Summary:
This PR resolves https://github.com/facebook/rocksdb/issues/10487 & https://github.com/facebook/rocksdb/issues/10536, user code needs to call Refresh() periodically.
The main code change is to support range deletions. A range tombstone iterator uses a sequence number as upper bound to decide which range tombstones are effective. During Iterator refresh, this sequence number upper bound needs to be updated for all range tombstone iterators under DBIter and LevelIterator. LevelIterator may create new table iterators and range tombstone iterator during scanning, so it needs to be aware of iterator refresh. The code path that propagates this change is `db_iter_->set_sequence(read_seq) -> MergingIterator::SetRangeDelReadSeqno() -> TruncatedRangeDelIterator::SetRangeDelReadSeqno() and LevelIterator::SetRangeDelReadSeqno()`.
This change also fixes an issue where range tombstone iterators created by LevelIterator may access ReadOptions::snapshot, even though we do not explicitly require users to keep a snapshot alive after creating an Iterator.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10594
Test Plan:
* New unit tests.
* Add Iterator::Refresh(snapshot) to stress test. Note that this change only adds tests for refreshing to the same snapshot since this is the main target use case.
TODO in a following PR:
* Stress test Iterator::Refresh() to different snapshots or no snapshot.
Reviewed By: ajkr
Differential Revision: D48456896
Pulled By: cbi42
fbshipit-source-id: 2e642c04e91235cc9542ef4cd37b3c20823bd779
Summary:
Update the logic in FilePrefetchBuffer to update `upper_bound_offset_` during reseek. During Reseek, `iterate_upper_bound` can be changed dynamically. So added an API to update that in FilePrefetchBuffer.
Added unit test to confirm the behavior.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11775
Test Plan:
- Check stress tests in case there is any failure after this diff.
- make crash_test -j32 with auto_readahead_size=1 passed locally
Reviewed By: anand1976
Differential Revision: D48815177
Pulled By: akankshamahajan15
fbshipit-source-id: 5f44fbb3af06c86a1c38f139c5fa4543891837f4
Summary:
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11823
Similarly to https://github.com/facebook/rocksdb/pull/11813, the patch is a small refactoring that eliminates some copy-paste around sorting the columns of entities by column name.
Reviewed By: jaykorean
Differential Revision: D49195504
fbshipit-source-id: d48c9f290e3203f838cc5949856c469ecf730008
Summary:
This should only affect iterator when
- user uses DeleteRange(),
- An iterator from level L has a non-ok status (such non-ok status may not be caught before the bug fix in https://github.com/facebook/rocksdb/pull/11783), and
- A range tombstone covers a key from level > L and triggers a reseek sets the status_ to OK in SeekImpl()/SeekPrevImpl() e.g. bd6a8340c3/table/merging_iterator.cc (L801)
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11786
Differential Revision: D48908830
Pulled By: cbi42
fbshipit-source-id: eb564be375af4e33dc27542eff753260186e6d5d
Summary:
This happens in (Compaction)MergingIterator layer, and can cause data loss during compaction or read/scan return incorrect result
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11782
Reviewed By: ajkr
Differential Revision: D48880575
Pulled By: cbi42
fbshipit-source-id: 2294ad284a6d653d3674bebe55380f12ee4b645b
Summary:
For a SST file that uses user-defined timestamp aware comparators, if a lower or upper bound is set, sst_dump tool doesn't handle it well. This PR adds support for that. While working on this `MaybeAddTimestampsToRange` is moved to the udt_util.h file to be shared.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11757
Test Plan:
make all check
for changes in db_impl.cc and db_impl_compaction_flush.cc
for changes in sst_file_dumper.cc, I manually tested this change handles specifying bounds for UDT use cases. It probably should have a unit test file eventually.
Reviewed By: ltamasi
Differential Revision: D48668048
Pulled By: jowlyzhang
fbshipit-source-id: 1560465f40e44668d6d82a7439fe9012be0e74a8
Summary:
wide_columns can now be pretty-printed in the following commands
- `./ldb dump_wal`
- `./ldb dump`
- `./ldb idump`
- `./ldb dump_live_files`
- `./ldb scan`
- `./sst_dump --command=scan`
There are opportunities to refactor to reduce some nearly identical code. This PR is initial change to add wide column support in `ldb` and `sst_dump` tool. More PRs to come for the refactor.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11754
Test Plan:
**New Tests added**
- `WideColumnsHelperTest::DumpWideColumns`
- `WideColumnsHelperTest::DumpSliceAsWideColumns`
**Changes added to existing tests**
- `ExternalSSTFileTest::BasicMixed` added to cover mixed case (This test should have been added in https://github.com/facebook/rocksdb/issues/11688). This test does not verify the ldb or sst_dump output. This test was used to create test SST files having some rows with wide columns and some without and the generated SST files were used to manually test sst_dump_tool.
- `createSST()` in `sst_dump_test` now takes `wide_column_one_in` to add wide column value in SST
**dump_wal**
```
./ldb dump_wal --walfile=/tmp/rocksdbtest-226125/db_wide_basic_test_2675429_2308393776696827948/000004.log --print_value --header
```
```
Sequence,Count,ByteSize,Physical Offset,Key(s) : value
1,1,59,0,PUT_ENTITY(0) : 0x:0x68656C6C6F 0x617474725F6E616D6531:0x666F6F 0x617474725F6E616D6532:0x626172
2,1,34,42,PUT_ENTITY(0) : 0x617474725F6F6E65:0x74776F 0x617474725F7468726565:0x666F7572
3,1,17,7d,PUT(0) : 0x7468697264 : 0x62617A
```
**idump**
```
./ldb --db=/tmp/rocksdbtest-226125/db_wide_basic_test_3481961_2308393776696827948/ idump
```
```
'first' seq:1, type:22 => :hello attr_name1:foo attr_name2:bar
'second' seq:2, type:22 => attr_one:two attr_three:four
'third' seq:3, type:1 => baz
Internal keys in range: 3
```
**SST Dump from dump_live_files**
```
./ldb --db=/tmp/rocksdbtest-226125/db_wide_basic_test_3481961_2308393776696827948/ compact
./ldb --db=/tmp/rocksdbtest-226125/db_wide_basic_test_3481961_2308393776696827948/ dump_live_files
```
```
...
==============================
SST Files
==============================
/tmp/rocksdbtest-226125/db_wide_basic_test_3481961_2308393776696827948/000013.sst level:1
------------------------------
Process /tmp/rocksdbtest-226125/db_wide_basic_test_3481961_2308393776696827948/000013.sst
Sst file format: block-based
'first' seq:0, type:22 => :hello attr_name1:foo attr_name2:bar
'second' seq:0, type:22 => attr_one:two attr_three:four
'third' seq:0, type:1 => baz
...
```
**dump**
```
./ldb --db=/tmp/rocksdbtest-226125/db_wide_basic_test_3481961_2308393776696827948/ dump
```
```
first ==> :hello attr_name1:foo attr_name2:bar
second ==> attr_one:two attr_three:four
third ==> baz
Keys in range: 3
```
**scan**
```
./ldb --db=/tmp/rocksdbtest-226125/db_wide_basic_test_3481961_2308393776696827948/ scan
```
```
first : :hello attr_name1:foo attr_name2:bar
second : attr_one:two attr_three:four
third : baz
```
**sst_dump**
```
./sst_dump --file=/tmp/rocksdbtest-226125/db_wide_basic_test_3481961_2308393776696827948/000013.sst --command=scan
```
```
options.env is 0x7ff54b296000
Process /tmp/rocksdbtest-226125/db_wide_basic_test_3481961_2308393776696827948/000013.sst
Sst file format: block-based
from [] to []
'first' seq:0, type:22 => :hello attr_name1:foo attr_name2:bar
'second' seq:0, type:22 => attr_one:two attr_three:four
'third' seq:0, type:1 => baz
```
Reviewed By: ltamasi
Differential Revision: D48837999
Pulled By: jaykorean
fbshipit-source-id: b0280f0589d2b9716bb9b50530ffcabb397d140f
Summary:
Fix seg fault in auto_readahead_size
```
db_stress:
internal_repo_rocksdb/repo/table/block_based/partitioned_index_iterator.h:70: virtual rocksdb::IndexValue rocksdb::PartitionedIndexIterator::value() const: Assertion `Valid()' failed.
```
During seek, after calculating readahead_size, db_stress can inject IOError resulting in failure to index_iter_->Seek and making index_iter_ invalid.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11761
Test Plan: Reproducible locally and passed with this fix
Reviewed By: anand1976
Differential Revision: D48696248
Pulled By: akankshamahajan15
fbshipit-source-id: 2be43bf56ad0fc2f95f9093c19c9a1b15a716091
Summary:
Implement trimming of readahead_size under a new option ReadOptions.auto_readahead_size. It'll trim the readahead_size during prefetching upto iterate_upper_bound offset only when ReadOptions.iterate_upper_bound is set, therefore reducing the prefetching of data beyond upper_bound.
It's enabled for both implicit auto readahead size and when ReadOptions.readahead_size is specified and for sync and async_io.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11684
Test Plan: Added new unit test
Reviewed By: anand1976
Differential Revision: D48479723
Pulled By: akankshamahajan15
fbshipit-source-id: 2b1703579caf779105e836b580866ffd7db076fc
Summary:
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11714
Fixes T161017540.
The staging build starts failing with an undefined symbol error:
```
ld.lld: error: undefined symbol: std::enable_if<rocksdb::ParsedFullFilterBlock::kCacheEntryRole == (rocksdb::CacheEntryRole)13 || true, rocksdb::Status>::type rocksdb::BlockBasedTable::MaybeReadBlockAndLoadToCache<rocksdb::ParsedFullFilterBlock>(rocksdb::FilePrefetchBuffer*, rocksdb::ReadOptions const&, rocksdb::BlockHandle const&, rocksdb::UncompressionDict const&, bool, rocksdb::CachableEntry<rocksdb::ParsedFullFilterBlock>*, rocksdb::GetContext*, rocksdb::BlockCacheLookupContext*, rocksdb::BlockContents*, bool) const
```
This is the `MaybeReadBlockAndLoadToCache` function where `TBlocklike = ParsedFullFilterBlock`. The trigger was an FDO profile update D48261413.
`MaybeReadBlockAndLoadToCache` is used in the same translation unit `block_based_table_reader.cc`, and also in another file `partitioned_filter_block.cc`. The later was the file that couldn't find the symbol. It seems after the FDO profile update, `MaybeReadBlockAndLoadToCache` may've got inlined into its caller in `block_based_table_reader.cc`. And with no knowledge of other usages, the symbol got stripped.
Explicitly instantiate the template similar to how `RetrieveBlock` was handled.
Reviewed By: pdillinger, akankshamahajan15
Differential Revision: D48400574
fbshipit-source-id: d4a80999bfb6ce4afa80678444139fcd8ae84aa4
Summary:
RocksDB provides APIs that enable creating SST files offline and then bulk loading them into the LSM tree quickly using metadata operations. Namely, clients can use the `SstFileWriter` class for the offline data preparation and then the IngestExternalFile family of APIs to perform the bulk loading. However, `SstFileWriter` currently does not support creating files with wide-column data in them. This PR adds `PutEntity` API implementation to `SstFileWriter` to support creating files with wide-column data.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11688
Test Plan: - `BasicWideColumn` test added in external_sst_file_test
Reviewed By: ltamasi
Differential Revision: D48243779
Pulled By: jaykorean
fbshipit-source-id: 1697e5bd67121a648c03946f867416a94be0cadf
Summary:
**Context/Summary:**
- Similar to https://github.com/facebook/rocksdb/pull/11288 but for user read such as `Get(), MultiGet(), DBIterator::XXX(), Verify(File)Checksum()`.
- For this, I refactored some user-facing `MultiGet` calls in `TransactionBase` and various types of `DB` so that it does not call a user-facing `Get()` but `GetImpl()` for passing the `ReadOptions::io_activity` check (see PR conversation)
- New user read stats breakdown are guarded by `kExceptDetailedTimers` since measurement shows they have 4-5% regression to the upstream/main.
- Misc
- More refactoring: with https://github.com/facebook/rocksdb/pull/11288, we complete passing `ReadOptions/IOOptions` to FS level. So we can now replace the previously [added](https://github.com/facebook/rocksdb/pull/9424) `rate_limiter_priority` parameter in `RandomAccessFileReader`'s `Read/MultiRead/Prefetch()` with `IOOptions::rate_limiter_priority`
- Also, `ReadAsync()` call time is measured in `SST_READ_MICRO` now
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11444
Test Plan:
- CI fake db crash/stress test
- Microbenchmarking
**Build** `make clean && ROCKSDB_NO_FBCODE=1 DEBUG_LEVEL=0 make -jN db_basic_bench`
- google benchmark version: 604f6fd3f4
- db_basic_bench_base: upstream
- db_basic_bench_pr: db_basic_bench_base + this PR
- asyncread_db_basic_bench_base: upstream + [db basic bench patch for IteratorNext](https://github.com/facebook/rocksdb/compare/main...hx235:rocksdb:micro_bench_async_read)
- asyncread_db_basic_bench_pr: asyncread_db_basic_bench_base + this PR
**Test**
Get
```
TEST_TMPDIR=/dev/shm ./db_basic_bench_{null_stat|base|pr} --benchmark_filter=DBGet/comp_style:0/max_data:134217728/per_key_size:256/enable_statistics:1/negative_query:0/enable_filter:0/mmap:1/threads:1 --benchmark_repetitions=1000
```
Result
```
Coming soon
```
AsyncRead
```
TEST_TMPDIR=/dev/shm ./asyncread_db_basic_bench_{base|pr} --benchmark_filter=IteratorNext/comp_style:0/max_data:134217728/per_key_size:256/enable_statistics:1/async_io:1/include_detailed_timers:0 --benchmark_repetitions=1000 > syncread_db_basic_bench_{base|pr}.out
```
Result
```
Base:
1956,1956,1968,1977,1979,1986,1988,1988,1988,1990,1991,1991,1993,1993,1993,1993,1994,1996,1997,1997,1997,1998,1999,2001,2001,2002,2004,2007,2007,2008,
PR (2.3% regression, due to measuring `SST_READ_MICRO` that wasn't measured before):
1993,2014,2016,2022,2024,2027,2027,2028,2028,2030,2031,2031,2032,2032,2038,2039,2042,2044,2044,2047,2047,2047,2048,2049,2050,2052,2052,2052,2053,2053,
```
Reviewed By: ajkr
Differential Revision: D45918925
Pulled By: hx235
fbshipit-source-id: 58a54560d9ebeb3a59b6d807639692614dad058a
Summary:
As titled, and also removed an undefined and unused member function in for ColumnFamilyData
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11683
Reviewed By: ajkr
Differential Revision: D48156290
Pulled By: jowlyzhang
fbshipit-source-id: cc99aaafe69db6611af3854cb2b2ebc5044941f7
Summary:
**Context/Summary:**
After https://github.com/facebook/rocksdb/pull/11631, file hint is not longer needed for compaction read. Therefore we can deprecate `Options::access_hint_on_compaction_start`. As this is a public API change, we should first mark the relevant APIs (including the Java's) deprecated and remove it in next major release 9.0.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11658
Test Plan: No code change
Reviewed By: ajkr
Differential Revision: D47997856
Pulled By: hx235
fbshipit-source-id: 16e015ae7728c224b1caef73143aa9915668f4ac
Summary:
## Context checksum
All RocksDB checksums currently use 32 bits of checking
power, which should be 1 in 4 billion false negative (FN) probability (failing to
detect corruption). This is true for random corruptions, and in some cases
small corruptions are guaranteed to be detected. But some possible
corruptions, such as in storage metadata rather than storage payload data,
would have a much higher FN rate. For example:
* Data larger than one SST block is replaced by data from elsewhere in
the same or another SST file. Especially with block_align=true, the
probability of exact block size match is probably around 1 in 100, making
the FN probability around that same. Without `block_align=true` the
probability of same block start location is probably around 1 in 10,000,
for FN probability around 1 in a million.
To solve this problem in new format_version=6, we add "context awareness"
to block checksum checks. The stored and expected checksum value is
modified based on the block's position in the file and which file it is in. The
modifications are cleverly chosen so that, for example
* blocks within about 4GB of each other are guaranteed to use different context
* blocks that are offset by exactly some multiple of 4GiB are guaranteed to use
different context
* files generated by the same process are guaranteed to use different context
for the same offsets, until wrap-around after 2^32 - 1 files
Thus, with format_version=6, if a valid SST block and checksum is misplaced,
its checksum FN probability should be essentially ideal, 1 in 4B.
## Footer checksum
This change also adds checksum protection to the SST footer (with
format_version=6), for the first time without relying on whole file checksum.
To prevent a corruption of the format_version in the footer (e.g. 6 -> 5) to
defeat the footer checksum, we change much of the footer data format
including an "extended magic number" in format_version 6 that would be
interpreted as empty index and metaindex block handles in older footer
versions. We also change the encoding of handles to free up space for
other new data in footer.
## More detail: making space in footer
In order to keep footer the same size in format_version=6 (avoid change to IO
patterns), we have to free up some space for new data. We do this two ways:
* Metaindex block handle is encoded down to 4 bytes (from 10) by assuming
it immediately precedes the footer, and by assuming it is < 4GB.
* Index block handle is moved into metaindex. (I don't know why it was
in footer to begin with.)
## Performance
In case of small performance penalty, I've made a "pay as you go" optimization
to compensate: replace `MutableCFOptions` in BlockBasedTableBuilder::Rep
with the only field used in that structure after construction: `prefix_extractor`.
This makes the PR an overall performance improvement (results below).
Nevertheless I'm seeing essentially no difference going from fv=5 to fv=6,
even including that improvement for both. That's based on extreme case table
write performance testing, many files with many blocks. This is relatively
checksum intensive (small blocks) and salt generation intensive (small files).
```
(for I in `seq 1 100`; do TEST_TMPDIR=/dev/shm/dbbench2 ./db_bench -benchmarks=fillseq -memtablerep=vector -disable_wal=1 -allow_concurrent_memtable_write=false -num=3000000 -compaction_style=2 -fifo_compaction_max_table_files_size_mb=10000 -fifo_compaction_allow_compaction=0 -write_buffer_size=100000 -compression_type=none -block_size=1000; done) 2>&1 | grep micros/op | tee out
awk '{ tot += $5; n += 1; } END { print int(1.0 * tot / n) }' < out
```
Each value below is ops/s averaged over 100 runs, run simultaneously with competing
configuration for load fairness
Before -> after (both fv=5): 483530 -> 483673 (negligible)
Re-run 1: 480733 -> 485427 (1.0% faster)
Re-run 2: 483821 -> 484541 (0.1% faster)
Before (fv=5) -> after (fv=6): 482006 -> 485100 (0.6% faster)
Re-run 1: 482212 -> 485075 (0.6% faster)
Re-run 2: 483590 -> 484073 (0.1% faster)
After fv=5 -> after fv=6: 483878 -> 485542 (0.3% faster)
Re-run 1: 485331 -> 483385 (0.4% slower)
Re-run 2: 485283 -> 483435 (0.4% slower)
Re-run 3: 483647 -> 486109 (0.5% faster)
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9058
Test Plan:
unit tests included (table_test, db_properties_test, salt in env_test). General DB tests
and crash test updated to test new format_version.
Also temporarily updated the default format version to 6 and saw some test failures. Almost all
were due to an inadvertent additional read in VerifyChecksum to verify the index block checksum,
though it's arguably a bug that VerifyChecksum does not appear to (re-)verify the index block
checksum, just assuming it was verified in opening the index reader (probably *usually* true but
probably not always true). Some other concerns about VerifyChecksum are left in FIXME
comments. The only remaining test failure on change of default (in block_fetcher_test) now
has a comment about how to upgrade the test.
The format compatibility test does not need updating because we have not updated the default
format_version.
Reviewed By: ajkr, mrambacher
Differential Revision: D33100915
Pulled By: pdillinger
fbshipit-source-id: 8679e3e572fa580181a737fd6d113ed53c5422ee
Summary:
... to improve data integrity validation during compaction.
A new option `compaction_verify_record_count` is introduced for this verification and is enabled by default. One exception when the verification is not done is when a compaction filter returns kRemoveAndSkipUntil which can cause CompactionIterator to seek until some key and hence not able to keep track of the number of keys processed.
For expected number of input keys, we sum over the number of total keys - number of range tombstones across compaction input files (`CompactionJob::UpdateCompactionStats()`). Table properties are consulted if `FileMetaData` is not initialized for some input file. Since table properties for all input files were also constructed during `DBImpl::NotifyOnCompactionBegin()`, `Compaction::GetTableProperties()` is introduced to reduce duplicated code.
For actual number of keys processed, each subcompaction will record its number of keys processed to `sub_compact->compaction_job_stats.num_input_records` and aggregated when all subcompactions finish (`CompactionJob::AggregateCompactionStats()`). In the case when some subcompaction encountered kRemoveAndSkipUntil from compaction filter and does not have accurate count, it propagates this information through `sub_compact->compaction_job_stats.has_num_input_records`.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11571
Test Plan:
* Add a new unit test `DBCompactionTest.VerifyRecordCount` for the corruption case.
* All other unit tests for non-corrupted case.
* Ran crash test for a few hours: `python3 ./tools/db_crashtest.py whitebox --simple`
Reviewed By: ajkr
Differential Revision: D47131965
Pulled By: cbi42
fbshipit-source-id: cc8e94565dd526c4347e9d3843ecf32f6727af92
Summary:
**Context/Summary**
As titled. The benefit of doing so is to explicitly call readahead() instead of relying page cache behavior for compaction read when we know that we most likely need readahead as compaction read is sequential read .
**Test**
Extended the existing UT to cover compaction read case
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11631
Reviewed By: ajkr
Differential Revision: D47681437
Pulled By: hx235
fbshipit-source-id: 78792f64985c4dc44aa8f2a9c41ab3e8bbc0bc90
Summary:
`sst_dump --command=verify` did not set read_options.verify_checksum to true so it was not verifying checksum.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11576
Test Plan:
ran the same command on an SST file with bad checksum:
```
sst_dump --command=verify --file=...sst_file_with_bad_block_checksum
Before this PR:
options.env is 0x6ba048
Process ...sst_file_with_bad_block_checksum
Sst file format: block-based
The file is ok
After this PR:
options.env is 0x7f43f6690000
Process ...sst_file_with_bad_block_checksum
Sst file format: block-based
... is corrupted: Corruption: block checksum mismatch: stored = 2170109798, computed = 2170097510, type = 4 ...
```
Reviewed By: ajkr
Differential Revision: D47136284
Pulled By: cbi42
fbshipit-source-id: 07d68db715c00347145e5b83d649aef2c3f2acd9
Summary:
Logically strip the user-defined timestamp when L0 files are created during flush when `AdvancedColumnFamilyOptions.persist_user_defined_timestamps` is false. Logically stripping timestamp here means replacing the original user-defined timestamp with a mininum timestamp, which for now is hard coded to be all zeros bytes.
While working on this, I caught a missing piece on the `BlockBuilder` level for this feature. The current quick path `std::min(buffer_size, last_key_size)` needs a bit tweaking to work for this feature. When user-defined timestamp is stripped during block building, on writing first entry or right after resetting, `buffer` is empty and `buffer_size` is zero as usual. However, in follow-up writes, depending on the size of the stripped user-defined timestamp, and the size of the value, what's in `buffer` can sometimes be smaller than `last_key_size`, leading `std::min(buffer_size, last_key_size)` to truncate the `last_key`. Previous test doesn't caught the bug because in those tests, the size of the stripped user-defined timestamps bytes is smaller than the length of the value. In order to avoid the conditional operation, this PR changed the original trivial `std::min` operation into an arithmetic operation. Since this is a change in a hot and performance critical path, I did the following benchmark to check no observable regression is introduced.
```TEST_TMPDIR=/dev/shm/rocksdb1 ./db_bench -benchmarks=fillseq -memtablerep=vector -allow_concurrent_memtable_write=false -num=50000000```
Compiled with DEBUG_LEVEL=0
Test vs. control runs simulaneous for better accuracy, units = ops/sec
PR vs base:
Round 1: 350652 vs 349055
Round 2: 365733 vs 364308
Round 3: 355681 vs 354475
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11557
Test Plan:
New timestamp specific test added or existing tests augmented, both are parameterized with `UserDefinedTimestampTestMode`:
`UserDefinedTimestampTestMode::kNormal` -> UDT feature enabled, write / read with min timestamp
`UserDefinedTimestampTestMode::kStripUserDefinedTimestamps` -> UDT feature enabled, write / read with min timestamp, set Options.persist_user_defined_timestamps to false.
```
make all check
./db_wal_test --gtest_filter="*WithTimestamp*"
./flush_job_test --gtest_filter="*WithTimestamp*"
./repair_test --gtest_filter="*WithTimestamp*"
./block_based_table_reader_test
```
Reviewed By: pdillinger
Differential Revision: D47027664
Pulled By: jowlyzhang
fbshipit-source-id: e729193b6334dfc63aaa736d684d907a022571f5
Summary:
1. Public API change: Replace `use_async_io` API in file_system with `SupportedOps` API which is used by underlying FileSystem to indicate to upper layers whether the FileSystem supports different operations introduced in `enum FSSupportedOps `. Right now operations are `async_io` and whether FS will provide its own buffer during reads or not. The api is changed to extend it to various FileSystem operations in one API rather than creating a separate API for each operation.
2. Provide support for underlying FS to pass their own buffer during Reads (async and sync read) instead of using RocksDB provided `scratch` (buffer) in `FSReadRequest`. Currently only MultiRead supports it and later will be extended to other reads as well (point lookup, scan etc). More details in how to enable in file_system.h
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11324
Test Plan: Tested locally
Reviewed By: anand1976
Differential Revision: D44465322
Pulled By: akankshamahajan15
fbshipit-source-id: 9ec9e08f839b5cc815e75d5dade6cd549998d0ec
Summary:
Start to record the value of the flag `AdvancedColumnFamilyOptions.persist_user_defined_timestamps` in the Manifest and table properties for a SST file when it is created. And use the recorded flag when creating a table reader for the SST file. This flag's default value is true, it is only explicitly recorded if it's false.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11515
Test Plan:
```
make all check
./version_edit_test
```
Reviewed By: ltamasi
Differential Revision: D46920386
Pulled By: jowlyzhang
fbshipit-source-id: 075c20363d3d2cc1368422ecc805617ed135cc26
Summary:
**Context/Summary:**
When db is upgrading to adopt [pr11406](https://github.com/facebook/rocksdb/pull/11406/), it's possible for RocksDB to infer a small tail size to prefetch for pre-upgrade files. Such small tail size would have caused 1 file read per index or filter partition if partitioned index or filer is used. This PR provides a UT to show this would not happen.
Misc: refactor the related UTs a bit to make this new UT more readable.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11522
Test Plan:
- New UT
If logic of upgrade is wrong e.g,
```
--- a/table/block_based/partitioned_index_reader.cc
+++ b/table/block_based/partitioned_index_reader.cc
@@ -166,7 +166,8 @@ Status PartitionIndexReader::CacheDependencies(
uint64_t prefetch_len = last_off - prefetch_off;
std::unique_ptr<FilePrefetchBuffer> prefetch_buffer;
if (tail_prefetch_buffer == nullptr || !tail_prefetch_buffer->Enabled() ||
- tail_prefetch_buffer->GetPrefetchOffset() > prefetch_off) {
+ (false && tail_prefetch_buffer->GetPrefetchOffset() > prefetch_off)) {
```
, then the UT will fail like below
```
[ RUN ] PrefetchTailTest/PrefetchTailTest.UpgradeToTailSizeInManifest/0
file/prefetch_test.cc:461: Failure
Expected: (db_open_file_read.count) < (num_index_partition), actual: 38 vs 33
Received signal 11 (Segmentation fault)
```
Reviewed By: pdillinger
Differential Revision: D46546707
Pulled By: hx235
fbshipit-source-id: 9897b0a975e9055963edac5451fd1cd9d6c45d0e
Summary:
Fix a use-after-move issue in block.cc and added some unit tests.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11505
Test Plan:
```
make all check
./block_test
```
Reviewed By: ltamasi
Differential Revision: D46506188
Pulled By: jowlyzhang
fbshipit-source-id: 316ed8ddd221c00b2bce2cf9fd47eea686cd74a5
Summary:
**Context:**
[PR11406](https://github.com/facebook/rocksdb/pull/11406/) caused more frequent read during db open reading files with no `tail_size` in the manifest as part of the upgrade to 11406. This is due to that PR introduced
- [smaller](https://github.com/facebook/rocksdb/pull/11406/files#diff-57ed8c49db2bdd4db7618646a177397674bbf25beacacecb104070071d30129fR833) prefetch tail buffer size compared to pre-11406 for small files (< 52 MB) when `tail_prefetch_stats` infers tail size to be 0 (usually happens when the stats does not have much historical data to infer early on)
- more read (up to # of partitioned filter/index) when such small prefetch tail buffer does not contain all the partitioned filter/index needed in CacheDependencies() since the [fallback logic](https://github.com/facebook/rocksdb/pull/11406/files#diff-d98f1a83de24412ad7f3527725dae7e28851c7222622c3cdb832d3cdf24bbf9fR165-R179) that prefetches all partitions at once will be [skipped](url) when such a small prefetch tail buffer is passed in
**Summary:**
- Revert the fallback prefetch buffer size change to preserve existing behavior fully during upgrading in `BlockBasedTable::PrefetchTail()`
- Use passed-in prefetch tail buffer in `CacheDependencies()` only if it has a smaller offset than the the offset of first partition filter/index, that is, at least as good as the existing prefetching behavior
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11516
Test Plan:
- db bench
Create db with small files prior to PR 11406
```
./db_bench -db=/tmp/testdb/ --partition_index_and_filters=1 --statistics=1 -benchmarks=fillseq -key_size=3200 -value_size=5 -num=1000000 -write_buffer_size=6550000 -disable_auto_compactions=true -compression_type=zstd`
```
Read db to see if post-pr has lower read qps (i.e, rocksdb.file.read.db.open.micros count) during db open.
```
./db_bench -use_direct_reads=1 --file_opening_threads=1 --threads=1 --use_existing_db=1 --seed=1682546046158958 --partition_index_and_filters=1 --statistics=1 --db=/tmp/testdb/ --benchmarks=readrandom --key_size=3200 --value_size=5 --num=100 --disable_auto_compactions=true --compression_type=zstd
```
Pre-PR:
```
rocksdb.file.read.db.open.micros P50 : 3.399023 P95 : 5.924468 P99 : 12.408333 P100 : 29.000000 COUNT : 611 SUM : 2539
```
Post-PR:
```
rocksdb.file.read.db.open.micros P50 : 593.736842 P95 : 861.605263 P99 : 1212.868421 P100 : 2663.000000 COUNT : 585 SUM : 345349
```
_Note: To control the starting offset of the prefetch tail buffer easier, I manually override the following to eliminate the effect of alignment_
```
class PosixRandomAccessFile : public FSRandomAccessFile {
virtual size_t GetRequiredBufferAlignment() const override {
- return logical_sector_size_;
+ return 1;
}
```
- CI
Reviewed By: pdillinger
Differential Revision: D46472566
Pulled By: hx235
fbshipit-source-id: 2fe14ac8d489d15b0e08e6f8fe4f46d5f110978e
Summary:
Add support to strip timestamp in block based table builder and pad timestamp in block based table reader.
On the write path, use the per column family option `AdvancedColumnFamilyOptions.persist_user_defined_timestamps` to indicate whether user-defined timestamps should be stripped for all block based tables created for the column family.
On the read path, added a per table `TableReadOption.user_defined_timestamps_persisted` to flag whether the user keys in the table contains user defined timestamps.
This patch is mostly passing the related flags down to the block building/parsing level with the exception of handling the `first_internal_key` in `IndexValue`, which is included in the `IndexBuilder` level. The value part of range deletion entries should have a similar handling, I haven't decided where to best fit this piece of logic, I will do it in a follow up.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11495
Test Plan:
Existing test `BlockBasedTableReaderTest` is parameterized to run with:
1) different UDT test modes: kNone, kNormal, kStripUserDefinedTimestamp
2) all four index types, when index type is `kTwoLevelIndexSearch`, also enables partitioned filters
3) parallel vs non-parallel compression
4) enable/disable compression dictionary.
Also added tests for API `BlockBasedTableReader::NewIterator`.
`PartitionedFilterBlockTest` is parameterized to run with different UDT test modes:kNone, kNormal, kStripUserDefinedTimestamp.
```
make all check
./block_based_table_reader_test
./partitioned_filter_block_test
```
Reviewed By: ltamasi
Differential Revision: D46344577
Pulled By: jowlyzhang
fbshipit-source-id: 93ac8542b19319d1298712b8bed908c8831ba675
Summary:
This patch adds support in `BlockBuilder` to strip user-defined timestamp from the `key` added via `Add(key, value)` and its equivalent APIs. The stripping logic is different when the key is either a user key or an internal key, so the `BlockBuilder` is created with a flag to indicate that. This patch also add support on the read path to APIs `NewIndexIterator`, `NewDataIterator` to support pad a min timestamp.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11472
Test Plan:
Three test modes are added to parameterize existing tests:
UserDefinedTimestampTestMode::kNone -> UDT feature is not enabled
UserDefinedTimestampTestMode::kNormal -> UDT feature enabled, write / read with min timestamp
UserDefinedTimestampTestMode::kStripUserDefinedTimestamps -> UDT feature enabled, write / read with min timestamp, set `persist_user_defined_timestamps` where it applies to false.
The tests read/write with min timestamp so that point read and range scan can correctly read values in all three test modes.
`block_test` are parameterized to run with above three test modes and some additional parameteriazation
```
make all check
./block_test --gtest_filter="P/BlockTest*"
./block_test --gtest_filter="P/IndexBlockTest*"
```
Reviewed By: ltamasi
Differential Revision: D46200539
Pulled By: jowlyzhang
fbshipit-source-id: 59f5d6b584639976b69c2943eba723bd47d9b3c0
Summary:
We want to know more about opportunities for better range filters, and the effectiveness of our own range filters. Currently the stats are very limited, essentially logging just hits and misses against prefix filters for range scans in BLOOM_FILTER_PREFIX_* without tracking the false positive rate. Perhaps confusingly, when prefix filters are used for point queries, the stats are currently going into the non-PREFIX tickers.
This change does several things:
* Introduce new stat tickers for seeks and related filtering, \*LEVEL_SEEK\*
* Most importantly, allows us to see opportunities for range filtering. Specifically, we can count how many times a seek in an SST file accesses at least one data block, and how many times at least one value() is then accessed. If a data block was accessed but no value(), we can generally assume that the key(s) seen was(were) not of interest so could have been filtered with the right kind of filter, avoiding the data block access.
* We can get the same level of detail when a filter (for now, prefix Bloom/ribbon) is used, or not. Specifically, we can infer a false positive rate for prefix filters (not available before) from the seek "false positive" rate: when a data block is accessed but no value() is called. (There can be other explanations for a seek false positive, but in typical iterator usage it would indicate a filter false positive.)
* For efficiency, I wanted to avoid making additional calls to the prefix extractor (or key comparisons, etc.), which would be required if we wanted to more precisely detect filter false positives. I believe that instrumenting value() is the best balance of efficiency vs. accurately measuring what we are often interested in.
* The stats are divided between last level and non-last levels, to help understand potential tiered storage use cases.
* The old BLOOM_FILTER_PREFIX_* stats have a different meaning: no longer referring to iterators but to point queries using prefix filters. BLOOM_FILTER_PREFIX_TRUE_POSITIVE is added for computing the prefix false positive rate on point queries, which can be due to filter false positives as well as different keys with the same prefix.
* Similarly, the non-PREFIX BLOOM_FILTER stats are now for whole key filtering only.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11460
Test Plan:
unit tests updated, including updating many to pop the stat value since last read to improve test
readability and maintainability.
Performance test shows a consistent small improvement with these changes, both with clang and with gcc. CPU profile indicates that RecordTick is using less CPU, and this makes sense at least for a high filter miss rate. Before, we were recording two ticks per filter miss in iterators (CHECKED & USEFUL) and now recording just one (FILTERED).
Create DB with
```
TEST_TMPDIR=/dev/shm ./db_bench -benchmarks=fillrandom -num=10000000 -disable_wal=1 -write_buffer_size=30000000 -bloom_bits=8 -compaction_style=2 -fifo_compaction_max_table_files_size_mb=10000 -fifo_compaction_allow_compaction=0 -prefix_size=8
```
And run simultaneous before&after with
```
TEST_TMPDIR=/dev/shm ./db_bench -readonly -benchmarks=seekrandom[-X1000] -num=10000000 -bloom_bits=8 -compaction_style=2 -fifo_compaction_max_table_files_size_mb=10000 -fifo_compaction_allow_compaction=0 -prefix_size=8 -seek_nexts=1 -duration=20 -seed=43 -threads=8 -cache_size=1000000000 -statistics
```
Before: seekrandom [AVG 275 runs] : 189680 (± 222) ops/sec; 18.4 (± 0.0) MB/sec
After: seekrandom [AVG 275 runs] : 197110 (± 208) ops/sec; 19.1 (± 0.0) MB/sec
Reviewed By: ajkr
Differential Revision: D46029177
Pulled By: pdillinger
fbshipit-source-id: cdace79a2ea548d46c5900b068c5b7c3a02e5822
Summary:
In IDE navigation I find it annoying that there are two statistics.h files (etc.) and often land on the wrong one. Here I migrate several headers to use the blah.h <- blah_impl.h <- blah.cc idiom. Although clang-format wants "blah.h" to be the top include for "blah.cc", I think overall this is an improvement.
No public API changes.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11408
Test Plan: existing tests
Reviewed By: ltamasi
Differential Revision: D45456696
Pulled By: pdillinger
fbshipit-source-id: 809d931253f3272c908cf5facf7e1d32fc507373
Summary:
Added a ticker stat, `BLOCK_CHECKSUM_MISMATCH_COUNT`, to count how many block checksum verifications detected a mismatch.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11438
Test Plan: new unit test
Reviewed By: pdillinger
Differential Revision: D45788179
Pulled By: ajkr
fbshipit-source-id: e2b44eba7c23b3e110ebe69eaa78a710dec2590f
Summary:
**Context:**
We prefetch the tail part of a SST file (i.e, the blocks after data blocks till the end of the file) during each SST file open in hope to prefetch all the stuff at once ahead of time for later read e.g, footer, meta index, filter/index etc. The existing approach to estimate the tail size to prefetch is through `TailPrefetchStats` heuristics introduced in https://github.com/facebook/rocksdb/pull/4156, which has caused small reads in unlucky case (e.g, small read into the tail buffer during table open in thread 1 under the same BlockBasedTableFactory object can make thread 2's tail prefetching use a small size that it shouldn't) and is hard to debug. Therefore we decide to record the exact tail size and use it directly to prefetch tail of the SST instead of relying heuristics.
**Summary:**
- Obtain and record in manifest the tail size in `BlockBasedTableBuilder::Finish()`
- For backward compatibility, we fall back to TailPrefetchStats and last to simple heuristics that the tail size is a linear portion of the file size - see PR conversation for more.
- Make`tail_start_offset` part of the table properties and deduct tail size to record in manifest for external files (e.g, file ingestion, import CF) and db repair (with no access to manifest).
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11406
Test Plan:
1. New UT
2. db bench
Note: db bench on /tmp/ where direct read is supported is too slow to finish and the default pinning setting in db bench is not helpful to profile # sst read of Get. Therefore I hacked the following to obtain the following comparison.
```
diff --git a/table/block_based/block_based_table_reader.cc b/table/block_based/block_based_table_reader.cc
index bd5669f0f..791484c1f 100644
--- a/table/block_based/block_based_table_reader.cc
+++ b/table/block_based/block_based_table_reader.cc
@@ -838,7 +838,7 @@ Status BlockBasedTable::PrefetchTail(
&tail_prefetch_size);
// Try file system prefetch
- if (!file->use_direct_io() && !force_direct_prefetch) {
+ if (false && !file->use_direct_io() && !force_direct_prefetch) {
if (!file->Prefetch(prefetch_off, prefetch_len, ro.rate_limiter_priority)
.IsNotSupported()) {
prefetch_buffer->reset(new FilePrefetchBuffer(
diff --git a/tools/db_bench_tool.cc b/tools/db_bench_tool.cc
index ea40f5fa0..39a0ac385 100644
--- a/tools/db_bench_tool.cc
+++ b/tools/db_bench_tool.cc
@@ -4191,6 +4191,8 @@ class Benchmark {
std::shared_ptr<TableFactory>(NewCuckooTableFactory(table_options));
} else {
BlockBasedTableOptions block_based_options;
+ block_based_options.metadata_cache_options.partition_pinning =
+ PinningTier::kAll;
block_based_options.checksum =
static_cast<ChecksumType>(FLAGS_checksum_type);
if (FLAGS_use_hash_search) {
```
Create DB
```
./db_bench --bloom_bits=3 --use_existing_db=1 --seed=1682546046158958 --partition_index_and_filters=1 --statistics=1 -db=/dev/shm/testdb/ -benchmarks=readrandom -key_size=3200 -value_size=512 -num=1000000 -write_buffer_size=6550000 -disable_auto_compactions=false -target_file_size_base=6550000 -compression_type=none
```
ReadRandom
```
./db_bench --bloom_bits=3 --use_existing_db=1 --seed=1682546046158958 --partition_index_and_filters=1 --statistics=1 -db=/dev/shm/testdb/ -benchmarks=readrandom -key_size=3200 -value_size=512 -num=1000000 -write_buffer_size=6550000 -disable_auto_compactions=false -target_file_size_base=6550000 -compression_type=none
```
(a) Existing (Use TailPrefetchStats for tail size + use seperate prefetch buffer in PartitionedFilter/IndexReader::CacheDependencies())
```
rocksdb.table.open.prefetch.tail.hit COUNT : 3395
rocksdb.sst.read.micros P50 : 5.655570 P95 : 9.931396 P99 : 14.845454 P100 : 585.000000 COUNT : 999905 SUM : 6590614
```
(b) This PR (Record tail size + use the same tail buffer in PartitionedFilter/IndexReader::CacheDependencies())
```
rocksdb.table.open.prefetch.tail.hit COUNT : 14257
rocksdb.sst.read.micros P50 : 5.173347 P95 : 9.015017 P99 : 12.912610 P100 : 228.000000 COUNT : 998547 SUM : 5976540
```
As we can see, we increase the prefetch tail hit count and decrease SST read count with this PR
3. Test backward compatibility by stepping through reading with post-PR code on a db generated pre-PR.
Reviewed By: pdillinger
Differential Revision: D45413346
Pulled By: hx235
fbshipit-source-id: 7d5e36a60a72477218f79905168d688452a4c064
Summary:
add option `block_protection_bytes_per_key` and implementation for block per key-value checksum. The main changes are
1. checksum construction and verification in block.cc/h
2. pass the option `block_protection_bytes_per_key` around (mainly for methods defined in table_cache.h)
3. unit tests/crash test updates
Tests:
* Added unit tests
* Crash test: `python3 tools/db_crashtest.py blackbox --simple --block_protection_bytes_per_key=1 --write_buffer_size=1048576`
Follow up (maybe as a separate PR): make sure corruption status returned from BlockIters are correctly handled.
Performance:
Turning on block per KV protection has a non-trivial negative impact on read performance and costs additional memory.
For memory, each block includes additional 24 bytes for checksum-related states beside checksum itself. For CPU, I set up a DB of size ~1.2GB with 5M keys (32 bytes key and 200 bytes value) which compacts to ~5 SST files (target file size 256 MB) in L6 without compression. I tested readrandom performance with various block cache size (to mimic various cache hit rates):
```
SETUP
make OPTIMIZE_LEVEL="-O3" USE_LTO=1 DEBUG_LEVEL=0 -j32 db_bench
./db_bench -benchmarks=fillseq,compact0,waitforcompaction,compact,waitforcompaction -write_buffer_size=33554432 -level_compaction_dynamic_level_bytes=true -max_background_jobs=8 -target_file_size_base=268435456 --num=5000000 --key_size=32 --value_size=200 --compression_type=none
BENCHMARK
./db_bench --use_existing_db -benchmarks=readtocache,readrandom[-X10] --num=5000000 --key_size=32 --disable_auto_compactions --reads=1000000 --block_protection_bytes_per_key=[0|1] --cache_size=$CACHESIZE
The readrandom ops/sec looks like the following:
Block cache size: 2GB 1.2GB * 0.9 1.2GB * 0.8 1.2GB * 0.5 8MB
Main 240805 223604 198176 161653 139040
PR prot_bytes=0 238691 226693 200127 161082 141153
PR prot_bytes=1 214983 193199 178532 137013 108211
prot_bytes=1 vs -10% -15% -10.8% -15% -23%
prot_bytes=0
```
The benchmark has a lot of variance, but there was a 5% to 25% regression in this benchmark with different cache hit rates.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11287
Reviewed By: ajkr
Differential Revision: D43970708
Pulled By: cbi42
fbshipit-source-id: ef98d898b71779846fa74212b9ec9e08b7183940
Summary:
Tweak some bounds and things, and reduce risk of surprise results by running on all supported compressions (mostly).
Also improves the precise compressibility of CompressibleString by using RandomBinaryString.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11396
Test Plan: updated tests
Reviewed By: ltamasi
Differential Revision: D45211938
Pulled By: pdillinger
fbshipit-source-id: 9dc1dd8574a60a9364efe18558be66d31a35598b
Summary:
## Option API updates
* Add new CompressionOptions::max_compressed_bytes_per_kb, which corresponds to 1024.0 / min allowable compression ratio. This avoids the hard-coded minimum ratio of 8/7.
* Remove unnecessary constructor for CompressionOptions.
* Document undocumented CompressionOptions. Use idiom for default values shown clearly in one place (not precariously repeated).
## Stat API updates
* Deprecate the BYTES_COMPRESSED, BYTES_DECOMPRESSED histograms. Histograms incur substantial extra space & time costs compared to tickers, and the distribution of uncompressed data block sizes tends to be uninteresting. If we're interested in that distribution, I don't see why it should be limited to blocks stored as compressed.
* Deprecate the NUMBER_BLOCK_NOT_COMPRESSED ticker, because the name is very confusing.
* New or existing tickers relevant to compression:
* BYTES_COMPRESSED_FROM
* BYTES_COMPRESSED_TO
* BYTES_COMPRESSION_BYPASSED
* BYTES_COMPRESSION_REJECTED
* COMPACT_WRITE_BYTES + FLUSH_WRITE_BYTES (both existing)
* NUMBER_BLOCK_COMPRESSED (existing)
* NUMBER_BLOCK_COMPRESSION_BYPASSED
* NUMBER_BLOCK_COMPRESSION_REJECTED
* BYTES_DECOMPRESSED_FROM
* BYTES_DECOMPRESSED_TO
We can compute a number of things with these stats:
* "Successful" compression ratio: BYTES_COMPRESSED_FROM / BYTES_COMPRESSED_TO
* Compression ratio of data on which compression was attempted: (BYTES_COMPRESSED_FROM + BYTES_COMPRESSION_REJECTED) / (BYTES_COMPRESSED_TO + BYTES_COMPRESSION_REJECTED)
* Compression ratio of data that could be eligible for compression: (BYTES_COMPRESSED_FROM + X) / (BYTES_COMPRESSED_TO + X) where X = BYTES_COMPRESSION_REJECTED + NUMBER_BLOCK_COMPRESSION_REJECTED
* Overall SST compression ratio (compression disabled vs. actual): (Y - BYTES_COMPRESSED_TO + BYTES_COMPRESSED_FROM) / Y where Y = COMPACT_WRITE_BYTES + FLUSH_WRITE_BYTES
Keeping _REJECTED separate from _BYPASSED helps us to understand "wasted" CPU time in compression.
## BlockBasedTableBuilder
Various small refactorings, optimizations, and name clean-ups.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11388
Test Plan:
unit tests added
* `options_settable_test.cc`: use non-deprecated idiom for configuring CompressionOptions from string. The old idiom is tested elsewhere and does not need to be updated to support the new field.
Reviewed By: ajkr
Differential Revision: D45128202
Pulled By: pdillinger
fbshipit-source-id: 5a652bf5c022b7ec340cf79018cccf0686962803
Summary:
**Context:**
The existing stat rocksdb.sst.read.micros does not reflect each of compaction and flush cases but aggregate them, which is not so helpful for us to understand IO read behavior of each of them.
**Summary**
- Update `StopWatch` and `RandomAccessFileReader` to record `rocksdb.sst.read.micros` and `rocksdb.file.{flush/compaction}.read.micros`
- Fixed the default histogram in `RandomAccessFileReader`
- New field `ReadOptions/IOOptions::io_activity`; Pass `ReadOptions` through paths under db open, flush and compaction to where we can prepare `IOOptions` and pass it to `RandomAccessFileReader`
- Use `thread_status_util` for assertion in `DbStressFSWrapper` for continuous testing on we are passing correct `io_activity` under db open, flush and compaction
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11288
Test Plan:
- **Stress test**
- **Db bench 1: rocksdb.sst.read.micros COUNT ≈ sum of rocksdb.file.read.flush.micros's and rocksdb.file.read.compaction.micros's.** (without blob)
- May not be exactly the same due to `HistogramStat::Add` only guarantees atomic not accuracy across threads.
```
./db_bench -db=/dev/shm/testdb/ -statistics=true -benchmarks="fillseq" -key_size=32 -value_size=512 -num=50000 -write_buffer_size=655 -target_file_size_base=655 -disable_auto_compactions=false -compression_type=none -bloom_bits=3 (-use_plain_table=1 -prefix_size=10)
```
```
// BlockBasedTable
rocksdb.sst.read.micros P50 : 2.009374 P95 : 4.968548 P99 : 8.110362 P100 : 43.000000 COUNT : 40456 SUM : 114805
rocksdb.file.read.flush.micros P50 : 1.871841 P95 : 3.872407 P99 : 5.540541 P100 : 43.000000 COUNT : 2250 SUM : 6116
rocksdb.file.read.compaction.micros P50 : 2.023109 P95 : 5.029149 P99 : 8.196910 P100 : 26.000000 COUNT : 38206 SUM : 108689
// PlainTable
Does not apply
```
- **Db bench 2: performance**
**Read**
SETUP: db with 900 files
```
./db_bench -db=/dev/shm/testdb/ -benchmarks="fillseq" -key_size=32 -value_size=512 -num=50000 -write_buffer_size=655 -disable_auto_compactions=true -target_file_size_base=655 -compression_type=none
```run till convergence
```
./db_bench -seed=1678564177044286 -use_existing_db=true -db=/dev/shm/testdb -benchmarks=readrandom[-X60] -statistics=true -num=1000000 -disable_auto_compactions=true -compression_type=none -bloom_bits=3
```
Pre-change
`readrandom [AVG 60 runs] : 21568 (± 248) ops/sec`
Post-change (no regression, -0.3%)
`readrandom [AVG 60 runs] : 21486 (± 236) ops/sec`
**Compaction/Flush**run till convergence
```
./db_bench -db=/dev/shm/testdb2/ -seed=1678564177044286 -benchmarks="fillseq[-X60]" -key_size=32 -value_size=512 -num=50000 -write_buffer_size=655 -disable_auto_compactions=false -target_file_size_base=655 -compression_type=none
rocksdb.sst.read.micros COUNT : 33820
rocksdb.sst.read.flush.micros COUNT : 1800
rocksdb.sst.read.compaction.micros COUNT : 32020
```
Pre-change
`fillseq [AVG 46 runs] : 1391 (± 214) ops/sec; 0.7 (± 0.1) MB/sec`
Post-change (no regression, ~-0.4%)
`fillseq [AVG 46 runs] : 1385 (± 216) ops/sec; 0.7 (± 0.1) MB/sec`
Reviewed By: ajkr
Differential Revision: D44007011
Pulled By: hx235
fbshipit-source-id: a54c89e4846dfc9a135389edf3f3eedfea257132
Summary:
Because of this failure with snappy 1.1.8, ROCKSDB_NO_FBCODE=1
```
Value 3531 is not in range [2000, 3525]
table/table_test.cc:4231: Failure
```
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11384
Test Plan: run updated test in failing configuration
Reviewed By: ajkr
Differential Revision: D45057161
Pulled By: pdillinger
fbshipit-source-id: 397054f08033315e2e2bd9410f1fa32ddbf3b9c8
Summary:
After https://github.com/facebook/rocksdb/issues/11301, I wasn't sure whether I had regressed block cache tracing with MultiGet. Demo PR https://github.com/facebook/rocksdb/issues/11330 shows the flawed state of tracing MultiGet before my change, and based on the unit test, there was essentially no change in tracing behavior with https://github.com/facebook/rocksdb/issues/11301. This change is to leave that code and behavior better than I found it.
This change is not intended to change any production behaviors except when block cache tracing is active, though might improve general read path efficiency by disabling some related tracking when such tracing is disabled.
More detail on production code:
* Refactoring to consolidate the construction of BlockCacheTraceRecord, and other related functionality, in block-based table reader, though it's somewhat awkward to preserve an optimization to avoid copying Slices into temporary strings in BlockCacheLookupContext.
* Accurately track cache hits and misses (etc.) for each data block accessed by a MultiGet(). (Previously reported hits as misses.)
* Reduced repeated checking of `block_cache_tracer_` state (by creating lookup_context only when active) for efficiency and to reduce the risk of corner case bugs where tracing is enabled or disabled for different parts of a read op. (See a TODO below)
* Improved estimate calculation for num_keys_in_block (see code comment)
Possible follow-up:
* `XXX:` use_cache=true means double cache query? (possible double-query of block cache when allow_mmap_reads=true)
* `TODO:` need more than one lookup_context here to track individual filter and index partition hits and misses
* `TODO:` optimize more state checks of `block_cache_tracer_` down to `lookup_context != nullptr`
* Pre-existing `XXX:` There appear to be 'break' statements above that bypass this writing of the block cache trace record
* Expand test coverage (see below)
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11339
Test Plan:
* Added a basic unit test for block cache tracing MultiGet, for now just covering one data block with two keys.
* Added HitMissCountingCache to independently verify that the actual block cache trace and expected block cache trace also agree with the actual number of cache hits / misses (nothing missing or mislabeled). For now only used with MultiGet test.
* Better testing of num_keys_in_block, for now just with MultiGet
* Misc improvements to table_test to improve clarity, such as making it clear that certain keys are auto-inserted at the start of every test.
Performance test:
Testing multireadrandom as in https://github.com/facebook/rocksdb/issues/11301, except averaging over distinct runs rather than [-X30] which doesn't seem to sufficiently reset after each run to work as an independent test run.
Base with revert of 11301: 3148926 ops/sec
Base: 3019146 ops/sec
New: 2999529 ops/sec
Possibly a tiny MultiGet CPU regression with this change. We are now always allocating an additional vector for the LookupContexts. I'm still contemplating options to try to correct the regression in https://github.com/facebook/rocksdb/issues/11301.
Testing readrandom:
Base with revert of 11301: 2311988
Base: 2281726
New: 2299722
Possibly a tiny Get CPU improvement with this change. We are now avoiding some unnecessary LookupContext population.
Reviewed By: akankshamahajan15
Differential Revision: D44557845
Pulled By: pdillinger
fbshipit-source-id: b841691799d2a48fb59cc8880dc7cbb1e107ae3d
Summary:
If RocksDB enables user-defined timestamp, then RocksDB read path can filter table files by the min/max timestamps of each file. If application wants to lookup a key that is the most recent and visible to a certain timestamp ts, then we can compare ts with the min_ts of each file. If ts < min_ts, then we know all keys in the file is not visible at time ts, then we do not have to open the file. This can also save an IO.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11332
Reviewed By: pdillinger
Differential Revision: D44763497
Pulled By: guowentian
fbshipit-source-id: abde346b9f18480fe03c04e4006e7d62aa9c22a8
Summary:
... which increases default number of shards from 16 to 64. Although the default block cache size is only recommended for applications where RocksDB is not performance-critical, under stress conditions, block cache mutex contention could become a performance bottleneck. This change of default should alleviate that.
Note that reducing the size of cache shards (recommended minimum 512MB) could cause thrashing, e.g. on filter blocks, so capacity needs to increase to safely increase number of shards.
The 8MB default dates back to 2011 or earlier (f779e7a5), when the most simultaneous threads you could get from a single CPU socket was 20 (e.g. Intel Xeon E7-8870). Now more than 100 is available.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11350
Test Plan: unit tests updated
Reviewed By: cbi42
Differential Revision: D44674873
Pulled By: pdillinger
fbshipit-source-id: 91ed3070789b42679283c7e6dc97c41a6a97bdf4
Summary:
Internally refactors SecondaryCache integration out of LRUCache specifically and into a wrapper/adapter class that works with various Cache implementations. Notably, this relies on separating the notion of async lookup handles from other cache handles, so that HyperClockCache doesn't have to deal with the problem of allocating handles from the hash table for lookups that might fail anyway, and might be on the same key without support for coalescing. (LRUCache's hash table can incorporate previously allocated handles thanks to its pointer indirection.) Specifically, I'm worried about the case in which hundreds of threads try to access the same block and probing in the hash table degrades to linear search on the pile of entries with the same key.
This change is a big step in the direction of supporting stacked SecondaryCaches, but there are obstacles to completing that. Especially, there is no SecondaryCache hook for evictions to pass from one to the next. It has been proposed that evictions be transmitted simply as the persisted data (as in SaveToCallback), but given the current structure provided by the CacheItemHelpers, that would require an extra copy of the block data, because there's intentionally no way to ask for a contiguous Slice of the data (to allow for flexibility in storage). `AsyncLookupHandle` and the re-worked `WaitAll()` should be essentially prepared for stacked SecondaryCaches, but several "TODO with stacked secondaries" issues remain in various places.
It could be argued that the stacking instead be done as a SecondaryCache adapter that wraps two (or more) SecondaryCaches, but at least with the current API that would require an extra heap allocation on SecondaryCache Lookup for a wrapper SecondaryCacheResultHandle that can transfer a Lookup between secondaries. We could also consider trying to unify the Cache and SecondaryCache APIs, though that might be difficult if `AsyncLookupHandle` is kept a fixed struct.
## cache.h (public API)
Moves `secondary_cache` option from LRUCacheOptions to ShardedCacheOptions so that it is applicable to HyperClockCache.
## advanced_cache.h (advanced public API)
* Add `Cache::CreateStandalone()` so that the SecondaryCache support wrapper can use it.
* Add `SetEvictionCallback()` / `eviction_callback_` so that the SecondaryCache support wrapper can use it. Only a single callback is supported for efficiency. If there is ever a need for more than one, hopefully that can be handled with a broadcast callback wrapper.
These are essentially the two "extra" pieces of `Cache` for pulling out specific SecondaryCache support from the `Cache` implementation. I think it's a good trade-off as these are reasonable, limited, and reusable "cut points" into the `Cache` implementations.
* Remove async capability from standard `Lookup()` (getting rid of awkward restrictions on pending Handles) and add `AsyncLookupHandle` and `StartAsyncLookup()`. As noted in the comments, the full struct of `AsyncLookupHandle` is exposed so that it can be stack allocated, for efficiency, though more data is being copied around than before, which could impact performance. (Lookup info -> AsyncLookupHandle -> Handle vs. Lookup info -> Handle)
I could foresee a future in which a Cache internally saves a pointer to the AsyncLookupHandle, which means it's dangerous to allow it to be copyable or even movable. It also means it's not compatible with std::vector (which I don't like requiring as an API parameter anyway), so `WaitAll()` expects any contiguous array of AsyncLookupHandles. I believe this is best for common case efficiency, while behaving well in other cases also. For example, `WaitAll()` has no effect on default-constructed AsyncLookupHandles, which look like a completed cache miss.
## cacheable_entry.h
A couple of functions are obsolete because Cache::Handle can no longer be pending.
## cache.cc
Provides default implementations for new or revamped Cache functions, especially appropriate for non-blocking caches.
## secondary_cache_adapter.{h,cc}
The full details of the Cache wrapper adding SecondaryCache support. Essentially replicates the SecondaryCache handling that was in LRUCache, but obviously refactored. There is a bit of logic duplication, where Lookup() is essentially a manually optimized version of StartAsyncLookup() and Wait(), but it's roughly a dozen lines of code.
## sharded_cache.h, typed_cache.h, charged_cache.{h,cc}, sim_cache.cc
Simply updated for Cache API changes.
## lru_cache.{h,cc}
Carefully remove SecondaryCache logic, implement `CreateStandalone` and eviction handler functionality.
## clock_cache.{h,cc}
Expose existing `CreateStandalone` functionality, add eviction handler functionality. Light refactoring.
## block_based_table_reader*
Mostly re-worked the only usage of async Lookup, which is in BlockBasedTable::MultiGet. Used arrays in place of autovector in some places for efficiency. Simplified some logic by not trying to process some cache results before they're all ready.
Created new function `BlockBasedTable::GetCachePriority()` to reduce some pre-existing code duplication (and avoid making it worse).
Fixed at least one small bug from the prior confusing mixture of async and sync Lookups. In MaybeReadBlockAndLoadToCache(), called by RetrieveBlock(), called by MultiGet() with wait=false, is_cache_hit for the block_cache_tracer entry would not be set to true if the handle was pending after Lookup and before Wait.
## Intended follow-up work
* Figure out if there are any missing stats or block_cache_tracer work in refactored BlockBasedTable::MultiGet
* Stacked secondary caches (see above discussion)
* See if we can make up for the small MultiGet performance regression.
* Study more performance with SecondaryCache
* Items evicted from over-full LRUCache in Release were not being demoted to SecondaryCache, and still aren't to minimize unit test churn. Ideally they would be demoted, but it's an exceptional case so not a big deal.
* Use CreateStandalone for cache reservations (save unnecessary hash table operations). Not a big deal, but worthy cleanup.
* Somehow I got the contract for SecondaryCache::Insert wrong in #10945. (Doesn't take ownership!) That API comment needs to be fixed, but didn't want to mingle that in here.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11301
Test Plan:
## Unit tests
Generally updated to include HCC in SecondaryCache tests, though HyperClockCache has some different, less strict behaviors that leads to some tests not really being set up to work with it. Some of the tests remain disabled with it, but I think we have good coverage without them.
## Crash/stress test
Updated to use the new combination.
## Performance
First, let's check for regression on caches without secondary cache configured. Adding support for the eviction callback is likely to have a tiny effect, but it shouldn't be worrisome. LRUCache could benefit slightly from less logic around SecondaryCache handling. We can test with cache_bench default settings, built with DEBUG_LEVEL=0 and PORTABLE=0.
```
(while :; do base/cache_bench --cache_type=hyper_clock_cache | grep Rough; done) | awk '{ sum += $9; count++; print $0; print "Average: " int(sum / count) }'
```
**Before** this and #11299 (which could also have a small effect), running for about an hour, before & after running concurrently for each cache type:
HyperClockCache: 3168662 (average parallel ops/sec)
LRUCache: 2940127
**After** this and #11299, running for about an hour:
HyperClockCache: 3164862 (average parallel ops/sec) (0.12% slower)
LRUCache: 2940928 (0.03% faster)
This is an acceptable difference IMHO.
Next, let's consider essentially the worst case of new CPU overhead affecting overall performance. MultiGet uses the async lookup interface regardless of whether SecondaryCache or folly are used. We can configure a benchmark where all block cache queries are for data blocks, and all are hits.
Create DB and test (before and after tests running simultaneously):
```
TEST_TMPDIR=/dev/shm ./db_bench -benchmarks=fillrandom -num=30000000 -disable_wal=1 -bloom_bits=16
TEST_TMPDIR=/dev/shm base/db_bench -benchmarks=multireadrandom[-X30] -readonly -multiread_batched -batch_size=32 -num=30000000 -bloom_bits=16 -cache_size=6789000000 -duration 20 -threads=16
```
**Before**:
multireadrandom [AVG 30 runs] : 3444202 (± 57049) ops/sec; 240.9 (± 4.0) MB/sec
multireadrandom [MEDIAN 30 runs] : 3514443 ops/sec; 245.8 MB/sec
**After**:
multireadrandom [AVG 30 runs] : 3291022 (± 58851) ops/sec; 230.2 (± 4.1) MB/sec
multireadrandom [MEDIAN 30 runs] : 3366179 ops/sec; 235.4 MB/sec
So that's roughly a 3% regression, on kind of a *worst case* test of MultiGet CPU. Similar story with HyperClockCache:
**Before**:
multireadrandom [AVG 30 runs] : 3933777 (± 41840) ops/sec; 275.1 (± 2.9) MB/sec
multireadrandom [MEDIAN 30 runs] : 3970667 ops/sec; 277.7 MB/sec
**After**:
multireadrandom [AVG 30 runs] : 3755338 (± 30391) ops/sec; 262.6 (± 2.1) MB/sec
multireadrandom [MEDIAN 30 runs] : 3785696 ops/sec; 264.8 MB/sec
Roughly a 4-5% regression. Not ideal, but not the whole story, fortunately.
Let's also look at Get() in db_bench:
```
TEST_TMPDIR=/dev/shm ./db_bench -benchmarks=readrandom[-X30] -readonly -num=30000000 -bloom_bits=16 -cache_size=6789000000 -duration 20 -threads=16
```
**Before**:
readrandom [AVG 30 runs] : 2198685 (± 13412) ops/sec; 153.8 (± 0.9) MB/sec
readrandom [MEDIAN 30 runs] : 2209498 ops/sec; 154.5 MB/sec
**After**:
readrandom [AVG 30 runs] : 2292814 (± 43508) ops/sec; 160.3 (± 3.0) MB/sec
readrandom [MEDIAN 30 runs] : 2365181 ops/sec; 165.4 MB/sec
That's showing roughly a 4% improvement, perhaps because of the secondary cache code that is no longer part of LRUCache. But weirdly, HyperClockCache is also showing 2-3% improvement:
**Before**:
readrandom [AVG 30 runs] : 2272333 (± 9992) ops/sec; 158.9 (± 0.7) MB/sec
readrandom [MEDIAN 30 runs] : 2273239 ops/sec; 159.0 MB/sec
**After**:
readrandom [AVG 30 runs] : 2332407 (± 11252) ops/sec; 163.1 (± 0.8) MB/sec
readrandom [MEDIAN 30 runs] : 2335329 ops/sec; 163.3 MB/sec
Reviewed By: ltamasi
Differential Revision: D44177044
Pulled By: pdillinger
fbshipit-source-id: e808e48ff3fe2f792a79841ba617be98e48689f5
Summary:
In preparation for factoring secondary cache support out of individual Cache implementations, we can get rid of the "in secondary cache" flag on entries through a workable hack: when an entry is promoted from secondary, it is inserted in primary using a helper that lacks secondary cache support, thus preventing re-insertion into secondary cache through existing logic.
This adds to the complexity of building CacheItemHelpers, because you always have to be able to get to an equivalent helper without secondary cache support, but that complexity is reasonably isolated within RocksDB typed_cache.h and test code.
gcc-7 seems to have problems with constexpr constructor referencing `this` so removed constexpr support on CacheItemHelper.
Also refactored some related test code to share common code / functionality.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11299
Test Plan: existing tests
Reviewed By: anand1976
Differential Revision: D44101453
Pulled By: pdillinger
fbshipit-source-id: 7a59d0a3938ee40159c90c3e65d7004f6a272345
Summary:
The existing PerfContext counter `internal_merge_count` only tracks the
Merge operands applied during range scans. The patch adds a new counter
called `internal_merge_count_point_lookups` to track the same metric
for point lookups (`Get` / `MultiGet` / `GetEntity` / `MultiGetEntity`), and
also fixes a couple of cases in the iterator where the existing counter wasn't
updated.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11284
Test Plan: `make check`
Reviewed By: jowlyzhang
Differential Revision: D43926082
Pulled By: ltamasi
fbshipit-source-id: 321566d8b4cf0a3b6c9b73b7a5c984fb9bb492e9
Summary:
Adds unit tests verifying that a block payload and checksum of all zeros is not falsely considered valid data. The test exhaustively checks that for blocks up to some length (default 20K, more exhaustively 10M) of all zeros do not produce a block checksum of all zeros.
Also small refactoring of an existing checksum test to use parameterized test. (Suggest hiding whitespace changes for review.)
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11260
Test Plan:
this is the test, manual run with
`ROCKSDB_THOROUGH_CHECKSUM_TEST=1` to verify up to 10M.
Reviewed By: hx235
Differential Revision: D43706192
Pulled By: pdillinger
fbshipit-source-id: 95e721c320ca928e7fa2400c2570fb359cc30b1f
Summary:
This makes it possible to eliminate some copies in `GetEntity` / `MultiGetEntity`,
in particular when `Merge`s or blobs are involved.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11248
Test Plan: `make check`
Reviewed By: akankshamahajan15
Differential Revision: D43544215
Pulled By: ltamasi
fbshipit-source-id: bc4c8955a24bbd8bc4ab098e72133ead757f9707
Summary:
A second attempt after https://github.com/facebook/rocksdb/issues/10802, with bug fixes and refactoring. This PR updates compaction logic to take range tombstones into account when determining whether to cut the current compaction output file (https://github.com/facebook/rocksdb/issues/4811). Before this change, only point keys were considered, and range tombstones could cause large compactions. For example, if the current compaction outputs is a range tombstone [a, b) and 2 point keys y, z, they would be added to the same file, and may overlap with too many files in the next level and cause a large compaction in the future. This PR also includes ajkr's effort to simplify the logic to add range tombstones to compaction output files in `AddRangeDels()` ([https://github.com/facebook/rocksdb/issues/11078](https://github.com/facebook/rocksdb/pull/11078#issuecomment-1386078861)).
The main change is for `CompactionIterator` to emit range tombstone start keys to be processed by `CompactionOutputs`. A new class `CompactionMergingIterator` is introduced to replace `MergingIterator` under `CompactionIterator` to enable emitting of range tombstone start keys. Further improvement after this PR include cutting compaction output at some grandparent boundary key (instead of the next output key) when cutting within a range tombstone to reduce overlap with grandparents.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11113
Test Plan:
* added unit test in db_range_del_test
* crash test with a small key range: `python3 tools/db_crashtest.py blackbox --simple --max_key=100 --interval=600 --write_buffer_size=262144 --target_file_size_base=256 --max_bytes_for_level_base=262144 --block_size=128 --value_size_mult=33 --subcompactions=10 --use_multiget=1 --delpercent=3 --delrangepercent=2 --verify_iterator_with_expected_state_one_in=2 --num_iterations=10`
Reviewed By: ajkr
Differential Revision: D42655709
Pulled By: cbi42
fbshipit-source-id: 8367e36ef5640e8f21c14a3855d4a8d6e360a34c
Summary:
The primary purpose of the FactoryFunc was to support LITE mode where the ObjectRegistry was not available. With the removal of LITE mode, the function was no longer required.
Note that the MergeOperator had some private classes defined in header files. To gain access to their constructors (and name methods), the class definitions were moved into header files.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11203
Reviewed By: cbi42
Differential Revision: D43160255
Pulled By: pdillinger
fbshipit-source-id: f3a465fd5d1a7049b73ecf31e4b8c3762f6dae6c
Summary:
From HISTORY.md: Added a subcode of `Status::Corruption`, `Status::SubCode::kMergeOperatorFailed`, for users to identify corruption failures originating in the merge operator, as opposed to RocksDB's internally identified data corruptions.
This is a followup to https://github.com/facebook/rocksdb/issues/11092, where we gave users the ability to keep running a DB despite merge operator failing. Now that the DB keeps running despite such failures, they want to be able to distinguish such failures from real corruptions.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11231
Test Plan: updated unit test
Reviewed By: akankshamahajan15
Differential Revision: D43396607
Pulled By: ajkr
fbshipit-source-id: 17fbcc779ad724dafada8abd73efd38e1c5208b9
Summary:
The new `MultiGetEntity` API can be used to get a consistent view of
a batch of keys, with the results presented as wide-column entities.
Similarly to `GetEntity` and the iterator's `columns` API, if the entry
corresponding to the key is a wide-column entity to start with, it is
returned as-is, and if it is a plain key-value, it is wrapped into an entity
with a single default column.
Implementation-wise, the new API shares the logic of the batched `MultiGet`
API (via the `MultiGetCommon` methods). Both single-CF and multi-CF
`MultiGetEntity` APIs are provided, and blobs are also supported.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11222
Test Plan: `make check`
Reviewed By: akankshamahajan15
Differential Revision: D43256950
Pulled By: ltamasi
fbshipit-source-id: 47fb2cb7e2d0470e3580f43fdb2fe9e51f0e7005
Summary:
The files in `port/`, such as `port_posix.h`, are layering over the system libraries, so shouldn't include the DB-specific files like `options.h`. This PR remove this dependency.
# How
The reason that `port_posix.h` (or `port_win.h`) include `options.h` is to use `CpuPriority`, as there is a method `SetCpuPriority()` in `port_posix.h` that uses `CpuPriority.`
- I think `SetCpuPriority()` make sense to exist in `port_posix.h` as it provides has platform-dependent implementation
- `CpuPriority` enum is defined in `env.h`, but used in `rocksdb/include` and `port/`.
Hence, let us define `CpuPriority` enum in a common file, say `port_defs.h`, such that both directories `rocksdb/include` and `port/` can include.
When we remove this dependency, some other files have compile errors because they can't find definitions, so add header files to resolve
# Test
make all check -j
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11214
Reviewed By: pdillinger
Differential Revision: D43196910
Pulled By: guowentian
fbshipit-source-id: 70deccb72844cfb08fcc994f76c6ef6df5d55ab9
Summary:
The definition of the Cache class should not be needed by the vast majority of RocksDB users, so I think it is just distracting to include it in cache.h, which is primarily needed for configuring and creating caches. This change moves the class to a new header advanced_cache.h. It is just cut-and-paste except for modifying the class API comment.
In general, operations on shared_ptr<Cache> should continue to work when only a forward declaration of Cache is available, as long as all the Cache instances provided are already shared_ptr. See https://stackoverflow.com/a/17650101/454544
Also, the most common way to customize a Cache is by wrapping an existing implementation, so it makes sense to provide CacheWrapper in the public API. This was a cut-and-paste job except removing the implementation of Name() so that derived classes must provide it.
Intended follow-up: consolidate Release() into one function to reduce customization bugs / confusion
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11192
Test Plan: `make check`
Reviewed By: anand1976
Differential Revision: D43055487
Pulled By: pdillinger
fbshipit-source-id: 7b05492df35e0f30b581b4c24c579bc275b6d110
Summary:
**Context/Summary:**
As instructed by convenience.h comments, a few deprecated APIs are removed.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11120
Test Plan:
- make check & CI
- eyeball check on test semantics.
Reviewed By: pdillinger
Differential Revision: D42937507
Pulled By: hx235
fbshipit-source-id: a9e4709387da01b1d0e9148c2e210f02e9746ee1
Summary:
Use the user key on sst file for blob verification for `Get` and `MultiGet` instead of the user key passed from caller.
Add tests for `Get` and `MultiGet` operations when user defined timestamp feature is enabled in a BlobDB.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11105
Test Plan:
make V=1 db_blob_basic_test
./db_blob_basic_test --gtest_filter="DBBlobTestWithTimestamp.*"
Reviewed By: ltamasi
Differential Revision: D42716487
Pulled By: jowlyzhang
fbshipit-source-id: 5987ecbb7e56ddf46d2467a3649369390789506a
Summary:
We haven't been actively mantaining RocksDB LITE recently and the size must have been gone up significantly. We are removing the support.
Most of changes were done through following comments:
unifdef -m -UROCKSDB_LITE `git grep -l ROCKSDB_LITE | egrep '[.](cc|h)'`
by Peter Dillinger. Others changes were manually applied to build scripts, CircleCI manifests, ROCKSDB_LITE is used in an expression and file db_stress_test_base.cc.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11147
Test Plan: See CI
Reviewed By: pdillinger
Differential Revision: D42796341
fbshipit-source-id: 4920e15fc2060c2cd2221330a6d0e5e65d4b7fe2
Summary:
Like other versions before, gcc 13 moved some includes around and as a result <cstdint> is no longer transitively included [1]. Explicitly include it for uint{32,64}_t.
[1] https://gcc.gnu.org/gcc-13/porting_to.html#header-dep-changes
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11118
Reviewed By: cbi42
Differential Revision: D42711356
Pulled By: ajkr
fbshipit-source-id: 5ea257b85b7017f40fd8fdbce965336da95c55b2
Summary:
Compressed block cache is replaced by compressed secondary cache. Remove the feature.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11117
Test Plan: See CI passes
Reviewed By: pdillinger
Differential Revision: D42700164
fbshipit-source-id: 6cbb24e460da29311150865f60ecb98637f9f67d
Summary:
Prior to this PR, `FullMergeV2()` can only return `false` to indicate failure, which causes any operation invoking it to fail. During a compaction, such a failure causes the compaction to fail and causes the DB to irreversibly enter read-only mode. Some users asked for a way to allow the merge operator to fail without such widespread damage.
To limit the blast radius of merge operator failures, this PR introduces the `MergeOperationOutput::op_failure_scope` API. When unpopulated (`kDefault`) or set to `kTryMerge`, the merge operator failure handling is the same as before. When set to `kMustMerge`, merge operator failure still causes failure to operations that must merge (`Get()`, iterator, `MultiGet()`, etc.). However, under `kMustMerge`, flushes/compactions can survive merge operator failures by outputting the unmerged input operands.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11092
Reviewed By: siying
Differential Revision: D42525673
Pulled By: ajkr
fbshipit-source-id: 951dc3bf190f86347dccf3381be967565cda52ee
Summary:
`InternalIteratorBase::is_mutable_` is not used any more, remove it.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11104
Reviewed By: ajkr
Differential Revision: D42582747
Pulled By: cbi42
fbshipit-source-id: d30bf75151fc8414df0ae112a6ec4943b5b7330b
Summary:
This reverts commit f02c708aa3 since it introduced several bugs (see https://github.com/facebook/rocksdb/issues/11078 and https://github.com/facebook/rocksdb/issues/11067 for attempts to fix them) and that I do not have a high confidence to fix all of them and ensure no further ones before the next release branch cut. There are also come existing issue found during bug fixing. We will work on it and try to merge it to the release after.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11089
Test Plan: existing CI.
Reviewed By: ajkr
Differential Revision: D42505972
Pulled By: cbi42
fbshipit-source-id: 2f66dcde6b85dc94977b317c2ce513872cfbc153
Summary:
This is several refactorings bundled into one to avoid having to incrementally re-modify uses of Cache several times. Overall, there are breaking changes to Cache class, and it becomes more of low-level interface for implementing caches, especially block cache. New internal APIs make using Cache cleaner than before, and more insulated from block cache evolution. Hopefully, this is the last really big block cache refactoring, because of rather effectively decoupling the implementations from the uses. This change also removes the EXPERIMENTAL designation on the SecondaryCache support in Cache. It seems reasonably mature at this point but still subject to change/evolution (as I warn in the API docs for Cache).
The high-level motivation for this refactoring is to minimize code duplication / compounding complexity in adding SecondaryCache support to HyperClockCache (in a later PR). Other benefits listed below.
* static_cast lines of code +29 -35 (net removed 6)
* reinterpret_cast lines of code +6 -32 (net removed 26)
## cache.h and secondary_cache.h
* Always use CacheItemHelper with entries instead of just a Deleter. There are several motivations / justifications:
* Simpler for implementations to deal with just one Insert and one Lookup.
* Simpler and more efficient implementation because we don't have to track which entries are using helpers and which are using deleters
* Gets rid of hack to classify cache entries by their deleter. Instead, the CacheItemHelper includes a CacheEntryRole. This simplifies a lot of code (cache_entry_roles.h almost eliminated). Fixes https://github.com/facebook/rocksdb/issues/9428.
* Makes it trivial to adjust SecondaryCache behavior based on kind of block (e.g. don't re-compress filter blocks).
* It is arguably less convenient for many direct users of Cache, but direct users of Cache are now rare with introduction of typed_cache.h (below).
* I considered and rejected an alternative approach in which we reduce customizability by assuming each secondary cache compatible value starts with a Slice referencing the uncompressed block contents (already true or mostly true), but we apparently intend to stack secondary caches. Saving an entry from a compressed secondary to a lower tier requires custom handling offered by SaveToCallback, etc.
* Make CreateCallback part of the helper and introduce CreateContext to work with it (alternative to https://github.com/facebook/rocksdb/issues/10562). This cleans up the interface while still allowing context to be provided for loading/parsing values into primary cache. This model works for async lookup in BlockBasedTable reader (reader owns a CreateContext) under the assumption that it always waits on secondary cache operations to finish. (Otherwise, the CreateContext could be destroyed while async operation depending on it continues.) This likely contributes most to the observed performance improvement because it saves an std::function backed by a heap allocation.
* Use char* for serialized data, e.g. in SaveToCallback, where void* was confusingly used. (We use `char*` for serialized byte data all over RocksDB, with many advantages over `void*`. `memcpy` etc. are legacy APIs that should not be mimicked.)
* Add a type alias Cache::ObjectPtr = void*, so that we can better indicate the intent of the void* when it is to be the object associated with a Cache entry. Related: started (but did not complete) a refactoring to move away from "value" of a cache entry toward "object" or "obj". (It is confusing to call Cache a key-value store (like DB) when it is really storing arbitrary in-memory objects, not byte strings.)
* Remove unnecessary key param from DeleterFn. This is good for efficiency in HyperClockCache, which does not directly store the cache key in memory. (Alternative to https://github.com/facebook/rocksdb/issues/10774)
* Add allocator to Cache DeleterFn. This is a kind of future-proofing change in case we get more serious about using the Cache allocator for memory tracked by the Cache. Right now, only the uncompressed block contents are allocated using the allocator, and a pointer to that allocator is saved as part of the cached object so that the deleter can use it. (See CacheAllocationPtr.) If in the future we are able to "flatten out" our Cache objects some more, it would be good not to have to track the allocator as part of each object.
* Removes legacy `ApplyToAllCacheEntries` and changes `ApplyToAllEntries` signature for Deleter->CacheItemHelper change.
## typed_cache.h
Adds various "typed" interfaces to the Cache as internal APIs, so that most uses of Cache can use simple type safe code without casting and without explicit deleters, etc. Almost all of the non-test, non-glue code uses of Cache have been migrated. (Follow-up work: CompressedSecondaryCache deserves deeper attention to migrate.) This change expands RocksDB's internal usage of metaprogramming and SFINAE (https://en.cppreference.com/w/cpp/language/sfinae).
The existing usages of Cache are divided up at a high level into these new interfaces. See updated existing uses of Cache for examples of how these are used.
* PlaceholderCacheInterface - Used for making cache reservations, with entries that have a charge but no value.
* BasicTypedCacheInterface<TValue> - Used for primary cache storage of objects of type TValue, which can be cleaned up with std::default_delete<TValue>. The role is provided by TValue::kCacheEntryRole or given in an optional template parameter.
* FullTypedCacheInterface<TValue, TCreateContext> - Used for secondary cache compatible storage of objects of type TValue. In addition to BasicTypedCacheInterface constraints, we require TValue::ContentSlice() to return persistable data. This simplifies usage for the normal case of simple secondary cache compatibility (can give you a Slice to the data already in memory). In addition to TCreateContext performing the role of Cache::CreateContext, it is also expected to provide a factory function for creating TValue.
* For each of these, there's a "Shared" version (e.g. FullTypedSharedCacheInterface) that holds a shared_ptr to the Cache, rather than assuming external ownership by holding only a raw `Cache*`.
These interfaces introduce specific handle types for each interface instantiation, so that it's easy to see what kind of object is controlled by a handle. (Ultimately, this might not be worth the extra complexity, but it seems OK so far.)
Note: I attempted to make the cache 'charge' automatically inferred from the cache object type, such as by expecting an ApproximateMemoryUsage() function, but this is not so clean because there are cases where we need to compute the charge ahead of time and don't want to re-compute it.
## block_cache.h
This header is essentially the replacement for the old block_like_traits.h. It includes various things to support block cache access with typed_cache.h for block-based table.
## block_based_table_reader.cc
Before this change, accessing the block cache here was an awkward mix of static polymorphism (template TBlocklike) and switch-case on a dynamic BlockType value. This change mostly unifies on static polymorphism, relying on minor hacks in block_cache.h to distinguish variants of Block. We still check BlockType in some places (especially for stats, which could be improved in follow-up work) but at least the BlockType is a static constant from the template parameter. (No more awkward partial redundancy between static and dynamic info.) This likely contributes to the overall performance improvement, but hasn't been tested in isolation.
The other key source of simplification here is a more unified system of creating block cache objects: for directly populating from primary cache and for promotion from secondary cache. Both use BlockCreateContext, for context and for factory functions.
## block_based_table_builder.cc, cache_dump_load_impl.cc
Before this change, warming caches was super ugly code. Both of these source files had switch statements to basically transition from the dynamic BlockType world to the static TBlocklike world. None of that mess is needed anymore as there's a new, untyped WarmInCache function that handles all the details just as promotion from SecondaryCache would. (Fixes `TODO akanksha: Dedup below code` in block_based_table_builder.cc.)
## Everything else
Mostly just updating Cache users to use new typed APIs when reasonably possible, or changed Cache APIs when not.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10975
Test Plan:
tests updated
Performance test setup similar to https://github.com/facebook/rocksdb/issues/10626 (by cache size, LRUCache when not "hyper" for HyperClockCache):
34MB 1thread base.hyper -> kops/s: 0.745 io_bytes/op: 2.52504e+06 miss_ratio: 0.140906 max_rss_mb: 76.4844
34MB 1thread new.hyper -> kops/s: 0.751 io_bytes/op: 2.5123e+06 miss_ratio: 0.140161 max_rss_mb: 79.3594
34MB 1thread base -> kops/s: 0.254 io_bytes/op: 1.36073e+07 miss_ratio: 0.918818 max_rss_mb: 45.9297
34MB 1thread new -> kops/s: 0.252 io_bytes/op: 1.36157e+07 miss_ratio: 0.918999 max_rss_mb: 44.1523
34MB 32thread base.hyper -> kops/s: 7.272 io_bytes/op: 2.88323e+06 miss_ratio: 0.162532 max_rss_mb: 516.602
34MB 32thread new.hyper -> kops/s: 7.214 io_bytes/op: 2.99046e+06 miss_ratio: 0.168818 max_rss_mb: 518.293
34MB 32thread base -> kops/s: 3.528 io_bytes/op: 1.35722e+07 miss_ratio: 0.914691 max_rss_mb: 264.926
34MB 32thread new -> kops/s: 3.604 io_bytes/op: 1.35744e+07 miss_ratio: 0.915054 max_rss_mb: 264.488
233MB 1thread base.hyper -> kops/s: 53.909 io_bytes/op: 2552.35 miss_ratio: 0.0440566 max_rss_mb: 241.984
233MB 1thread new.hyper -> kops/s: 62.792 io_bytes/op: 2549.79 miss_ratio: 0.044043 max_rss_mb: 241.922
233MB 1thread base -> kops/s: 1.197 io_bytes/op: 2.75173e+06 miss_ratio: 0.103093 max_rss_mb: 241.559
233MB 1thread new -> kops/s: 1.199 io_bytes/op: 2.73723e+06 miss_ratio: 0.10305 max_rss_mb: 240.93
233MB 32thread base.hyper -> kops/s: 1298.69 io_bytes/op: 2539.12 miss_ratio: 0.0440307 max_rss_mb: 371.418
233MB 32thread new.hyper -> kops/s: 1421.35 io_bytes/op: 2538.75 miss_ratio: 0.0440307 max_rss_mb: 347.273
233MB 32thread base -> kops/s: 9.693 io_bytes/op: 2.77304e+06 miss_ratio: 0.103745 max_rss_mb: 569.691
233MB 32thread new -> kops/s: 9.75 io_bytes/op: 2.77559e+06 miss_ratio: 0.103798 max_rss_mb: 552.82
1597MB 1thread base.hyper -> kops/s: 58.607 io_bytes/op: 1449.14 miss_ratio: 0.0249324 max_rss_mb: 1583.55
1597MB 1thread new.hyper -> kops/s: 69.6 io_bytes/op: 1434.89 miss_ratio: 0.0247167 max_rss_mb: 1584.02
1597MB 1thread base -> kops/s: 60.478 io_bytes/op: 1421.28 miss_ratio: 0.024452 max_rss_mb: 1589.45
1597MB 1thread new -> kops/s: 63.973 io_bytes/op: 1416.07 miss_ratio: 0.0243766 max_rss_mb: 1589.24
1597MB 32thread base.hyper -> kops/s: 1436.2 io_bytes/op: 1357.93 miss_ratio: 0.0235353 max_rss_mb: 1692.92
1597MB 32thread new.hyper -> kops/s: 1605.03 io_bytes/op: 1358.04 miss_ratio: 0.023538 max_rss_mb: 1702.78
1597MB 32thread base -> kops/s: 280.059 io_bytes/op: 1350.34 miss_ratio: 0.023289 max_rss_mb: 1675.36
1597MB 32thread new -> kops/s: 283.125 io_bytes/op: 1351.05 miss_ratio: 0.0232797 max_rss_mb: 1703.83
Almost uniformly improving over base revision, especially for hot paths with HyperClockCache, up to 12% higher throughput seen (1597MB, 32thread, hyper). The improvement for that is likely coming from much simplified code for providing context for secondary cache promotion (CreateCallback/CreateContext), and possibly from less branching in block_based_table_reader. And likely a small improvement from not reconstituting key for DeleterFn.
Reviewed By: anand1976
Differential Revision: D42417818
Pulled By: pdillinger
fbshipit-source-id: f86bfdd584dce27c028b151ba56818ad14f7a432
Summary:
Reading uncompression dict block always uses sync reads, while data blocks may use async reads and prefetching. This causes problems in FilePrefetchBuffer. So avoid mixing the two by reading the uncompression dict straight from the file.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11050
Test Plan: Crash test
Reviewed By: akankshamahajan15
Differential Revision: D42194682
Pulled By: anand1976
fbshipit-source-id: aaa8b396fdfe966b157e210f5ef8501c45b7b69e
Summary:
This PR is the first step for Issue https://github.com/facebook/rocksdb/issues/4811. Currently compaction output files are cut at point keys, and the decision is made mainly in `CompactionOutputs::ShouldStopBefore()`. This makes it possible for range tombstones to cause large compactions that does not respect `max_compaction_bytes`. For example, we can have a large range tombstone that overlaps with too many files from the next level. Another example is when there is a gap between a range tombstone and another key. The first issue may be more acceptable, as a lot of data is deleted. This PR address the second issue by calling `ShouldStopBefore()` for range tombstone start keys. The main change is for `CompactionIterator` to emit range tombstone start keys to be processed by `CompactionOutputs`. A new `CompactionMergingIterator` is introduced and only used under `CompactionIterator` for this purpose. Further improvement after this PR include 1) cut compaction output at some grandparent boundary key instead of at the next point key or range tombstone start key and 2) cut compaction output file within a large range tombstone (it may be easier and reasonable to only do it for range tombstones at the end of a compaction output).
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10802
Test Plan:
- added unit tests in db_range_del_test.
- stress test: `python3 tools/db_crashtest.py whitebox --[simple|enable_ts] --verify_iterator_with_expected_state_one_in=5 --delrangepercent=5 --prefixpercent=2 --writepercent=58 --readpercen=21 --duration=36000 --range_deletion_width=1000000`
Reviewed By: ajkr, jay-zhuang
Differential Revision: D40308827
Pulled By: cbi42
fbshipit-source-id: a8fd6f70a3f09d0ef7a40e006f6c964bba8c00df
Summary:
Previously, you could get a format_version error if SST file size was too small in manifest, or a weird "too short" error if too big in manifest. Now we ensure:
* Magic number error is reported first if we attempt to open an SST file and the footer is completely bad.
* Footer errors are reported with affected file.
* If manifest file size doesn't match actual, then the error includes expected and actual sizes (if an error is reported; in some cases we allow the file to be too big)
Pull Request resolved: https://github.com/facebook/rocksdb/pull/11009
Test Plan:
unit tests added, some manual
Previously, the code for "file too short" in footer processing was only covered by some tests attempting to verify SST checksums on non-SST files (fixed).
Reviewed By: siying
Differential Revision: D41656272
Pulled By: pdillinger
fbshipit-source-id: 3da32702eb5aaedbea0e5e74742ad57edd7ad3df
Summary:
In MergingIterator, if a range tombstone's start or end key is added to minHeap/maxHeap, the key is copied. This PR removes the copying of range tombstone keys by adding InternalKey comparator that compares `Slice` for internal key and `ParsedInternalKey` directly.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10878
Test Plan:
- existing UT
- ran all flavors of stress test through sandcastle
- benchmarks: I did not get improvement when compiling with DEBUG_LEVEL=0, and saw many noise. With `OPTIMIZE_LEVEL="-O3" USE_LTO=1` I do see improvement.
```
# Favorable set up: half of the writes are DeleteRange.
TEST_TMPDIR=/tmp/rocksdb-rangedel-test-all-tombstone ./db_bench --benchmarks=fillseq,levelstats --writes_per_range_tombstone=1 --max_num_range_tombstones=1000000 --range_tombstone_width=2 --num=1000000 --max_bytes_for_level_base=4194304 --disable_auto_compactions --write_buffer_size=33554432 --key_size=50
# benchmark command
TEST_TMPDIR=/tmp/rocksdb-rangedel-test-all-tombstone ./db_bench --benchmarks=readseq[-W1][-X5],levelstats --use_existing_db=true --cache_size=3221225472 --disable_auto_compactions=true --avoid_flush_during_recovery=true --seek_nexts=100 --reads=1000000 --num=1000000 --threads=25
# main
readseq [AVG 5 runs] : 26017977 (± 371077) ops/sec; 3721.9 (± 53.1) MB/sec
readseq [MEDIAN 5 runs] : 26096905 ops/sec; 3733.2 MB/sec
# this PR
readseq [AVG 5 runs] : 27481724 (± 568758) ops/sec; 3931.3 (± 81.4) MB/sec
readseq [MEDIAN 5 runs] : 27323957 ops/sec; 3908.7 MB/sec
```
Reviewed By: ajkr
Differential Revision: D40711170
Pulled By: cbi42
fbshipit-source-id: 708cb584e2bd085a9ce0d2ef6a420489f721717f
Summary:
Currently, `iterate_upper_bound` is not checked for range tombstone keys in MergingIterator. This may impact performance when there is a large number of range tombstones right after `iterate_upper_bound`. This PR fixes this issue by checking `iterate_upper_bound` in MergingIterator for range tombstone keys.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10966
Test Plan:
- added unit test
- stress test: `python3 tools/db_crashtest.py whitebox --simple --verify_iterator_with_expected_state_one_in=5 --delrangepercent=5 --prefixpercent=18 --writepercent=48 --readpercen=15 --duration=36000 --range_deletion_width=100`
- ran different stress tests over sandcastle
- Falcon team ran some test traffic and saw reduced CPU usage on processing range tombstones.
Reviewed By: ajkr
Differential Revision: D41414172
Pulled By: cbi42
fbshipit-source-id: 9b2c29eb3abb99327c6a649bdc412e70d863f981
Summary:
Compressed block cache depends on reading the block compression marker beyond the payload block size. Only the payload bytes were being saved and loaded from SecondaryCache -> boom!
This removes some unnecessary code attempting to combine these two competing features. Note that BlockContents was previously used for block-based filter in block cache, but that support has been removed.
Also marking block_cache_compressed as deprecated in this commit as we expect it to be replaced with SecondaryCache.
This problem was discovered during refactoring but didn't want to combine bug fix with that refactoring.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10944
Test Plan: test added that fails on base revision (at least with ASAN)
Reviewed By: akankshamahajan15
Differential Revision: D41205578
Pulled By: pdillinger
fbshipit-source-id: 1b29d36c7a6552355ac6511fcdc67038ef4af29f
Summary:
The patch refines/reworks `MergeHelper::TimedFullMerge(WithEntity)`
a bit in two ways. First, it eliminates the recently introduced `TimedFullMerge`
overload, which makes the responsibilities clearer by making sure the query
result (`value` for `Get`, `columns` for `GetEntity`) is set uniformly in
`SaveValue` and `GetContext`. Second, it changes the interface of
`TimedFullMergeWithEntity` so it exposes its result in a serialized form; this
is a more decoupled design which will come in handy when adding support
for `Merge` with wide-column entities to `DBIter`.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10932
Test Plan: `make check`
Reviewed By: akankshamahajan15
Differential Revision: D41129399
Pulled By: ltamasi
fbshipit-source-id: 69d8da358c77d4fc7e8c40f4dafc2c129a710677
Summary:
The patch fixes a bug where `GetContext::Merge` (and `MergeEntity`) does not update the ticker `READ_NUM_MERGE_OPERANDS` because it implicitly uses the default parameter value of `update_num_ops_stats=false` when calling `MergeHelper::TimedFullMerge`. Also, to prevent such issues going forward, the PR removes the default parameter values from the `TimedFullMerge` methods. In addition, it removes an unused/unnecessary parameter from `TimedFullMergeWithEntity`, and does some cleanup at the call sites of these methods.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10925
Test Plan: `make check`
Reviewed By: riversand963
Differential Revision: D41096453
Pulled By: ltamasi
fbshipit-source-id: fc60646d32b4d516b8fe81e265c3f020a32fd7f8
Summary:
The patch adds `Merge` support for wide-column entities to the point lookup
APIs, i.e. `Get`, `MultiGet`, `GetEntity`, and `GetMergeOperands`. (I plan to
update the iterator and compaction logic in separate PRs.) In terms of semantics,
the `Merge` operation is applied to the default (anonymous) column; any other
columns in the entity are unaffected.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10916
Test Plan: `make check`
Reviewed By: riversand963
Differential Revision: D40962311
Pulled By: ltamasi
fbshipit-source-id: 244bc9d172be1af2f204796b2f89104e4d2fa373
Summary:
This PR implements the originally disabled `Merge()` APIs when user-defined timestamp is enabled.
Simplest usage:
```cpp
// assume string append merge op is used with '.' as delimiter.
// ts1 < ts2
db->Put(WriteOptions(), "key", ts1, "v0");
db->Merge(WriteOptions(), "key", ts2, "1");
ReadOptions ro;
ro.timestamp = &ts2;
db->Get(ro, "key", &value);
ASSERT_EQ("v0.1", value);
```
Some code comments are added for clarity.
Note: support for timestamp in `DB::GetMergeOperands()` will be done in a follow-up PR.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10819
Test Plan: make check
Reviewed By: ltamasi
Differential Revision: D40603195
Pulled By: riversand963
fbshipit-source-id: f96d6f183258f3392d80377025529f7660503013
Summary:
Currently, a memtable's stats `num_deletes_` is incremented only if the entry is a regular delete (kTypeDeletion). We need to fix it by accounting for kTypeSingleDeletion and kTypeDeletionWithTimestamp.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10886
Test Plan: make check
Reviewed By: ltamasi
Differential Revision: D40740754
Pulled By: riversand963
fbshipit-source-id: 7bde62cd6df136585bc5bfb1c426c7a8276c08e1
Summary:
The for loop is marked as unreachable code because it will never call the increment. Switch it to `if`.
```
\table\merging_iterator.cc(823): error C2220: the following warning is treated as an error
\table\merging_iterator.cc(823): warning C4702: unreachable code
\table\merging_iterator.cc(1030): error C2220: the following warning is treated as an error
\table\merging_iterator.cc(1030): warning C4702: unreachable code
```
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10897
Reviewed By: cbi42
Differential Revision: D40811790
Pulled By: jay-zhuang
fbshipit-source-id: fe8fd3e7cf3d6f710360c402b79763854d5120df
Summary:
The PR fixes the handling of `Merge`s in `GetEntity`. Note that `Merge` is not yet
supported for wide-column entities written using `PutEntity`; this change is
about returning correct (i.e. consistent with `Get`) results in cases like when the
base value is a plain old key-value written using `Put` or when there is no real base
value because we hit either a tombstone or the beginning of history.
Implementation-wise, the patch introduces a new wrapper around the existing
`MergeHelper::TimedFullMerge` that can store the merge result in either a string
(for the purposes of `Get`) or a `PinnableWideColumns` instance (for `GetEntity`).
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10894
Test Plan: `make check`
Reviewed By: riversand963
Differential Revision: D40782708
Pulled By: ltamasi
fbshipit-source-id: 3d700d56b2ef81f02ba1e2d93f6481bf13abcc90
Summary:
Right now in MergingIterator, for each range tombstone start and end key, we pop one end from heap and push the other end into the heap. This involves extra downheap and upheap cost. In the likely cases when a range tombstone iterator emits relatively adjacent keys, these keys should have similar order within all keys in the heap. This can happen when there is a burst of consecutive range tombstones, and most of the keys covered by them are dropped already. This PR uses `replace_top()` when inserting new range tombstone keys, which is more efficient in these common cases.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10877
Test Plan:
- existing UT
- ran all flavors of stress test through sandcastle
- benchmark:
```
# Set up: --writes_per_range_tombstone=1 means one point write and one delete range
TEST_TMPDIR=/tmp/rocksdb-rangedel-test-all-tombstone ./db_bench --benchmarks=fillseq,levelstats --writes_per_range_tombstone=1 --max_num_range_tombstones=1000000 --range_tombstone_width=2 --num=100000000 --writes=800000 --max_bytes_for_level_base=4194304 --disable_auto_compactions --write_buffer_size=33554432 --key_size=64
Level Files Size(MB)
--------------------
0 8 152
1 0 0
2 0 0
3 0 0
4 0 0
5 0 0
6 0 0
# Benchmark
TEST_TMPDIR=/tmp/rocksdb-rangedel-test-all-tombstone/ ./db_bench --benchmarks=readseq[-W1][-X5],levelstats --use_existing_db=true --cache_size=3221225472 --num=100000000 --reads=1000000 --disable_auto_compactions=true --avoid_flush_during_recovery=true
# Pre PR
readseq [AVG 5 runs] : 1432116 (± 59664) ops/sec; 224.0 (± 9.3) MB/sec
readseq [MEDIAN 5 runs] : 1454886 ops/sec; 227.5 MB/sec
# Post PR
readseq [AVG 5 runs] : 1944425 (± 29521) ops/sec; 304.1 (± 4.6) MB/sec
readseq [MEDIAN 5 runs] : 1959430 ops/sec; 306.5 MB/sec
```
Reviewed By: ajkr
Differential Revision: D40710936
Pulled By: cbi42
fbshipit-source-id: cb782fb9cdcd26c0c3eb9443215a4ef4d2f79022
Summary:
Add some extra information in outputs of "sst_dump --command=raw" to help debug some issues. Right now, encoded block handle is printed out. It is more useful to directly print out offset and size.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10873
Test Plan: Manually run it against a file and check the output.
Reviewed By: anand1976
Differential Revision: D40742289
fbshipit-source-id: 04d7de26e7f27e1595a7cc3ac1c1082e4e835b93
Summary:
FragmentedRangeTombstoneList has a member variable `seq_set_` that contains the sequence numbers of all range tombstones in a set. The set is constructed in `FragmentTombstones()` and is used only in `FragmentedRangeTombstoneList::ContainsRange()` which only happens during compaction. This PR moves the initialization of `seq_set_` to `FragmentedRangeTombstoneList::ContainsRange()`. This should speed up `FragmentTombstones()` when the range tombstone list is used for read/scan requests. Microbench shows the speed improvement to be ~45%.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10848
Test Plan:
- Existing tests and stress test: `python3 tools/db_crashtest.py whitebox --simple --verify_iterator_with_expected_state_one_in=5`.
- Microbench: update `range_del_aggregator_bench` to benchmark speed of `FragmentTombstones()`:
```
./range_del_aggregator_bench --num_range_tombstones=1000 --tombstone_start_upper_bound=50000000 --num_runs=10000 --tombstone_width_mean=200 --should_deletes_per_run=100 --use_compaction_range_del_aggregator=true
Before this PR:
=========================
Fragment Tombstones: 270.286 us
AddTombstones: 1.28933 us
ShouldDelete (first): 0.525528 us
ShouldDelete (rest): 0.0797519 us
After this PR: time to fragment tombstones is pushed to AddTombstones() which only happen during compaction.
=========================
Fragment Tombstones: 149.879 us
AddTombstones: 102.131 us
ShouldDelete (first): 0.565871 us
ShouldDelete (rest): 0.0729444 us
```
- db_bench: this should improve speed for fragmenting range tombstones for mutable memtable:
```
./db_bench --benchmarks=readwhilewriting --writes_per_range_tombstone=100 --max_write_buffer_number=100 --min_write_buffer_number_to_merge=100 --writes=500000 --reads=250000 --disable_auto_compactions --max_num_range_tombstones=100000 --finish_after_writes --write_buffer_size=1073741824 --threads=25
Before this PR:
readwhilewriting : 18.301 micros/op 1310445 ops/sec 4.769 seconds 6250000 operations; 28.1 MB/s (41001 of 250000 found)
After this PR:
readwhilewriting : 16.943 micros/op 1439376 ops/sec 4.342 seconds 6250000 operations; 23.8 MB/s (28977 of 250000 found)
```
Reviewed By: ajkr
Differential Revision: D40646227
Pulled By: cbi42
fbshipit-source-id: ea471667edb258f67d01cfd828588e80a89e4083
Summary:
`#include "db/range_tombstone_fragmenter.h"` seems to break some internal test for 7.8 release. I'm removing it from sst_file_reader.h for now to unblock release. This should be fine as it is only used in a unit test for DeleteRange with timestamp. In addition, it does not seem to be useful to support delete range for sst file writer, since the range tombstone won't cover any key (its sequence number is 0). So maybe we can remove it in the future.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10847
Test Plan: CI.
Reviewed By: akankshamahajan15
Differential Revision: D40620865
Pulled By: cbi42
fbshipit-source-id: be44b2f31e062bff87ed1b8d94482c3f7eaa370c
Summary:
Refactor the classes, APIs and data structures for block cache tracing to allow a user provided trace writer to be used. Currently, only a TraceWriter is supported, with a default built-in implementation of FileTraceWriter. The TraceWriter, however, takes a flat trace record and is thus only suitable for file tracing. This PR introduces an abstract BlockCacheTraceWriter class that takes a structured BlockCacheTraceRecord. The BlockCacheTraceWriter implementation can then format and log the record in whatever way it sees fit. The default BlockCacheTraceWriterImpl does file tracing using a user provided TraceWriter.
`DB::StartBlockTrace` will internally redirect to changed `BlockCacheTrace::StartBlockCacheTrace`.
New API `DB::StartBlockTrace` is also added that directly takes `BlockCacheTraceWriter` pointer.
This same philosophy can be applied to KV and IO tracing as well.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10811
Test Plan:
existing unit tests
Old API DB::StartBlockTrace checked with db_bench tool
create database
```
./db_bench --benchmarks="fillseq" \
--key_size=20 --prefix_size=20 --keys_per_prefix=0 --value_size=100 \
--cache_index_and_filter_blocks --cache_size=1048576 \
--disable_auto_compactions=1 --disable_wal=1 --compression_type=none \
--min_level_to_compress=-1 --compression_ratio=1 --num=10000000
```
To trace block cache accesses when running readrandom benchmark:
```
./db_bench --benchmarks="readrandom" --use_existing_db --duration=60 \
--key_size=20 --prefix_size=20 --keys_per_prefix=0 --value_size=100 \
--cache_index_and_filter_blocks --cache_size=1048576 \
--disable_auto_compactions=1 --disable_wal=1 --compression_type=none \
--min_level_to_compress=-1 --compression_ratio=1 --num=10000000 \
--threads=16 \
-block_cache_trace_file="/tmp/binary_trace_test_example" \
-block_cache_trace_max_trace_file_size_in_bytes=1073741824 \
-block_cache_trace_sampling_frequency=1
```
Reviewed By: anand1976
Differential Revision: D40435289
Pulled By: akankshamahajan15
fbshipit-source-id: fa2755f4788185e19f4605e731641cfd21ab3282
Summary:
The motivations for this change include
* Free up space in ClockHandle so that we can add data for secondary cache handling while still keeping within single cache line (64 byte) size.
* This change frees up space by eliminating the need for the `hash` field by making the fixed-size key itself a hash, using a 128-bit bijective (lossless) hash.
* Generally more customizability of ShardedCache (such as hashing) without worrying about virtual call overheads
* ShardedCache now uses static polymorphism (template) instead of dynamic polymorphism (virtual overrides) for the CacheShard. No obvious performance benefit is seen from the change (as mostly expected; most calls to virtual functions in CacheShard could already be optimized to static calls), but offers more flexibility without incurring the runtime cost of adhering to a common interface (without type parameters or static callbacks).
* You'll also notice less `reinterpret_cast`ing and other boilerplate in the Cache implementations, as this can go in ShardedCache.
More detail:
* Don't have LRUCacheShard maintain `std::shared_ptr<SecondaryCache>` copies (extra refcount) when LRUCache can be in charge of keeping a `shared_ptr`.
* Renamed `capacity_mutex_` to `config_mutex_` to better represent the scope of what it guards.
* Some preparation for 64-bit hash and indexing in LRUCache, but didn't include the full change because of slight performance regression.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10801
Test Plan:
Unit test updates were non-trivial because of major changes to the ClockCacheShard interface in handling of key vs. hash.
Performance:
Create with `TEST_TMPDIR=/dev/shm ./db_bench -benchmarks=fillrandom -num=30000000 -disable_wal=1 -bloom_bits=16`
Test with
```
TEST_TMPDIR=/dev/shm ./db_bench -benchmarks=readrandom[-X1000] -readonly -num=30000000 -bloom_bits=16 -cache_index_and_filter_blocks=1 -cache_size=610000000 -duration 20 -threads=16
```
Before: `readrandom [AVG 150 runs] : 321147 (± 253) ops/sec`
After: `readrandom [AVG 150 runs] : 321530 (± 326) ops/sec`
So possibly ~0.1% improvement.
And with `-cache_type=hyper_clock_cache`:
Before: `readrandom [AVG 30 runs] : 614126 (± 7978) ops/sec`
After: `readrandom [AVG 30 runs] : 645349 (± 8087) ops/sec`
So roughly 5% improvement!
Reviewed By: anand1976
Differential Revision: D40252236
Pulled By: pdillinger
fbshipit-source-id: ff8fc70ef569585edc95bcbaaa0386f61355ae5b
Summary:
Instead of existing calls to ps from gnu_parallel, call a new wrapper that does ps, looks for unit test like processes, and uses pstack or gdb to print thread stack traces. Also, using `ps -wwf` instead of `ps -wf` ensures output is not cut off.
For security, CircleCI runs with security restrictions on ptrace (/proc/sys/kernel/yama/ptrace_scope = 1), and this change adds a work-around to `InstallStackTraceHandler()` (only used by testing tools) to allow any process from the same user to debug it. (I've also touched >100 files to ensure all the unit tests call this function.)
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10828
Test Plan: local manual + temporary infinite loop in a unit test to observe in CircleCI
Reviewed By: hx235
Differential Revision: D40447634
Pulled By: pdillinger
fbshipit-source-id: 718a4c4a5b54fa0f9af2d01a446162b45e5e84e1
Summary:
Add user-defined timestamp support for range deletion. The new API is `DeleteRange(opt, cf, begin_key, end_key, ts)`. Most of the change is to update the comparator to compare without timestamp. Other than that, major changes are
- internal range tombstone data structures (`FragmentedRangeTombstoneList`, `RangeTombstone`, etc.) to store timestamps.
- Garbage collection of range tombstones and range tombstone covered keys during compaction.
- Get()/MultiGet() to return the timestamp of a range tombstone when needed.
- Get/Iterator with range tombstones bounded by readoptions.timestamp.
- timestamp crash test now issues DeleteRange by default.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10661
Test Plan:
- Added unit test: `make check`
- Stress test: `python3 tools/db_crashtest.py --enable_ts whitebox --readpercent=57 --prefixpercent=4 --writepercent=25 -delpercent=5 --iterpercent=5 --delrangepercent=4`
- Ran `db_bench` to measure regression when timestamp is not enabled. The tests are for write (with some range deletion) and iterate with DB fitting in memory: `./db_bench--benchmarks=fillrandom,seekrandom --writes_per_range_tombstone=200 --max_write_buffer_number=100 --min_write_buffer_number_to_merge=100 --writes=500000 --reads=500000 --seek_nexts=10 --disable_auto_compactions -disable_wal=true --max_num_range_tombstones=1000`. Did not see consistent regression in no timestamp case.
| micros/op | fillrandom | seekrandom |
| --- | --- | --- |
|main| 2.58 |10.96|
|PR 10661| 2.68 |10.63|
Reviewed By: riversand963
Differential Revision: D39441192
Pulled By: cbi42
fbshipit-source-id: f05aca3c41605caf110daf0ff405919f300ddec2
Summary:
Try to align the compaction output file boundaries to the next level ones
(grandparent level), to reduce the level compaction write-amplification.
In level compaction, there are "wasted" data at the beginning and end of the
output level files. Align the file boundary can avoid such "wasted" compaction.
With this PR, it tries to align the non-bottommost level file boundaries to its
next level ones. It may cut file when the file size is large enough (at least
50% of target_file_size) and not too large (2x target_file_size).
db_bench shows about 12.56% compaction reduction:
```
TEST_TMPDIR=/data/dbbench2 ./db_bench --benchmarks=fillrandom,readrandom -max_background_jobs=12 -num=400000000 -target_file_size_base=33554432
# baseline:
Flush(GB): cumulative 25.882, interval 7.216
Cumulative compaction: 285.90 GB write, 162.36 MB/s write, 269.68 GB read, 153.15 MB/s read, 2926.7 seconds
# with this change:
Flush(GB): cumulative 25.882, interval 7.753
Cumulative compaction: 249.97 GB write, 141.96 MB/s write, 233.74 GB read, 132.74 MB/s read, 2534.9 seconds
```
The compaction simulator shows a similar result (14% with 100G random data).
As a side effect, with this PR, the SST file size can exceed the
target_file_size, but is capped at 2x target_file_size. And there will be
smaller files. Here are file size statistics when loading 100GB with the target
file size 32MB:
```
baseline this_PR
count 1.656000e+03 1.705000e+03
mean 3.116062e+07 3.028076e+07
std 7.145242e+06 8.046139e+06
```
The feature is enabled by default, to revert to the old behavior disable it
with `AdvancedColumnFamilyOptions.level_compaction_dynamic_file_size = false`
Also includes https://github.com/facebook/rocksdb/issues/1963 to cut file before skippable grandparent file. Which is for
use case like user adding 2 or more non-overlapping data range at the same
time, it can reduce the overlapping of 2 datasets in the lower levels.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10655
Reviewed By: cbi42
Differential Revision: D39552321
Pulled By: jay-zhuang
fbshipit-source-id: 640d15f159ab0cd973f2426cfc3af266fc8bdde2
Summary:
Currently, without this fix, DBImpl::GetLatestSequenceForKey() may not return the latest sequence number for merge operands of the key. This can cause conflict checking during optimistic transaction commit phase to fail. Fix it by always returning the latest sequence number of the key, also considering range tombstones.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10724
Test Plan: make check
Reviewed By: cbi42
Differential Revision: D39756847
Pulled By: riversand963
fbshipit-source-id: 0764c3dd4cb24960b37e18adccc6e7feed0e6876
