Summary:
The SeqnoToTimeMapping class (RocksDB internal) used by the preserve_internal_time_seconds / preclude_last_level_data_seconds options was essentially in a prototype state with some significant flaws that would risk biting us some day. This is a big, complicated change because both the implementation and the behavioral requirements of the class needed to be upgraded together. In short, this makes SeqnoToTimeMapping more internally responsible for maintaining good invariants, so that callers don't easily encounter dangerous scenarios.
* Some API functions were confusingly named and structured, so I fully refactored the APIs to use clear naming (e.g. `DecodeFrom` and `CopyFromSeqnoRange`), object states, function preconditions, etc.
* Previously the object could informally be sorted / compacted or not, and there was limited checking or enforcement on these states. Now there's a well-defined "enforced" state that is consistently checked in debug mode for applicable operations. (I attempted to create a separate "builder" class for unenforced states, but IIRC found that more cumbersome for existing uses than it was worth.)
* Previously operations would coalesce data in a way that was better for `GetProximalTimeBeforeSeqno` than for `GetProximalSeqnoBeforeTime` which is odd because the latter is the only one used by DB code currently (what is the seqno cut-off for data definitely older than this given time?). This is now reversed to consistently favor `GetProximalSeqnoBeforeTime`, with that logic concentrated in one place: `SeqnoToTimeMapping::SeqnoTimePair::Merge()`. Unfortunately, a lot of unit test logic was specifically testing the old, suboptimal behavior.
* Previously, the natural behavior of SeqnoToTimeMapping was to THROW AWAY data needed to get reasonable answers to the important `GetProximalSeqnoBeforeTime` queries. This is because SeqnoToTimeMapping only had a FIFO policy for staying within the entry capacity (except in aggregate+sort+serialize mode). If the DB wasn't extremely careful to avoid gathering too many time mappings, it could lose track of where the seqno cutoff was for cold data (`GetProximalSeqnoBeforeTime()` returning 0) and preventing all further data migration to the cold tier--until time passes etc. for mappings to catch up with FIFO purging of them. (The problem is not so acute because SST files contain relevant snapshots of the mappings, but the problem would apply to long-lived memtables.)
* Now the SeqnoToTimeMapping class has fully-integrated smarts for keeping a sufficiently complete history, within capacity limits, to give good answers to `GetProximalSeqnoBeforeTime` queries.
* Fixes old `// FIXME: be smarter about how we erase to avoid data falling off the front prematurely.`
* Fix an apparent bug in how entries are selected for storing into SST files. Previously, it only selected entries within the seqno range of the file, but that would easily leave a gap at the beginning of the timeline for data in the file for the purposes of answering GetProximalXXX queries with reasonable accuracy. This could probably lead to the same problem discussed above in naively throwing away entries in FIFO order in the old SeqnoToTimeMapping. The updated testing of GetProximalSeqnoBeforeTime in BasicSeqnoToTimeMapping relies on the fixed behavior.
* Fix a potential compaction CPU efficiency/scaling issue in which each compaction output file would iterate over and sort all seqno-to-time mappings from all compaction input files. Now we distill the input file entries to a constant size before processing each compaction output file.
Intended follow-up (me or others):
* Expand some direct testing of SeqnoToTimeMapping APIs. Here I've focused on updating existing tests to make sense.
* There are likely more gaps in availability of needed SeqnoToTimeMapping data when the DB shuts down and is restarted, at least with WAL.
* The data tracked in the DB could be kept more accurate and limited if it used the oldest seqno of unflushed data. This might require some more API refactoring.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/12253
Test Plan: unit tests updated
Reviewed By: jowlyzhang
Differential Revision: D52913733
Pulled By: pdillinger
fbshipit-source-id: 020737fcbbe6212f6701191a6ab86565054c9593
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:
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:
We have a lot of confusing code because of mixed, sometimes
completely opposite uses of of the term "raw block" or "raw contents",
sometimes within the same source file. For example, in `BlockBasedTableBuilder`,
`raw_block_contents` and `raw_size` generally referred to uncompressed block
contents and size, while `WriteRawBlock` referred to writing a block that
is already compressed if it is going to be. Meanwhile, in
`BlockBasedTable`, `raw_block_contents` either referred to a (maybe
compressed) block with trailer, or a maybe compressed block maybe
without trailer. (Note: left as follow-up work to use C++ typing to
better sort out the various kinds of BlockContents.)
This change primarily tries to apply some consistent terminology around
the kinds of block representations, avoiding the unclear "raw". (Any
meaning of "raw" assumes some bias toward the storage layer or toward
the logical data layer.) Preferred terminology:
* **Serialized block** - bytes that go into storage. For block-based table
(usually the case) this includes the block trailer. WART: block `size` may or
may not include the trailer; need to be clear about whether it does or not.
* **Maybe compressed block** - like a serialized block, but without the
trailer (or no promise of including a trailer). Must be accompanied by a
CompressionType.
* **Uncompressed block** - "payload" bytes that are either stored with no
compression, used as input to compression function, or result of
decompression function.
* **Parsed block** - an in-memory form of a block in block cache, as it is
used by the table reader. Different C++ types are used depending on the
block type (see block_like_traits.h).
Other refactorings:
* Misc corrections/improvements of internal API comments
* Remove a few misleading / unhelpful / redundant comments.
* Use move semantics in some places to simplify contracts
* Use better parameter names to indicate which parameters are used for
outputs
* Remove some extraneous `extern`
* Various clean-ups to `CacheDumperImpl` (mostly unnecessary code)
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10408
Test Plan: existing tests
Reviewed By: akankshamahajan15
Differential Revision: D38172617
Pulled By: pdillinger
fbshipit-source-id: ccb99299f324ac5ca46996d34c5089621a4f260c
Summary:
Which will be used for tiered storage to preclude hot data from
compacting to the cold tier (the last level).
Internally, adding seqno to time mapping. A periodic_task is scheduled
to record the current_seqno -> current_time in certain cadence. When
memtable flush, the mapping informaiton is stored in sstable property.
During compaction, the mapping information are merged and get the
approximate time of sequence number, which is used to determine if a key
is recently inserted or not and preclude it from the last level if it's
recently inserted (within the `preclude_last_level_data_seconds`).
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10338
Test Plan: CI
Reviewed By: siying
Differential Revision: D37810187
Pulled By: jay-zhuang
fbshipit-source-id: 6953be7a18a99de8b1cb3b162d712f79c2b4899f
Summary:
We have three related concepts:
* BlockType: an internal enum conceptually indicating a type of SST file
block
* CacheEntryRole: a user-facing enum for categorizing block cache entries,
which is also involved in associated cache entries with an appropriate
deleter. Can include categories for non-block cache entries (e.g. memory
reservations).
* TBlocklike: a C++ type for the actual type behind a void* cache entry.
We had some existing code ugliness because BlockType did not imply
TBlocklike, because of various kinds of "filter" block. This refactoring
fixes that with new BlockTypes.
More clean-up can come in later work.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10098
Test Plan: existing tests
Reviewed By: akankshamahajan15
Differential Revision: D36897945
Pulled By: pdillinger
fbshipit-source-id: 3ae496b5caa81e0a0ed85e873eb5b525e2d9a295
Summary:
When table_options.prepopulate_block_cache is set to
BlockBasedTableOptions::PrepopulateBlockCache::kFlushOnly and
table_options.partition_filters is also set true, then there is
segmentation failure when top level filter is fetched because its
entered with wrong type in cache.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9263
Test Plan:
Updated unit tests;
Ran db_stress: make crash_test -j32
Reviewed By: pdillinger
Differential Revision: D32936566
Pulled By: akankshamahajan15
fbshipit-source-id: 8bd79e53830d3e3c1bb79787e1ffbc3cb46d4426
Summary:
To prepare for adding checksum to footer and "context aware"
checksums. This also brings closely related code much closer together.
Recently added `BlockBasedTableBuilder::ComputeBlockTrailer` for testing
is made obsolete in the refactoring, as testing the checksums can happen
at a lower level of abstraction.
Also now checking for unrecognized checksum type on reading footer,
rather than later on use.
Also removed an obsolete function delcaration.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9113
Test Plan:
existing tests worked before refactoring to remove
`ComputeBlockTrailer`. And then refactored+improved tests using it.
Reviewed By: mrambacher
Differential Revision: D32090149
Pulled By: pdillinger
fbshipit-source-id: 2879da683c1498ea85a3b70dace9b6d9f6b47b6e
Summary:
XXH3 - latest hash function that is extremely fast on large
data, easily faster than crc32c on most any x86_64 hardware. In
integrating this hash function, I have handled the compression type byte
in a non-standard way to avoid using the streaming API (extra data
movement and active code size because of hash function complexity). This
approach got a thumbs-up from Yann Collet.
Existing functionality change:
* reject bad ChecksumType in options with InvalidArgument
This change split off from https://github.com/facebook/rocksdb/issues/9058 because context-aware checksum is
likely to be handled through different configuration than ChecksumType.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9069
Test Plan:
tests updated, and substantially expanded. Unit tests now check
that we don't accidentally change the values generated by the checksum
algorithms ("schema test") and that we properly handle
invalid/unrecognized checksum types in options or in file footer.
DBTestBase::ChangeOptions (etc.) updated from two to one configuration
changing from default CRC32c ChecksumType. The point of this test code
is to detect possible interactions among features, and the likelihood of
some bad interaction being detected by including configurations other
than XXH3 and CRC32c--and then not detected by stress/crash test--is
extremely low.
Stress/crash test also updated (manual run long enough to see it accepts
new checksum type). db_bench also updated for microbenchmarking
checksums.
### Performance microbenchmark (PORTABLE=0 DEBUG_LEVEL=0, Broadwell processor)
./db_bench -benchmarks=crc32c,xxhash,xxhash64,xxh3,crc32c,xxhash,xxhash64,xxh3,crc32c,xxhash,xxhash64,xxh3
crc32c : 0.200 micros/op 5005220 ops/sec; 19551.6 MB/s (4096 per op)
xxhash : 0.807 micros/op 1238408 ops/sec; 4837.5 MB/s (4096 per op)
xxhash64 : 0.421 micros/op 2376514 ops/sec; 9283.3 MB/s (4096 per op)
xxh3 : 0.171 micros/op 5858391 ops/sec; 22884.3 MB/s (4096 per op)
crc32c : 0.206 micros/op 4859566 ops/sec; 18982.7 MB/s (4096 per op)
xxhash : 0.793 micros/op 1260850 ops/sec; 4925.2 MB/s (4096 per op)
xxhash64 : 0.410 micros/op 2439182 ops/sec; 9528.1 MB/s (4096 per op)
xxh3 : 0.161 micros/op 6202872 ops/sec; 24230.0 MB/s (4096 per op)
crc32c : 0.203 micros/op 4924686 ops/sec; 19237.1 MB/s (4096 per op)
xxhash : 0.839 micros/op 1192388 ops/sec; 4657.8 MB/s (4096 per op)
xxhash64 : 0.424 micros/op 2357391 ops/sec; 9208.6 MB/s (4096 per op)
xxh3 : 0.162 micros/op 6182678 ops/sec; 24151.1 MB/s (4096 per op)
As you can see, especially once warmed up, xxh3 is fastest.
### Performance macrobenchmark (PORTABLE=0 DEBUG_LEVEL=0, Broadwell processor)
Test
for I in `seq 1 50`; do for CHK in 0 1 2 3 4; do TEST_TMPDIR=/dev/shm/rocksdb$CHK ./db_bench -benchmarks=fillseq -memtablerep=vector -allow_concurrent_memtable_write=false -num=30000000 -checksum_type=$CHK 2>&1 | grep 'micros/op' | tee -a results-$CHK & done; wait; done
Results (ops/sec)
for FILE in results*; do echo -n "$FILE "; awk '{ s += $5; c++; } END { print 1.0 * s / c; }' < $FILE; done
results-0 252118 # kNoChecksum
results-1 251588 # kCRC32c
results-2 251863 # kxxHash
results-3 252016 # kxxHash64
results-4 252038 # kXXH3
Reviewed By: mrambacher
Differential Revision: D31905249
Pulled By: pdillinger
fbshipit-source-id: cb9b998ebe2523fc7c400eedf62124a78bf4b4d1
Summary:
Insert warm blocks (data, uncompressed dict, index and filter blocks) during flush in Block cache which is enabled under option BlockBasedTableOptions.prepopulate_block_cache.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/8561
Test Plan: Added unit test
Reviewed By: anand1976
Differential Revision: D29773411
Pulled By: akankshamahajan15
fbshipit-source-id: 6631123c10134340ef0bd7e90baafaa6deba0e66
Summary:
This PR prepopulates warm/hot data blocks which are already in memory
into block cache at the time of flush. On a flush, the data block that is
in memory (in memtables) get flushed to the device. If using Direct IO,
additional IO is incurred to read this data back into memory again, which
is avoided by enabling newly added option.
Right now, this is enabled only for flush for data blocks. We plan to
expand this option to cover compactions in the future and for other types
of blocks.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/8242
Test Plan: Add new unit test
Reviewed By: anand1976
Differential Revision: D28521703
Pulled By: akankshamahajan15
fbshipit-source-id: 7219d6958821cedce689a219c3963a6f1a9d5f05
Summary:
Greatly reduced the not-quite-copy-paste giant parameter lists
of rocksdb::NewTableBuilder, rocksdb::BuildTable,
BlockBasedTableBuilder::Rep ctor, and BlockBasedTableBuilder ctor.
Moved weird separate parameter `uint32_t column_family_id` of
TableFactory::NewTableBuilder into TableBuilderOptions.
Re-ordered parameters to TableBuilderOptions ctor, so that `uint64_t
target_file_size` is not randomly placed between uint64_t timestamps
(was easy to mix up).
Replaced a couple of fields of BlockBasedTableBuilder::Rep with a
FilterBuildingContext. The motivation for this change is making it
easier to pass along more data into new fields in FilterBuildingContext
(follow-up PR).
Pull Request resolved: https://github.com/facebook/rocksdb/pull/8240
Test Plan: ASAN make check
Reviewed By: mrambacher
Differential Revision: D28075891
Pulled By: pdillinger
fbshipit-source-id: fddb3dbb8260a0e8bdcbb51b877ebabf9a690d4f
Summary:
Previously it only applied to block-based tables generated by flush. This restriction
was undocumented and blocked a new use case. Now compression sampling
applies to all block-based tables we generate when it is enabled.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/8105
Test Plan: new unit test
Reviewed By: riversand963
Differential Revision: D27317275
Pulled By: ajkr
fbshipit-source-id: cd9fcc5178d6515e8cb59c6facb5ac01893cb5b0
Summary:
For dictionary compression, we need to collect some representative samples of the data to be compressed, which we use to either generate or train (when `CompressionOptions::zstd_max_train_bytes > 0`) a dictionary. Previously, the strategy was to buffer all the data blocks during flush, and up to the target file size during compaction. That strategy allowed us to randomly pick samples from as wide a range as possible that'd be guaranteed to land in a single output file.
However, some users try to make huge files in memory-constrained environments, where this strategy can cause OOM. This PR introduces an option, `CompressionOptions::max_dict_buffer_bytes`, that limits how much data blocks are buffered before we switch to unbuffered mode (which means creating the per-SST dictionary, writing out the buffered data, and compressing/writing new blocks as soon as they are built). It is not strict as we currently buffer more than just data blocks -- also keys are buffered. But it does make a step towards giving users predictable memory usage.
Related changes include:
- Changed sampling for dictionary compression to select unique data blocks when there is limited availability of data blocks
- Made use of `BlockBuilder::SwapAndReset()` to save an allocation+memcpy when buffering data blocks for building a dictionary
- Changed `ParseBoolean()` to accept an input containing characters after the boolean. This is necessary since, with this PR, a value for `CompressionOptions::enabled` is no longer necessarily the final component in the `CompressionOptions` string.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7970
Test Plan:
- updated `CompressionOptions` unit tests to verify limit is respected (to the extent expected in the current implementation) in various scenarios of flush/compaction to bottommost/non-bottommost level
- looked at jemalloc heap profiles right before and after switching to unbuffered mode during flush/compaction. Verified memory usage in buffering is proportional to the limit set.
Reviewed By: pdillinger
Differential Revision: D26467994
Pulled By: ajkr
fbshipit-source-id: 3da4ef9fba59974e4ef40e40c01611002c861465
Summary:
This commit makes https://github.com/facebook/rocksdb/issues/6262's code change tidier and easier to understand by:
1. Wrapping parallel compression initialization and termination into
common methods;
2. Wrapping BlockRep initialization, push/pop into common methods;
3. Wrapping file size estimation into common methods;
4. Fixing function declarations that use non-const reference;
5. Fixing some uninitialized variables;
6. Fixing first_block data race;
7. Making BlockRep::status check in BlockBasedTableBuilder::Finish only present
if ok();
8. Making assert(ok()) in BlockBasedTableBuilder::CompressAndVerifyBlock only
present in non-parallel compression mode. In parallel compression mode,
compression will abort if status is not OK;
9. Eliminating potential data race caused by BlockBasedTableBuilder::GetStatus()
and BlockBasedTableBuilder::GetIOStatus() by returning status copy instead of
unprotected reference.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6888
Reviewed By: ajkr
Differential Revision: D21957110
Pulled By: jay-zhuang
fbshipit-source-id: 3a29892f249209513f030349756cecd7736eae80
Summary:
`db_id` and `db_session_id` are now part of the table properties for all formats and stored in SST files. This adds about 99 bytes to each new SST file.
The `TablePropertiesNames` for these two identifiers are `rocksdb.creating.db.identity` and `rocksdb.creating.session.identity`.
In addition, SST files generated from SstFileWriter and Repairer have DB identity “SST Writer” and “DB Repairer”, respectively. Their DB session IDs are generated in the same way as `DB::GetDbSessionId`.
A table property test is added.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6983
Test Plan: make check and some manual tests.
Reviewed By: zhichao-cao
Differential Revision: D22048826
Pulled By: gg814
fbshipit-source-id: afdf8c11424a6f509b5c0b06dafad584a80103c9
Summary:
There is no systematic mechanism to prevent BlockBasedTableBuilder's status to be set from non-OK to OK. Adding a mechanism to force this will help us prevent failures in the future.
The solution is to only make it possible to set the status code if the status code to set is not OK.
Since the status code passed to CompressAndVerifyBlock() is changed, a mini refactoring is done too so that the output arguments are changed from reference to pointers, based on Google C++ Style.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6776
Test Plan: Run all existing test.
Reviewed By: pdillinger
Differential Revision: D21314382
fbshipit-source-id: 27000c10f1e4c121661e026548d6882066409375
Summary:
With https://github.com/facebook/rocksdb/issues/6262, TSAN complains about data race of some variables. Those variables are used to estimate file size and are accessed in writer and background threads. Since file size estimation doesn't have to be 100% accurate, we make some variables atomic and use relaxed memory order.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6636
Test Plan: Run all tests with TSAN.
Reviewed By: anand1976
Differential Revision: D20820635
fbshipit-source-id: 1ea45ff38be15e33674ffe06b7d42fc9fe161ea5
Summary:
This PR adds support for pipelined & parallel compression optimization for `BlockBasedTableBuilder`. This optimization makes block building, block compression and block appending a pipeline, and uses multiple threads to accelerate block compression. Users can set `CompressionOptions::parallel_threads` greater than 1 to enable compression parallelism.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6262
Reviewed By: ajkr
Differential Revision: D20651306
fbshipit-source-id: 62125590a9c15b6d9071def9dc72589c1696a4cb
Summary:
In the current implementation, sst file checksum is calculated by a shared checksum function object, which may make some checksum function hard to be applied here such as SHA1. In this implementation, each sst file will have its own checksum generator obejct, created by FileChecksumGenFactory. User needs to implement its own FilechecksumGenerator and Factory to plugin the in checksum calculation method.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6600
Test Plan: tested with make asan_check
Reviewed By: riversand963
Differential Revision: D20717670
Pulled By: zhichao-cao
fbshipit-source-id: 2a74c1c280ac11a07a1980185b43b671acaa71c6
Summary:
In the current code base, we use Status to get and store the returned status from the call. Specifically, for IO related functions, the current Status cannot reflect the IO Error details such as error scope, error retryable attribute, and others. With the implementation of https://github.com/facebook/rocksdb/issues/5761, we have the new Wrapper for IO, which returns IOStatus instead of Status. However, the IOStatus is purged at the lower level of write path and transferred to Status.
The first job of this PR is to pass the IOStatus to the write path (flush, WAL write, and Compaction). The second job is to identify the Retryable IO Error as HardError, and set the bg_error_ as HardError. In this case, the DB Instance becomes read only. User is informed of the Status and need to take actions to deal with it (e.g., call db->Resume()).
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6487
Test Plan: Added the testing case to error_handler_fs_test. Pass make asan_check
Reviewed By: anand1976
Differential Revision: D20685017
Pulled By: zhichao-cao
fbshipit-source-id: ff85f042896243abcd6ef37877834e26f36b6eb0
Summary:
When dynamically linking two binaries together, different builds of RocksDB from two sources might cause errors. To provide a tool for user to solve the problem, the RocksDB namespace is changed to a flag which can be overridden in build time.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6433
Test Plan: Build release, all and jtest. Try to build with ROCKSDB_NAMESPACE with another flag.
Differential Revision: D19977691
fbshipit-source-id: aa7f2d0972e1c31d75339ac48478f34f6cfcfb3e
Summary:
In the current code base, RocksDB generate the checksum for each block and verify the checksum at usage. Current PR enable SST file checksum. After a SST file is generated by Flush or Compaction, RocksDB generate the SST file checksum and store the checksum value and checksum method name in the vs_info and MANIFEST as part for the FileMetadata.
Added the enable_sst_file_checksum to Options to enable or disable file checksum. Added sst_file_checksum to Options such that user can plugin their own SST file checksum calculate method via overriding the SstFileChecksum class. The checksum information inlcuding uint32_t checksum value and a checksum name (string). A new tool is added to LDB such that user can dump out a list of file checksum information from MANIFEST. If user enables the file checksum but does not provide the sst_file_checksum instance, RocksDB will use the default crc32checksum implemented in table/sst_file_checksum_crc32c.h
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6216
Test Plan: Added the testing case in table_test and ldb_cmd_test to verify checksum is correct in different level. Pass make asan_check.
Differential Revision: D19171461
Pulled By: zhichao-cao
fbshipit-source-id: b2e53479eefc5bb0437189eaa1941670e5ba8b87
Summary:
This change enables custom implementations of FilterPolicy to
wrap a variety of NewBloomFilterPolicy and select among them based on
contextual information such as table level and compaction style.
* Moves FilterBuildingContext to public API and elaborates it with more
useful data. (It would be nice to put more general options-like data,
but at the time this object is constructed, we are using internal APIs
ImmutableCFOptions and MutableCFOptions and don't have easy access to
ColumnFamilyOptions that I can tell.)
* Renames BloomFilterPolicy::GetFilterBitsBuilderInternal to
GetBuilderWithContext, because it's now public.
* Plumbs through the table's "level_at_creation" for filter building
context.
* Simplified some tests by adding GetBuilder() to
MockBlockBasedTableTester.
* Adds test as DBBloomFilterTest.ContextCustomFilterPolicy, including
sample wrapper class LevelAndStyleCustomFilterPolicy.
* Fixes a cross-test bug in DBBloomFilterTest.OptimizeFiltersForHits
where it does not reset perf context.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6088
Test Plan: make check, valgrind on db_bloom_filter_test
Differential Revision: D18697817
Pulled By: pdillinger
fbshipit-source-id: 5f987a2d7b07cc7a33670bc08ca6b4ca698c1cf4
Summary:
Use delete to disable automatic generated methods instead of private, and put the constructor together for more clear.This modification cause the unused field warning, so add unused attribute to disable this warning.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/5009
Differential Revision: D17288733
fbshipit-source-id: 8a767ce096f185f1db01bd28fc88fef1cdd921f3
