
FastCrc32c | 4.35 | 230,897 | 25.53
2) Crc32c computation CPU cost (inclusive samples percentage)
PER RUN
Implementation | run | TotalSamples | Crc32c percentage
3-way  | 1  | 4,572,250,000 | 4.37%
3-way  | 2  | 3,779,250,000 | 4.62%
3-way  | 3  | 4,129,500,000 | 4.48%
FastCrc32c   | 1  | 4,663,500,000 | 11.24%
FastCrc32c   | 2  | 4,047,500,000 | 12.34%
FastCrc32c   | 3  | 4,366,750,000 | 11.68%
**# Test Plan**
make -j64 corruption_test && ./corruption_test
By default it uses 3-way SSE algorithm
NO_THREEWAY_CRC32C=1 make -j64 corruption_test && ./corruption_test
make clean && DEBUG_LEVEL=0 make -j64 db_bench
make clean && DEBUG_LEVEL=0 NO_THREEWAY_CRC32C=1 make -j64 db_bench
Closes https://github.com/facebook/rocksdb/pull/3173
Differential Revision: D6330882
Pulled By: yingsu00
fbshipit-source-id: 8ec3d89719533b63b536a736663ca6f0dd4482e9
Summary:
added `ThreadType::BOTTOM_PRIORITY` which is used in the `ThreadStatus` object to indicate the thread is used for bottom-pri compactions. Previously there was a bug where we mislabeled such threads as `ThreadType::LOW_PRIORITY`.
Closes https://github.com/facebook/rocksdb/pull/3270
Differential Revision: D6559428
Pulled By: ajkr
fbshipit-source-id: 96b1a50a9c19492b1a5fd1b77cf7061a6f9f1d1c
Summary:
There were a few places where MSVC's implicit truncation warnings were getting triggered, which was causing the MSVC build to fail due to warnings being treated as errors. This resolves the issues by making the truncations in some places explicit, and by making it so there are no truncations of literals.
Fixes#3239
Supersedes #3259
Closes https://github.com/facebook/rocksdb/pull/3273
Reviewed By: yiwu-arbug
Differential Revision: D6569204
Pulled By: Orvid
fbshipit-source-id: c188cf1cf98d9acb6d94b71875041cc81f8ff088
Summary:
DeleteScheduler::MarkAsTrash() don't handle existing .trash files correctly
This cause rocksdb to not being able to delete existing .trash files on restart
Closes https://github.com/facebook/rocksdb/pull/3261
Differential Revision: D6548003
Pulled By: IslamAbdelRahman
fbshipit-source-id: c3800639412e587a690062c63076a5a08881e0e6
Summary:
I started adding gflags support for cmake on linux and got frustrated that I'd need to duplicate the build_detect_platform logic, which determines namespace based on attempting compilation. We can do it differently -- use the GFLAGS_NAMESPACE macro if available, and if not, that indicates it's an old gflags version without configurable namespace so we can simply hardcode "google".
Closes https://github.com/facebook/rocksdb/pull/3212
Differential Revision: D6456973
Pulled By: ajkr
fbshipit-source-id: 3e6d5bde3ca00d4496a120a7caf4687399f5d656
Summary:
Allow users to configure the trash-to-DB size ratio limit, so
that ratelimits for deletes can be enforced even when larger portions of
the database are being deleted.
Closes https://github.com/facebook/rocksdb/pull/3158
Differential Revision: D6304897
Pulled By: gdavidsson
fbshipit-source-id: a28dd13059ebab7d4171b953ed91ce383a84d6b3
Summary:
This confused some users who were getting compression type from the logs.
Closes https://github.com/facebook/rocksdb/pull/3153
Differential Revision: D6294964
Pulled By: ajkr
fbshipit-source-id: 3c813376d33682dc6ccafc9a78df1a2e2528985e
Summary:
The current implementation of PinnableSlice move assignment have an issue #3163. We are moving away from it instead of try to get the move assignment right, since it is too tricky.
Closes https://github.com/facebook/rocksdb/pull/3164
Differential Revision: D6319201
Pulled By: yiwu-arbug
fbshipit-source-id: 8f3279021f3710da4a4caa14fd238ed2df902c48
Summary:
After move assignment, we need to re-initialized the moved PinnableSlice.
Also update blob_db_impl.cc to not reuse the moved PinnableSlice since it is supposed to be in an undefined state after move.
Closes https://github.com/facebook/rocksdb/pull/3127
Differential Revision: D6238585
Pulled By: yiwu-arbug
fbshipit-source-id: bd99f2e37406c4f7de160c7dee6a2e8126bc224e
Summary:
util/concurrent_arena.h:
CID 1396145 (#1 of 1): Uninitialized pointer field (UNINIT_CTOR)
2. uninit_member: Non-static class member free_begin_ is not initialized in this constructor nor in any functions that it calls.
94 Shard() : allocated_and_unused_(0) {}
util/dynamic_bloom.cc:
1. Condition hash_func == NULL, taking true branch.
CID 1322821 (#1 of 1): Uninitialized pointer field (UNINIT_CTOR)
3. uninit_member: Non-static class member data_ is not initialized in this constructor nor in any functions that it calls.
47 hash_func_(hash_func == nullptr ? &BloomHash : hash_func) {}
48
util/file_reader_writer.h:
204 private:
205 AlignedBuffer buffer_;
member_not_init_in_gen_ctor: The compiler-generated constructor for this class does not initialize buffer_offset_.
206 uint64_t buffer_offset_;
CID 1418246 (#1 of 1): Uninitialized scalar field (UNINIT_CTOR)
member_not_init_in_gen_ctor: The compiler-generated constructor for this class does not initialize buffer_len_.
207 size_t buffer_len_;
208};
util/thread_local.cc:
341#endif
CID 1322795 (#1 of 1): Uninitialized scalar field (UNINIT_CTOR)
3. uninit_member: Non-static class member pthread_key_ is not initialized in this constructor nor in any functions that it calls.
342}
40struct ThreadData {
2. uninit_member: Non-static class member next is not initialized in this constructor nor in any functions that it calls.
CID 1400668 (#1 of 1): Uninitialized pointer field (UNINIT_CTOR)
4. uninit_member: Non-static class member prev is not initialized in this constructor nor in any functions that it calls.
41 explicit ThreadData(ThreadLocalPtr::StaticMeta* _inst) : entries(), inst(_inst) {}
42 std::vector<Entry> entries;
1. member_decl: Class member declaration for next.
43 ThreadData* next;
3. member_decl: Class member declaration for prev.
44 ThreadData* prev;
45 ThreadLocalPtr::StaticMeta* inst;
46};
Closes https://github.com/facebook/rocksdb/pull/3123
Differential Revision: D6233566
Pulled By: sagar0
fbshipit-source-id: aa2068790ea69787a0035c0db39d59b0c25108db
Summary:
Instead of using samples directly, we now support passing the samples through zstd's dictionary generator when `CompressionOptions::zstd_max_train_bytes` is set to nonzero. If set to zero, we will use the samples directly as the dictionary -- same as before.
Note this is the first step of #2987, extracted into a separate PR per reviewer request.
Closes https://github.com/facebook/rocksdb/pull/3057
Differential Revision: D6116891
Pulled By: ajkr
fbshipit-source-id: 70ab13cc4c734fa02e554180eed0618b75255497
Summary:
SstFileManager move files that need to be deleted into a trash directory.
Deprecate this behaviour and instead add ".trash" extension to files that need to be deleted
Closes https://github.com/facebook/rocksdb/pull/2970
Differential Revision: D5976805
Pulled By: IslamAbdelRahman
fbshipit-source-id: 27374ece4315610b2792c30ffcd50232d4c9a343
Summary:
ColumnFamilyOptions::compaction_options_fifo and all its sub-fields can be set dynamically now.
Some of the ways in which the fifo compaction options can be set are:
- `SetOptions({{"compaction_options_fifo", "{max_table_files_size=1024}"}})`
- `SetOptions({{"compaction_options_fifo", "{ttl=600;}"}})`
- `SetOptions({{"compaction_options_fifo", "{max_table_files_size=1024;ttl=600;}"}})`
- `SetOptions({{"compaction_options_fifo", "{max_table_files_size=51;ttl=49;allow_compaction=true;}"}})`
Most of the code has been made generic enough so that it could be reused later to make universal options (and other such nested defined-types) dynamic with very few lines of parsing/serializing code changes.
Introduced a few new functions like `ParseStruct`, `SerializeStruct` and `GetStringFromStruct`.
The duplicate code in `GetStringFromDBOptions` and `GetStringFromColumnFamilyOptions` has been moved into `GetStringFromStruct`. So they become just simple wrappers now.
Closes https://github.com/facebook/rocksdb/pull/3006
Differential Revision: D6058619
Pulled By: sagar0
fbshipit-source-id: 1e8f78b3374ca5249bb4f3be8a6d3bb4cbc52f92
Summary:
When I impl my own comparator, and build in release mode.
The following compile error occurs.
undefined reference to `typeinfo for rocksdb::Comparator'
This fix allows users build with RTTI off when has their own comparator.
Closes https://github.com/facebook/rocksdb/pull/3008
Differential Revision: D6077354
Pulled By: yiwu-arbug
fbshipit-source-id: 914c26dbab72f0ad1f0e15f8666a3fb2f10bfed8
Summary:
it turns out that, with older GCC shipped from centos7, the SSE42
intrinsics are not available even with "target" specified. so we
need to pass "-msse42" for checking compiler's sse4.2 support and
for building crc32c.cc which uses sse4.2 intrinsics for crc32.
Signed-off-by: Kefu Chai <tchaikov@gmail.com>
Closes https://github.com/facebook/rocksdb/pull/2950
Differential Revision: D6032298
Pulled By: siying
fbshipit-source-id: 124c946321043661b3fb0a70b6cdf4c9c5126ab4
Summary:
Dynamic adjustment of rate limit according to demand for background I/O. It increases by a factor when limiter is drained too frequently, and decreases by the same factor when limiter is not drained frequently enough. The parameters for this behavior are fixed in `GenericRateLimiter::Tune`. Other changes:
- make rate limiter's `Env*` configurable for testing
- track num drain intervals in RateLimiter so we don't have to rely on stats, which may be shared across different DB instances from the ones that share the RateLimiter.
Closes https://github.com/facebook/rocksdb/pull/2899
Differential Revision: D5858704
Pulled By: ajkr
fbshipit-source-id: cc2bac30f85e7f6fd63655d0a6732ef9ed7403b1
Summary:
Previously the thread pool might be non-empty after joining since concurrent submissions could spawn new threads. This problem didn't affect our background flush/compaction thread pools because the `shutting_down_` flag prevented new jobs from being submitted during/after joining. But I wanted to be able to reuse the `ThreadPool` without such external synchronization.
Closes https://github.com/facebook/rocksdb/pull/2953
Differential Revision: D5951920
Pulled By: ajkr
fbshipit-source-id: 0efec7d0056d36d1338367da75e8b0c089bbc973
Summary:
Merging iterator invokes InternalKeyComparator.Compare() frequently to heap merge. By making InternalKeyComparator final and merging iterator to directly use InternalKeyComparator rather than through Iterator interface, we can give compiler a choice to avoid one more virtual function call if possible. I ran readseq benchmark in memory-only use case to make sure the performance at least doesn't regress.
I have to disable the final key word in debug build, as a hack test class depends on overriding the class.
Closes https://github.com/facebook/rocksdb/pull/2860
Differential Revision: D5800461
Pulled By: siying
fbshipit-source-id: ab876f22a09bb5c560740911412336e0e25ccb53
Summary:
if we enable SSE42 globally when compiling the tree for preparing a
portable binary, which could be running on CPU w/o SSE42 instructions
even the GCC on the building host is able to emit SSE42 code, this leads
to illegal instruction errors on machines not supporting SSE42. to solve
this problem, crc32 detects the supported instruction at runtime, and
selects the supported CRC32 implementation according to the result of
`cpuid`. but intrinics like "_mm_crc32_u64()" will not be available
unless the "target" machine is appropriately specified in the command
line, like "-msse42", or using the "target" attribute.
we could pass "-msse42" only when compiling crc32c.cc, and allow the
compiler to generate the SSE42 instructions, but we are still at the
risk of executing illegal instructions on machines does not support
SSE42 if the compiler emits code that is not guarded by our runtime
detection. and we need to do the change in both Makefile and CMakefile.
or, we can use GCC's "target" attribute to enable the machine specific
instructions on certain function. in this way, we have finer grained
control of the used "target". and no need to change the makefiles. so
we don't need to duplicate the changes on both makefile and cmake as
the previous approach.
this problem surfaces when preparing a package for GNU/Linux distribution,
and we only applies to optimization for SSE42, so using a feature
only available on GCC/Clang is not that formidable.
Closes https://github.com/facebook/rocksdb/pull/2807
Differential Revision: D5786084
Pulled By: siying
fbshipit-source-id: bca5c0f877b8d6fb55f58f8f122254a26422843d
Summary:
Right now, if direct I/O is enabled, prefetching the last 512KB cannot be applied, except compaction inputs or readahead is enabled for iterators. This can create a lot of I/O for HDD cases. To solve the problem, the 512KB is prefetched in block based table if direct I/O is enabled. The prefetched buffer is passed in totegher with random access file reader, so that we try to read from the buffer before reading from the file. This can be extended in the future to support flexible user iterator readahead too.
Closes https://github.com/facebook/rocksdb/pull/2708
Differential Revision: D5593091
Pulled By: siying
fbshipit-source-id: ee36ff6d8af11c312a2622272b21957a7b5c81e7
Summary:
When we had a single thread pool for compactions, a thread could be busy for a long time (minutes) executing a compaction involving the bottom level. In multi-instance setups, the entire thread pool could be consumed by such bottom-level compactions. Then, top-level compactions (e.g., a few L0 files) would be blocked for a long time ("head-of-line blocking"). Such top-level compactions are critical to prevent compaction stalls as they can quickly reduce number of L0 files / sorted runs.
This diff introduces a bottom-priority queue for universal compactions including the bottom level. This alleviates the head-of-line blocking situation for fast, top-level compactions.
- Added `Env::Priority::BOTTOM` thread pool. This feature is only enabled if user explicitly configures it to have a positive number of threads.
- Changed `ThreadPoolImpl`'s default thread limit from one to zero. This change is invisible to users as we call `IncBackgroundThreadsIfNeeded` on the low-pri/high-pri pools during `DB::Open` with values of at least one. It is necessary, though, for bottom-pri to start with zero threads so the feature is disabled by default.
- Separated `ManualCompaction` into two parts in `PrepickedCompaction`. `PrepickedCompaction` is used for any compaction that's picked outside of its execution thread, either manual or automatic.
- Forward universal compactions involving last level to the bottom pool (worker thread's entry point is `BGWorkBottomCompaction`).
- Track `bg_bottom_compaction_scheduled_` so we can wait for bottom-level compactions to finish. We don't count them against the background jobs limits. So users of this feature will get an extra compaction for free.
Closes https://github.com/facebook/rocksdb/pull/2580
Differential Revision: D5422916
Pulled By: ajkr
fbshipit-source-id: a74bd11f1ea4933df3739b16808bb21fcd512333
Summary:
Replace dynamic_cast<> so that users can choose to build with RTTI off, so that they can save several bytes per object, and get tiny more memory available.
Some nontrivial changes:
1. Add Comparator::GetRootComparator() to get around the internal comparator hack
2. Add the two experiemental functions to DB
3. Add TableFactory::GetOptionString() to avoid unnecessary casting to get the option string
4. Since 3 is done, move the parsing option functions for table factory to table factory files too, to be symmetric.
Closes https://github.com/facebook/rocksdb/pull/2645
Differential Revision: D5502723
Pulled By: siying
fbshipit-source-id: fd13cec5601cf68a554d87bfcf056f2ffa5fbf7c
Summary:
This fixes OOMs that we (logdevice) are currently having in production.
SkipListRep constructor does a couple small allocations from ConcurrentArena (see InlineSkipList constructor). ConcurrentArena would sometimes allocate an entire block for that, which is a few megabytes (we use Options::arena_block_size = 4 MB). So an empty memtable can take take 4 MB of memory. We have ~40k column families (spread across 15 DB instances), so 4 MB per empty memtable easily OOMs a machine for us.
This PR makes ConcurrentArena always allocate from Arena's inline block when possible. So as long as InlineSkipList's initial allocations are below 2 KB there would be no blocks allocated for empty memtables.
Closes https://github.com/facebook/rocksdb/pull/2569
Differential Revision: D5404029
Pulled By: al13n321
fbshipit-source-id: 568ec22a3fd1a485c06123f6b2dfc5e9ef67cd23
Summary:
This reverts the previous commit 1d7048c598, which broke the build.
Did a `git revert 1d7048c`.
Closes https://github.com/facebook/rocksdb/pull/2627
Differential Revision: D5476473
Pulled By: sagar0
fbshipit-source-id: 4756ff5c0dfc88c17eceb00e02c36176de728d06
