Summary:
L0 score is based on size target and number of files. The size target
used is `max_bytes_for_level_base`. However, the base level's size can
dynamically expand in write burst mode. In fact, it can expand so much
that L0->Lbase becomes the highest fanout in target sizes. This doesn't
make sense from an efficiency perspective, so this PR bounds the
L0->Lbase fanout to the smoothed level multiplier. The L0 scoring based
on file count remains unchanged.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7325
Test Plan:
contrived benchmark that exhibits the problem:
```
$ TEST_TMPDIR=/data/users/andrewkr/ ./db_bench -benchmarks=filluniquerandom,readrandom -write_buffer_size=1048576 -target_file_size_base=1048576 -max_bytes_for_level_base=4194304 -level0_file_num_compaction_trigger=4 -level_compaction_dynamic_level_bytes=true -compression_type=none -max_background_jobs=12 -rate_limiter_bytes_per_sec=104857600 -benchmark_write_rate_limit=10485760 -num=100000000
```
Results:
- "Burst W-Amp" is the write-amp near the end of the fillrandom benchmark
- "Total W-Amp" is the write-amp after readrandom has run a while and all levels no longer need compaction
Branch | Burst W-Amp | Total W-Amp | fillrandom (MB/s)
-- | -- | -- | --
master | 20.2 | 21.5 | 4.7
dynamic-l0-score | 12.6 | 14.1 | 7.2
Reviewed By: siying
Differential Revision: D23412935
Pulled By: ajkr
fbshipit-source-id: f91f2067188e432dd39deab02f1c56f195057a0e
Summary:
Replace FSRandomAccessFile pointer with FSRandomAccessFilePtr
object in RandomAccessFileReader.
This new object wraps FSRandomAccessFile pointer.
Objective: If tracing is enabled, FSRandomAccessFile Ptr returns
FSRandomAccessFileTracingWrapper pointer that includes all necessary
information in IORecord and calls underlying FileSystem and invokes
IOTracer to dump that record in a binary file. If tracing is disabled
then, underlying FileSystem pointer is returned directly.
FSRandomAccessFilePtr wrapper class is added to bypass the FSRandomAccessFileWrapper when
tracing is disabled.
Test Plan: make check -j64
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7192
Reviewed By: anand1976
Differential Revision: D23356867
Pulled By: akankshamahajan15
fbshipit-source-id: 48f31168166a17a7444b40be44a9a9d4a5c7182c
Summary:
More tests now pass. When in doubt, I added a TODO comment to check what should happen with an ignored error.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7305
Reviewed By: akankshamahajan15
Differential Revision: D23301262
Pulled By: ajkr
fbshipit-source-id: 5f120edc7393560aefc0633250277bbc7e8de9e6
Summary:
This test uses database functionality and required more extensive work to get it to pass than the other tests. The DB functionality required for this test now passes the check.
When it was unclear what the proper behavior was for unchecked status codes, a TODO was added.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7283
Reviewed By: akankshamahajan15
Differential Revision: D23251497
Pulled By: ajkr
fbshipit-source-id: 52b79629bdafa0a58de8ead1d1d66f141b331523
Summary:
This diff contains following changes:
1. Replace `FSSequentialFile` pointer with `FSSequentialFilePtr` object that wraps `FSSequentialFile` pointer in `SequenceFileReader`.
Objective: If tracing is enabled, `FSSequentialFilePtr` returns `FSSequentialFileTracingWrapper` pointer that includes all necessary information in `IORecord` and calls underlying FileSystem and invokes `IOTracer` to dump that record in a binary file. If tracing is disabled then, underlying `FileSystem` pointer is returned directly. `FSSequentialFilePtr` wrapper class is added to bypass the `FSSequentialFileTracingWrapper` when tracing is disabled.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7190
Test Plan:
make check -j64
COMPILE_WITH_TSAN=1 make check -j64
Reviewed By: anand1976
Differential Revision: D23059616
Pulled By: akankshamahajan15
fbshipit-source-id: 1564b94dd1297cd0fbfe2ed5c9cc3e20f7395301
Summary:
As part of the IOTracing project, this PR
1. Caches "FileSystemPtr" object(wrapper class that returns file system pointer based on tracing enabled) instead of "FileSystem" pointer.
2. FileSystemPtr object is created using FileSystem pointer and IOTracer
pointer.
3. IOTracer shared_ptr is created in DBImpl and it is passed to different classes through constructor.
4. When tracing is enabled through DB::StartIOTrace, FileSystemPtr
returns FileSystemTracingWrapper pointer for tracing purpose and when
it is disabled underlying FileSystem pointer is returned.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7180
Test Plan:
make check -j64
COMPILE_WITH_TSAN=1 make check -j64
Reviewed By: anand1976
Differential Revision: D22987117
Pulled By: akankshamahajan15
fbshipit-source-id: 6073617e4c2d5bc363914f3a1f55ae3b0a58fbf1
Summary:
`VersionStorageInfo::AddFile` currently has a debug-mode consistency
check to make sure the newly added file does not overlap with the
previous one (for levels below L0). Considering that
`VersionBuilder::CheckConsistency` also performs similar checks (in
fact, those checks are more comprehensive and cover L0 as well), this
check is redundant. The patch removes it and also cleans up `AddFile` a
little.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7237
Test Plan: `make check`
Reviewed By: riversand963
Differential Revision: D23041937
Pulled By: ltamasi
fbshipit-source-id: e00665f3b83bfd17f86c54c238800f3d77d739bd
Summary:
`WalAddition`, `WalDeletion` are defined in `wal_version.h` and used in `VersionEdit`.
`WalAddition` is used to represent events of creating a new WAL (no size, just log number), or closing a WAL (with size).
`WalDeletion` is used to represent events of deleting or archiving a WAL, it means the WAL is no longer alive (won't be replayed during recovery).
`WalSet` is the set of alive WALs kept in `VersionSet`.
1. Why use `WalDeletion` instead of relying on `MinLogNumber` to identify outdated WALs
On recovery, we can compute `MinLogNumber()` based on the log numbers kept in MANIFEST, any log with number < MinLogNumber can be ignored. So it seems that we don't need to persist `WalDeletion` to MANIFEST, since we can ignore the WALs based on MinLogNumber.
But the `MinLogNumber()` is actually a lower bound, it does not exactly mean that logs starting from MinLogNumber must exist. This is because in a corner case, when a column family is empty and never flushed, its log number is set to the largest log number, but not persisted in MANIFEST. So let's say there are 2 column families, when creating the DB, the first WAL has log number 1, so it's persisted to MANIFEST for both column families. Then CF 0 is empty and never flushed, CF 1 is updated and flushed, so a new WAL with log number 2 is created and persisted to MANIFEST for CF 1. But CF 0's log number in MANIFEST is still 1. So on recovery, MinLogNumber is 1, but since log 1 only contains data for CF 1, and CF 1 is flushed, log 1 might have already been deleted from disk.
We can make `MinLogNumber()` be the exactly minimum log number that must exist, by persisting the most recent log number for empty column families that are not flushed. But if there are N such column families, then every time a new WAL is created, we need to add N records to MANIFEST.
In current design, a record is persisted to MANIFEST only when WAL is created, closed, or deleted/archived, so the number of WAL related records are bounded to 3x number of WALs.
2. Why keep `WalSet` in `VersionSet` instead of applying the `VersionEdit`s to `VersionStorageInfo`
`VersionEdit`s are originally designed to track the addition and deletion of SST files. The SST files are related to column families, each column family has a list of `Version`s, and each `Version` keeps the set of active SST files in `VersionStorageInfo`.
But WALs are a concept of DB, they are not bounded to specific column families. So logically it does not make sense to store WALs in a column family's `Version`s.
Also, `Version`'s purpose is to keep reference to SST / blob files, so that they are not deleted until there is no version referencing them. But a WAL is deleted regardless of version references.
So we keep the WALs in `VersionSet` for the purpose of writing out the DB state's snapshot when creating new MANIFESTs.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7164
Test Plan:
make version_edit_test && ./version_edit_test
make wal_edit_test && ./wal_edit_test
Reviewed By: ltamasi
Differential Revision: D22677936
Pulled By: cheng-chang
fbshipit-source-id: 5a3b6890140e572ffd79eb37e6e4c3c32361a859
Summary:
IteratorIterator::IsOutOfBound() and IteratorIterator::MayBeOutOfUpperBound() are two functions that related to upper bound check. It is hard for users to reason about this complexity. Consolidate the two functions into one and assign an enum as results to improve readability.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7200
Test Plan: Run all existing test. Would run crash test with atomic for a while.
Reviewed By: anand1976
Differential Revision: D22833181
fbshipit-source-id: a0c724267056adbd0476bde74650e6c7226077e6
Summary:
Previously, a `ReadOptions` object was stored in every `BlockBasedTableIterator`
and every `LevelIterator`. This redundancy consumes extra memory,
resulting in the `Arena` making more allocations, and iteration
observing worse cache performance.
This PR migrates callers of `NewInternalIterator()` and
`MakeInputIterator()` to provide a `ReadOptions` object guaranteed to
outlive the returned iterator. When the iterator's lifetime will be managed by the
user, this lifetime guarantee is achieved by storing the `ReadOptions`
value in `ArenaWrappedDBIter`. Then, sub-iterators of `NewInternalIterator()` and
`MakeInputIterator()` can hold a reference-to-const `ReadOptions`.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7210
Test Plan:
- `make check` under ASAN and valgrind
- benchmark: on a DB with 2 L0 files and 3 L1+ levels, this PR reduced `Arena` allocation 4792 -> 4160 bytes.
Reviewed By: anand1976
Differential Revision: D22861323
Pulled By: ajkr
fbshipit-source-id: 54aebb3e89c872eeab0f5793b4b6e42878d093ce
Summary:
NextAndGetResult() is not implemented in memtable and is very simply implemented in level iterator. The result is that for a normal leveled iterator, performance regression will be observed for calling PrepareValue() for most iterator Next(). Mitigate the problem by implementing the function for both iterators. In level iterator, the implementation cannot be perfect as when calling file iterator's SeekToFirst() we don't have information about whether the value is prepared. Fortunately, the first key should not cause a big portion of the CPu.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7179
Test Plan: Run normal crash test for a while.
Reviewed By: anand1976
Differential Revision: D22783840
fbshipit-source-id: c19f45cdf21b756190adef97a3b66ccde3936e05
Summary:
There currently exist multiple `GetChildren()` calls in `DBImpl::Recover()`, which can be expensive in cases of distributed file systems.
This pull request try to call `DBImpl::Recover()` of each necessary directory only _once_ and reuse the results in the places of repeated calls in current code.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7044
Test Plan:
Run `make check` and use the default test suite. The modified code should be semantically identical to the current code. As a proof of this solution, we may optionally deploy the system onto a (real or simulated) distributed system and expect reduced latency caused by manifest fetching.
(WIP)
Reviewed By: riversand963
Differential Revision: D22419925
Pulled By: roghnin
fbshipit-source-id: d3774fbfbc246c5527101bc16747eb5c90919886
Summary:
After https://github.com/facebook/rocksdb/issues/6949 , VersionSet::io_status_ can be concurrently accessed by multiple
threads without lock, causing tsan test to fail. For example, a bg flush thread
resets io_status_ before calling LogAndApply(), while another thread already in
the process of LogAndApply() reads io_status_. This is a bug.
We do not have to reset io_status_ each time we call LogAndApply(). io_status_
is part of the state of VersionSet, and it indicates the outcome of preceding
MANIFEST/CURRENT files IO operations. Its value should be updated only when:
1. MANIFEST/CURRENT files IO fail for the first time.
2. MANIFEST/CURRENT files IO succeed as part of recovering from a prior
failure without process restart, e.g. calling Resume().
Test Plan (devserver):
COMPILE_WITH_TSAN=1 make check
COMPILE_WITH_TSAN=1 make db_test2
./db_test2 --gtest_filter=DBTest2.CompactionStall
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7034
Reviewed By: zhichao-cao
Differential Revision: D22247137
Pulled By: riversand963
fbshipit-source-id: 77b83e05390f3ee3cd2d96d3fdd6fe4f225e3216
Summary:
This PR provides preliminary support for handling IO error during MANIFEST write.
File write/sync is not guaranteed to be atomic. If we encounter an IOError while writing/syncing to the MANIFEST file, we cannot be sure about the state of the MANIFEST file. The version edits may or may not have reached the file. During cleanup, if we delete the newly-generated SST files referenced by the pending version edit(s), but the version edit(s) actually are persistent in the MANIFEST, then next recovery attempt will process the version edits(s) and then fail since the SST files have already been deleted.
One approach is to truncate the MANIFEST after write/sync error, so that it is safe to delete the SST files. However, file truncation may not be supported on certain file systems. Therefore, we take the following approach.
If an IOError is detected during MANIFEST write/sync, we disable file deletions for the faulty database. Depending on whether the IOError is retryable (set by underlying file system), either RocksDB or application can call `DB::Resume()`, or simply shutdown and restart. During `Resume()`, RocksDB will try to switch to a new MANIFEST and write all existing in-memory version storage in the new file. If this succeeds, then RocksDB may proceed. If all recovery is completed, then file deletions will be re-enabled.
Note that multiple threads can call `LogAndApply()` at the same time, though only one of them will be going through the process MANIFEST write, possibly batching the version edits of other threads. When the leading MANIFEST writer finishes, all of the MANIFEST writing threads in this batch will have the same IOError. They will all call `ErrorHandler::SetBGError()` in which file deletion will be disabled.
Possible future directions:
- Add an `ErrorContext` structure so that it is easier to pass more info to `ErrorHandler`. Currently, as in this example, a new `BackgroundErrorReason` has to be added.
Test plan (dev server):
make check
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6949
Reviewed By: anand1976
Differential Revision: D22026020
Pulled By: riversand963
fbshipit-source-id: f3c68a2ef45d9b505d0d625c7c5e0c88495b91c8
Summary:
https://github.com/facebook/rocksdb/issues/5411 refactored `VersionSet::Recover` but introduced a bug, explained as follows.
Before, once a checksum mismatch happens, `reporter` will set `s` to be non-ok. Therefore, Recover will stop processing the MANIFEST any further.
```
// Correct
// Inside Recover
LogReporter reporter;
reporter.status = &s;
log::Reader reader(..., reporter);
while (reader.ReadRecord() && s.ok()) {
...
}
```
The bug is that, the local variable `s` in `ReadAndRecover` won't be updated by `reporter` while reading the MANIFEST. It is possible that the reader sees a checksum mismatch in a record, but `ReadRecord` retries internally read and finds the next valid record. The mismatched record will be ignored and no error is reported.
```
// Incorrect
// Inside Recover
LogReporter reporter;
reporter.status = &s;
log::Reader reader(..., reporter);
s = ReadAndRecover(reader, ...);
// Inside ReadAndRecover
Status s; // Shadows the s in Recover.
while (reader.ReadRecord() && s.ok()) {
...
}
```
`LogReporter` can use a separate `log_read_status` to track the errors while reading the MANIFEST. RocksDB can process more MANIFEST entries only if `log_read_status.ok()`.
Test plan (devserver):
make check
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6996
Reviewed By: ajkr
Differential Revision: D22105746
Pulled By: riversand963
fbshipit-source-id: b22f717a423457a41ca152a242abbb64cf91fc38
Summary:
Memory pinned by `pin_l0_filter_and_index_blocks_in_cache` needs to be predictable based on user config. This PR makes sure
we do not pin extra memory for large files generated by intra-L0 (see https://github.com/facebook/rocksdb/issues/6889).
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6911
Test Plan: unit test
Reviewed By: siying
Differential Revision: D21835818
Pulled By: ajkr
fbshipit-source-id: a11a088549d06bed8aacc2548d266e5983f0ead4
Summary:
When MultiGet is called with duplicate keys, and the key matches the
largest key in an SST file and the value type is merge, only the first
instance of the duplicate key is returned with correct results. This is
due to the incorrect assumption that if a key in a batch is equal to the
largest key in the file, the next key cannot be present in that file.
Tests:
Add a new unit test
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6953
Reviewed By: cheng-chang
Differential Revision: D21935898
Pulled By: anand1976
fbshipit-source-id: a2cc327a15150e23fd997546ca64d1c33021cb4c
Summary:
The implementation of GetApproximateSizes was inconsistent in
its treatment of the size of non-data blocks of SST files, sometimes
including and sometimes now. This was at its worst with large portion
of table file used by filters and querying a small range that crossed
a table boundary: the size estimate would include large filter size.
It's conceivable that someone might want only to know the size in terms
of data blocks, but I believe that's unlikely enough to ignore for now.
Similarly, there's no evidence the internal function AppoximateOffsetOf
is used for anything other than a one-sided ApproximateSize, so I intend
to refactor to remove redundancy in a follow-up commit.
So to fix this, GetApproximateSizes (and implementation details
ApproximateSize and ApproximateOffsetOf) now consistently include in
their returned sizes a portion of table file metadata (incl filters
and indexes) based on the size portion of the data blocks in range. In
other words, if a key range covers data blocks that are X% by size of all
the table's data blocks, returned approximate size is X% of the total
file size. It would technically be more accurate to attribute metadata
based on number of keys, but that's not computationally efficient with
data available and rarely a meaningful difference.
Also includes miscellaneous comment improvements / clarifications.
Also included is a new approximatesizerandom benchmark for db_bench.
No significant performance difference seen with this change, whether ~700 ops/sec with cache_index_and_filter_blocks and small cache or ~150k ops/sec without cache_index_and_filter_blocks.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6784
Test Plan:
Test added to DBTest.ApproximateSizesFilesWithErrorMargin.
Old code running new test...
[ RUN ] DBTest.ApproximateSizesFilesWithErrorMargin
db/db_test.cc:1562: Failure
Expected: (size) <= (11 * 100), actual: 9478 vs 1100
Other tests updated to reflect consistent accounting of metadata.
Reviewed By: siying
Differential Revision: D21334706
Pulled By: pdillinger
fbshipit-source-id: 6f86870e45213334fedbe9c73b4ebb1d8d611185
Summary:
Does what it says on the can: the patch adds a hash map to `VersionStorageInfo`
that maps file numbers to file locations, i.e. (level, position in level) pairs. This
will enable stricter consistency checks in `VersionBuilder`. The patch also fixes
all the unit tests that used duplicate file numbers in a version (which would trigger
an assertion with the new code).
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6862
Test Plan:
`make check`
`make whitebox_crash_test`
Reviewed By: riversand963
Differential Revision: D21670446
Pulled By: ltamasi
fbshipit-source-id: 2eac249945cf33d8fb8597b26bfff5221e1a861a
Summary:
1. Add a value_size in read options which limits the cumulative value size of keys read in batches. Once the size exceeds read_options.value_size, all the remaining keys are returned with status Abort without further fetching any key.
2. Add a unit test case MultiGetBatchedValueSizeSimple the reads keys from memory and sst files.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6826
Test Plan:
1. make check -j64
2. Add a new unit test case
Reviewed By: anand1976
Differential Revision: D21471483
Pulled By: akankshamahajan15
fbshipit-source-id: dea51b8e76d5d1df38ece8cdb29933b1d798b900
Summary:
When using ldb, users cannot turn on force consistency check in most commands, while they cannot use checksonsistnecy with --try_load_options. The change fixes both by:
1. checkconsistency now calls OpenDB() so that it gets all the options loading and sanitized options logic
2. use options.check_consistency_checks = true by default, and add a --disable_consistency_checks to turn it off.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6802
Test Plan: Add a new unit test. Some manual tests with corrupted DBs.
Reviewed By: pdillinger
Differential Revision: D21388051
fbshipit-source-id: 8d122732d391b426e3982a1c3232a8e3763ffad0
Summary:
Delete triggered compaction in universal compaction mode was causing a corruption when scheduled in parallel with other compactions.
1. When num_levels = 1, a file marked for compaction may be picked along with all older files in L0, without checking if any of them are already being compaction. This can cause unpredictable results like resurrection of older versions of keys or deleted keys.
2. When num_levels > 1, a delete triggered compaction would not get scheduled if it overlaps with a running regular compaction. However, the reverse is not true. This is due to the fact that in ```UniversalCompactionBuilder::CalculateSortedRuns```, it assumes that entire sorted runs are picked for compaction and only checks the first file in a sorted run to determine conflicts. This is violated by a delete triggered compaction as it works on a subset of a sorted run.
Fix the bug for num_levels > 1, and disable the feature for now when num_levels = 1. After disabling this feature, files would still get marked for compaction, but no compaction would get scheduled.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6799
Reviewed By: pdillinger
Differential Revision: D21431286
Pulled By: anand1976
fbshipit-source-id: ae9f0bdb1d6ae2f10284847db731c23f43af164a
Summary:
With consistency check enabled, VersionBuilder::SaveTo() may return error once
corruption is detected while building versions. We should handle these errors.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6801
Test Plan: make check
Reviewed By: siying
Differential Revision: D21385045
Pulled By: riversand963
fbshipit-source-id: 98f6424e2a4699b62befa21e9fe00e70a771118e
Summary:
The patch adds logic that returns the set of live blob files from
`Version::AddLiveFiles` and `VersionSet::AddLiveFiles` (in addition to
live table files), and also cleans up the code a bit, for example, by
exposing only the numbers of table files as opposed to the earlier
`FileDescriptor`s that no clients used. Moreover, the patch extends
the `GetLiveFiles` API so that it also exposes blob files in the current version.
Similarly to https://github.com/facebook/rocksdb/pull/6755,
this is a building block for identifying and purging obsolete blob files.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6785
Test Plan: `make check`
Reviewed By: riversand963
Differential Revision: D21336210
Pulled By: ltamasi
fbshipit-source-id: fc1aede8a49eacd03caafbc5f6f9ce43b6270821
Summary:
We are swallowing some file consistency checking failures. This is not expected. We are fixing two cases: DB reopen and manifest dump.
More places are not fixed and need follow-up.
Error from CheckConsistencyForDeletes() is also swallowed, which is not fixed in this PR.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6793
Test Plan: Add a unit test to cover the reopen case.
Reviewed By: riversand963
Differential Revision: D21366525
fbshipit-source-id: eb438a322237814e8d5125f916a3c6de97f39ded
Summary:
The patch adds logic to keep track of obsolete blob files. A blob file becomes
obsolete when the last `shared_ptr` that points to the corresponding
`SharedBlobFileMetaData` object goes away, which, in turn, happens when the
last `Version` that contains the blob file is destroyed. No longer needed blob
files are added to the obsolete list in `VersionSet` using a custom deleter to
avoid unnecessary coupling between `SharedBlobFileMetaData` and `VersionSet`.
Obsolete blob files are returned by `VersionSet::GetObsoleteFiles` and stored
in `JobContext`.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6755
Test Plan: `make check`
Reviewed By: riversand963
Differential Revision: D21233155
Pulled By: ltamasi
fbshipit-source-id: 47757e06fdc0127f27ed57f51abd27893d9a7b7a
Summary:
Context: Index type `kBinarySearchWithFirstKey` added the ability for sst file iterator to sometimes report a key from index without reading the corresponding data block. This is useful when sst blocks are cut at some meaningful boundaries (e.g. one block per key prefix), and many seeks land between blocks (e.g. for each prefix, the ranges of keys in different sst files are nearly disjoint, so a typical seek needs to read a data block from only one file even if all files have the prefix). But this added a new error condition, which rocksdb code was really not equipped to deal with: `InternalIterator::value()` may fail with an IO error or Status::Incomplete, but it's just a method returning a Slice, with no way to report error instead. Before this PR, this type of error wasn't handled at all (an empty slice was returned), and kBinarySearchWithFirstKey implementation was considered a prototype.
Now that we (LogDevice) have experimented with kBinarySearchWithFirstKey for a while and confirmed that it's really useful, this PR is adding the missing error handling.
It's a pretty inconvenient situation implementation-wise. The error needs to be reported from InternalIterator when trying to access value. But there are ~700 call sites of `InternalIterator::value()`, most of which either can't hit the error condition (because the iterator is reading from memtable or from index or something) or wouldn't benefit from the deferred loading of the value (e.g. compaction iterator that reads all values anyway). Adding error handling to all these call sites would needlessly bloat the code. So instead I made the deferred value loading optional: only the call sites that may use deferred loading have to call the new method `PrepareValue()` before calling `value()`. The feature is enabled with a new bool argument `allow_unprepared_value` to a bunch of methods that create iterators (it wouldn't make sense to put it in ReadOptions because it's completely internal to iterators, with virtually no user-visible effect). Lmk if you have better ideas.
Note that the deferred value loading only happens for *internal* iterators. The user-visible iterator (DBIter) always prepares the value before returning from Seek/Next/etc. We could go further and add an API to defer that value loading too, but that's most likely not useful for LogDevice, so it doesn't seem worth the complexity for now.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6621
Test Plan: make -j5 check . Will also deploy to some logdevice test clusters and look at stats.
Reviewed By: siying
Differential Revision: D20786930
Pulled By: al13n321
fbshipit-source-id: 6da77d918bad3780522e918f17f4d5513d3e99ee
Summary:
Towards making compaction logic compatible with user timestamp.
When computing boundaries and overlapping ranges for inputs of compaction, We need to compare SSTs by user key without timestamp.
Test plan (devserver):
```
make check
```
Several individual tests:
```
./version_set_test --gtest_filter=VersionStorageInfoTimestampTest.GetOverlappingInputs
./db_with_timestamp_compaction_test
./db_with_timestamp_basic_test
```
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6645
Reviewed By: ltamasi
Differential Revision: D20960012
Pulled By: riversand963
fbshipit-source-id: ad377fa9eb481bf7a8a3e1824aaade48cdc653a4
Summary:
Does what it says on the can. Similarly to table files, we need to re-persist
the metadata of live blob files whenever a new manifest file is opened.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6630
Test Plan: `make check`
Reviewed By: riversand963
Differential Revision: D20802126
Pulled By: ltamasi
fbshipit-source-id: 5738692d898790293bf09d66e9997369bbf89566
Summary:
In the current code base, we use Status to get and store the returned status from the call. Specifically, for IO related functions, the current Status cannot reflect the IO Error details such as error scope, error retryable attribute, and others. With the implementation of https://github.com/facebook/rocksdb/issues/5761, we have the new Wrapper for IO, which returns IOStatus instead of Status. However, the IOStatus is purged at the lower level of write path and transferred to Status.
The first job of this PR is to pass the IOStatus to the write path (flush, WAL write, and Compaction). The second job is to identify the Retryable IO Error as HardError, and set the bg_error_ as HardError. In this case, the DB Instance becomes read only. User is informed of the Status and need to take actions to deal with it (e.g., call db->Resume()).
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6487
Test Plan: Added the testing case to error_handler_fs_test. Pass make asan_check
Reviewed By: anand1976
Differential Revision: D20685017
Pulled By: zhichao-cao
fbshipit-source-id: ff85f042896243abcd6ef37877834e26f36b6eb0
Summary:
The patch adds a couple of classes to represent metadata about
blob files: `SharedBlobFileMetaData` contains the information elements
that are immutable (once the blob file is closed), e.g. blob file number,
total number and size of blob files, checksum method/value, while
`BlobFileMetaData` contains attributes that can vary across versions like
the amount of garbage in the file. There is a single `SharedBlobFileMetaData`
for each blob file, which is jointly owned by the `BlobFileMetaData` objects
that point to it; `BlobFileMetaData` objects, in turn, are owned by `Version`s
and can also be shared if the (immutable _and_ mutable) state of the blob file
is the same in two versions.
In addition, the patch adds the blob file metadata to `VersionStorageInfo`, and extends
`VersionBuilder` so that it can apply blob file related `VersionEdit`s (i.e. those
containing `BlobFileAddition`s and/or `BlobFileGarbage`), and save blob file metadata
to a new `VersionStorageInfo`. Consistency checks are also extended to ensure
that table files point to blob files that are part of the `Version`, and that all blob files
that are part of any given `Version` have at least some _non_-garbage data in them.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6597
Test Plan: `make check`
Reviewed By: riversand963
Differential Revision: D20656803
Pulled By: ltamasi
fbshipit-source-id: f1f74d135045b3b42d0146f03ee576ef0a4bfd80
Summary:
Add timestamp support for MultiGet().
timestamp from readoptions is honored, and timestamps can be returned along with values.
MultiReadRandom perf test (10 minutes) on the same development machine ram drive with the same DB data shows no regression (within marge of error). The test is adapted from https://github.com/facebook/rocksdb/wiki/RocksDB-In-Memory-Workload-Performance-Benchmarks.
base line (commit 17bef7d3a):
multireadrandom : 104.173 micros/op 307167 ops/sec; (5462999 of 5462999 found)
This PR:
multireadrandom : 104.199 micros/op 307095 ops/sec; (5307999 of 5307999 found)
.\db_bench --db=r:\rocksdb.github --num_levels=6 --key_size=20 --prefix_size=20 --keys_per_prefix=0 --value_size=100 --cache_size=2147483648 --cache_numshardbits=6 --compression_type=none --compression_ratio=1 --min_level_to_compress=-1 --disable_seek_compaction=1 --hard_rate_limit=2 --write_buffer_size=134217728 --max_write_buffer_number=2 --level0_file_num_compaction_trigger=8 --target_file_size_base=134217728 --max_bytes_for_level_base=1073741824 --disable_wal=0 --wal_dir=r:\rocksdb.github\WAL_LOG --sync=0 --verify_checksum=1 --statistics=0 --stats_per_interval=0 --stats_interval=1048576 --histogram=0 --use_plain_table=1 --open_files=-1 --memtablerep=prefix_hash --bloom_bits=10 --bloom_locality=1 --duration=600 --benchmarks=multireadrandom --use_existing_db=1 --num=25000000 --threads=32 --allow_concurrent_memtable_write=0
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6483
Reviewed By: anand1976
Differential Revision: D20498373
Pulled By: riversand963
fbshipit-source-id: 8505f22bc40fd791bc7dd05e48d7e67c91edb627
Summary:
When applying a new version in non DB open case, optimize_filters_for_hits is used for max_threads, which is clearly a bug. It is not clear what the indented value in the first place, but it value 1 makes sense here, which would create no extra threads. This bug is not expected to cause user visible problems, assuming C++ implicitly cast bool to 0 or 1.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6576
Test Plan: Run all exsiting test.
Reviewed By: ajkr
Differential Revision: D20602467
fbshipit-source-id: 40b2cd8619aba09ae9242b36c415464db3c9b737
Summary:
The current Env/FileSystem API separation has a couple of issues -
1. It requires the user to specify 2 options - ```Options::env``` and ```Options::file_system``` - which means they have to make code changes to benefit from the new APIs. Furthermore, there is a risk of accessing the same APIs in two different ways, through Env in the old way and through FileSystem in the new way. The two may not always match, for example, if env is ```PosixEnv``` and FileSystem is a custom implementation. Any stray RocksDB calls to env will use the ```PosixEnv``` implementation rather than the file_system implementation.
2. There needs to be a simple way for the FileSystem developer to instantiate an Env for backward compatibility purposes.
This PR solves the above issues and simplifies the migration in the following ways -
1. Embed a shared_ptr to the ```FileSystem``` in the ```Env```, and remove ```Options::file_system``` as a configurable option. This way, no code changes will be required in application code to benefit from the new API. The default Env constructor uses a ```LegacyFileSystemWrapper``` as the embedded ```FileSystem```.
1a. - This also makes it more robust by ensuring that even if RocksDB
has some stray calls to Env APIs rather than FileSystem, they will go
through the same object and thus there is no risk of getting out of
sync.
2. Provide a ```NewCompositeEnv()``` API that can be used to construct a
PosixEnv with a custom FileSystem implementation. This eliminates an
indirection to call Env APIs, and relieves the FileSystem developer of
the burden of having to implement wrappers for the Env APIs.
3. Add a couple of missing FileSystem APIs - ```SanitizeEnvOptions()``` and
```NewLogger()```
Tests:
1. New unit tests
2. make check and make asan_check
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6552
Reviewed By: riversand963
Differential Revision: D20592038
Pulled By: anand1976
fbshipit-source-id: c3801ad4153f96d21d5a3ae26c92ba454d1bf1f7
Summary:
There are situations when RocksDB tries to recover, but the db is in an inconsistent state due to SST files referenced in the MANIFEST being missing. In this case, previous RocksDB will just fail the recovery and return a non-ok status.
This PR enables another possibility. During recovery, RocksDB checks possible MANIFEST files, and try to recover to the most recent state without missing table file. `VersionSet::Recover()` applies version edits incrementally and "materializes" a version only when this version does not reference any missing table file. After processing the entire MANIFEST, the version created last will be the latest version.
`DBImpl::Recover()` calls `VersionSet::Recover()`. Afterwards, WAL replay will *not* be performed.
To use this capability, set `options.best_efforts_recovery = true` when opening the db. Best-efforts recovery is currently incompatible with atomic flush.
Test plan (on devserver):
```
$make check
$COMPILE_WITH_ASAN=1 make all && make check
```
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6334
Reviewed By: anand1976
Differential Revision: D19778960
Pulled By: riversand963
fbshipit-source-id: c27ea80f29bc952e7d3311ecf5ee9c54393b40a8
Summary:
Each time RocksDB switches to a new MANIFEST file from old one, it calls WriteCurrentStateToManifest() which writes a 'snapshot' of the current in-memory state of versions to the beginning of the new manifest as a bunch of version edits. We can distinguish these version edits from other version edits written during normal operations with a custom, safe-to-ignore tag.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6530
Test Plan: added test to version_edit_test, pass make asan_check
Reviewed By: riversand963
Differential Revision: D20524516
Pulled By: zhichao-cao
fbshipit-source-id: f1de102f5499bfa88dae3caa2f32c7f42cf904db
Summary:
Preliminary support for iterator with user timestamp. Current implementation does not consider merge operator and reverse iterator. Auto compaction is also disabled in unit tests.
Create an iterator with timestamp.
```
...
read_opts.timestamp = &ts;
auto* iter = db->NewIterator(read_opts);
// target is key without timestamp.
for (iter->Seek(target); iter->Valid(); iter->Next()) {}
for (iter->SeekToFirst(); iter->Valid(); iter->Next()) {}
delete iter;
read_opts.timestamp = &ts1;
// lower_bound and upper_bound are without timestamp.
read_opts.iterate_lower_bound = &lower_bound;
read_opts.iterate_upper_bound = &upper_bound;
auto* iter1 = db->NewIterator(read_opts);
// Do Seek or SeekToFirst()
delete iter1;
```
Test plan (dev server)
```
$make check
```
Simple benchmarking (dev server)
1. The overhead introduced by this PR even when timestamp is disabled.
key size: 16 bytes
value size: 100 bytes
Entries: 1000000
Data reside in main memory, and try to stress iterator.
Repeated three times on master and this PR.
- Seek without next
```
./db_bench -db=/dev/shm/rocksdbtest-1000 -benchmarks=fillseq,seekrandom -enable_pipelined_write=false -disable_wal=true -format_version=3
```
master: 159047.0 ops/sec
this PR: 158922.3 ops/sec (2% drop in throughput)
- Seek and next 10 times
```
./db_bench -db=/dev/shm/rocksdbtest-1000 -benchmarks=fillseq,seekrandom -enable_pipelined_write=false -disable_wal=true -format_version=3 -seek_nexts=10
```
master: 109539.3 ops/sec
this PR: 107519.7 ops/sec (2% drop in throughput)
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6255
Differential Revision: D19438227
Pulled By: riversand963
fbshipit-source-id: b66b4979486f8474619f4aa6bdd88598870b0746
Summary:
In the current code base, we can use Directory from Env to manage directory (e.g, Fsync()). The PR https://github.com/facebook/rocksdb/issues/5761 introduce the File System as a new Env API. So we further replace the Directory class in DB with FSDirectory such that we can have more IO information from IOStatus returned by FSDirectory.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6468
Test Plan: pass make asan_check
Differential Revision: D20195261
Pulled By: zhichao-cao
fbshipit-source-id: 93962cb9436852bfcfb76e086d9e7babd461cbe1
Summary:
Added new Get() methods that return timestamp. Dummy implementation is given so that classes derived from DB don't need to be touched to provide their implementation. MultiGet is not included.
ReadRandom perf test (10 minutes) on the same development machine ram drive with the same DB data shows no regression (within marge of error). The test is adapted from https://github.com/facebook/rocksdb/wiki/RocksDB-In-Memory-Workload-Performance-Benchmarks.
base line (commit 72ee067b9):
101.712 micros/op 314602 ops/sec; 36.0 MB/s (5658999 of 5658999 found)
This PR:
100.288 micros/op 319071 ops/sec; 36.5 MB/s (5674999 of 5674999 found)
./db_bench --db=r:\rocksdb.github --num_levels=6 --key_size=20 --prefix_size=20 --keys_per_prefix=0 --value_size=100 --cache_size=2147483648 --cache_numshardbits=6 --compression_type=none --compression_ratio=1 --min_level_to_compress=-1 --disable_seek_compaction=1 --hard_rate_limit=2 --write_buffer_size=134217728 --max_write_buffer_number=2 --level0_file_num_compaction_trigger=8 --target_file_size_base=134217728 --max_bytes_for_level_base=1073741824 --disable_wal=0 --wal_dir=r:\rocksdb.github\WAL_LOG --sync=0 --verify_checksum=1 --delete_obsolete_files_period_micros=314572800 --max_background_compactions=4 --max_background_flushes=0 --level0_slowdown_writes_trigger=16 --level0_stop_writes_trigger=24 --statistics=0 --stats_per_interval=0 --stats_interval=1048576 --histogram=0 --use_plain_table=1 --open_files=-1 --mmap_read=1 --mmap_write=0 --memtablerep=prefix_hash --bloom_bits=10 --bloom_locality=1 --duration=600 --benchmarks=readrandom --use_existing_db=1 --num=25000000 --threads=32
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6409
Differential Revision: D20200086
Pulled By: riversand963
fbshipit-source-id: 490edd74d924f62bd8ae9c29c2a6bbbb8410ca50
Summary:
Currently, a new MANIFEST file is assigned a new file number when 1) no
MANIFEST is open, or 2) current MANIFEST file size exceeds a threshold. This is
not sufficient. There are cases when the caller explicitly specifies that a new
MANIFEST be created. For example, if user sets options.write_dbid_to_manifest = true,
and there are WAL files, then RocksDB will run into an issue during recovery.
`DBImpl::Recover()` will call `LogAndApply()` to write dbid. At this point, the db being
recovered creates a new MANIFEST, say, MANIFEST-000003. Since there are WALs,
`DBImpl::RecoverLogFiles` will be called. Towards the end of this function, we call
`LogAndApply(new_descriptor_log=true)`, which explicitly creates a new MANIFEST.
However, the manifest_file_number is wrong before this fix. Consequently, RocksDB
opens an existing, non-empty file for append, effectively truncating the file to zero.
If a crash occurs, then there will be data loss.
Test Plan (devserver):
make check
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6426
Test Plan: make check
Differential Revision: D19951866
Pulled By: riversand963
fbshipit-source-id: 4b1b9fc28d4fe2ac12764b388ef9e61f05e766da
Summary:
When dynamically linking two binaries together, different builds of RocksDB from two sources might cause errors. To provide a tool for user to solve the problem, the RocksDB namespace is changed to a flag which can be overridden in build time.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6433
Test Plan: Build release, all and jtest. Try to build with ROCKSDB_NAMESPACE with another flag.
Differential Revision: D19977691
fbshipit-source-id: aa7f2d0972e1c31d75339ac48478f34f6cfcfb3e
Summary:
Previously, when recovering version set, LoadTableHandlers failures are ignored.
If paranoid_checks is true, this failure should not be ignored, otherwise, the opened db might be in an inconsistent state.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6368
Test Plan: make check
Differential Revision: D19713459
Pulled By: cheng-chang
fbshipit-source-id: 68cb94f4f2cc43f8b024b14755193cd45cfcad55
Summary:
Seems like this caused the following test failure on AppVeyor:
DBTest2.CrashInRecoveryMultipleCF
c:\projects\rocksdb\db\db_test_util.cc(107): error: DestroyDB(dbname_, options)
IO error: Failed to delete: C:\projects\rocksdb\db_tests\\testrocksdb-3112//db_test2_10791409581227174103/000013.sst: Access is denied.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6410
Test Plan: Wait to see whether the AppVeyor test passes.
Differential Revision: D19879872
Pulled By: cheng-chang
fbshipit-source-id: 59a9c55ca88566e9210c0b715ecc45a4fd9afe26
Summary:
Unrevert the previous fix to propagate error status, and an additional fix to not treat a memtable lookup MergeInProgress status as an error.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6403
Test Plan:
Unit tests
Tried running stress tests but couldn't repro the stress failure
Differential Revision: D19846721
Pulled By: anand1976
fbshipit-source-id: 7db10cccbdc863d9b559497f0a46b608d2488ca4
Summary:
This reverts commit d70011bccc. The commit is causing some stress test failure due to unexpected Status::MergeInProgress() return for some keys.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6401
Differential Revision: D19826623
Pulled By: anand1976
fbshipit-source-id: edd634cede9cb7bdd2cb8f46e662ea709b16d2f1
Summary:
Add a utility class `Defer` to defer the execution of a function until the Defer object goes out of scope.
Used in VersionSet:: ProcessManifestWrites as an example.
The inline comments for class `Defer` have more details.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6382
Test Plan: `make defer_test version_set_test && ./defer_test && ./version_set_test`
Differential Revision: D19797538
Pulled By: cheng-chang
fbshipit-source-id: b1a9b7306e4fd4f48ec2ab55783caa561a315f0f
Summary:
https://github.com/facebook/rocksdb/pull/6383 surfaced an issue with
`VersionSet`/`ReactiveVersionSet` and `AtomicGroupReadBuffer::AddEdit`
(which was added in https://github.com/facebook/rocksdb/pull/5411):
`AddEdit` moves the `VersionEdit` passed to it into `replay_buffer_`,
however, the client `VersionSet` classes keep using it afterwards. This
*seemed to* work before the refactoring but it really did not: since
`VersionEdit` used to have a user-declared destructor, no move
constructor/move assignment operator was generated, and the `move` in
`AddEdit` was really a copy. The patch makes the copy explicit. Note: it
should be possible to rework this logic so that we can get away
with the move but for now, this should fix the issue.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6400
Test Plan:
`make check`
`make analyze`
Differential Revision: D19824466
Pulled By: ltamasi
fbshipit-source-id: f38033967daf2a39c78dcd6e12978bafe37632b4
Andrew Kryczka