rocksdb/db/flush_job_test.cc

822 lines
31 KiB
C++
Raw Normal View History

// Copyright (c) 2011-present, Facebook, Inc. All rights reserved.
// This source code is licensed under both the GPLv2 (found in the
// COPYING file in the root directory) and Apache 2.0 License
// (found in the LICENSE.Apache file in the root directory).
#include "db/flush_job.h"
#include <algorithm>
#include <array>
#include <map>
#include <string>
#include "db/blob/blob_index.h"
#include "db/column_family.h"
#include "db/db_impl/db_impl.h"
#include "db/version_set.h"
#include "file/writable_file_writer.h"
#include "rocksdb/cache.h"
#include "rocksdb/file_system.h"
#include "rocksdb/write_buffer_manager.h"
#include "table/mock_table.h"
#include "test_util/testharness.h"
#include "test_util/testutil.h"
#include "util/random.h"
#include "util/string_util.h"
namespace ROCKSDB_NAMESPACE {
// TODO(icanadi) Mock out everything else:
// 1. VersionSet
// 2. Memtable
class FlushJobTestBase : public testing::Test {
protected:
FlushJobTestBase(std::string dbname, const Comparator* ucmp)
: env_(Env::Default()),
Create a CustomEnv class; Add WinFileSystem; Make LegacyFileSystemWrapper private (#7703) Summary: This PR does the following: -> Creates a WinFileSystem class. This class is the Windows equivalent of the PosixFileSystem and will be used on Windows systems. -> Introduces a CustomEnv class. A CustomEnv is an Env that takes a FileSystem as constructor argument. I believe there will only ever be two implementations of this class (PosixEnv and WinEnv). There is still a CustomEnvWrapper class that takes an Env and a FileSystem and wraps the Env calls with the input Env but uses the FileSystem for the FileSystem calls -> Eliminates the public uses of the LegacyFileSystemWrapper. With this change in place, there are effectively the following patterns of Env: - "Base Env classes" (PosixEnv, WinEnv). These classes implement the core Env functions (e.g. Threads) and have a hard-coded input FileSystem. These classes inherit from CompositeEnv, implement the core Env functions (threads) and delegate the FileSystem-like calls to the input file system. - Wrapped Composite Env classes (MemEnv). These classes take in an Env and a FileSystem. The core env functions are re-directed to the wrapped env. The file system calls are redirected to the input file system - Legacy Wrapped Env classes. These classes take in an Env input (but no FileSystem). The core env functions are re-directed to the wrapped env. A "Legacy File System" is created using this env and the file system calls directed to the env itself. With these changes in place, the PosixEnv becomes a singleton -- there is only ever one created. Any other use of the PosixEnv is via another wrapped env. This cleans up some of the issues with the env construction and destruction. Additionally, there were places in the code that required had an Env when they required a FileSystem. Many of these places would wrap the Env with a LegacyFileSystemWrapper instead of using the env->GetFileSystem(). These places were changed, thereby removing layers of additional redirection (LegacyFileSystem --> Env --> Env::FileSystem). Pull Request resolved: https://github.com/facebook/rocksdb/pull/7703 Reviewed By: zhichao-cao Differential Revision: D25762190 Pulled By: anand1976 fbshipit-source-id: 1a088e97fc916f28ac69c149cd1dcad0ab31704b
2021-01-06 18:48:24 +00:00
fs_(env_->GetFileSystem()),
dbname_(std::move(dbname)),
ucmp_(ucmp),
options_(),
db_options_(options_),
column_family_names_({kDefaultColumnFamilyName, "foo", "bar"}),
table_cache_(NewLRUCache(50000, 16)),
write_buffer_manager_(db_options_.db_write_buffer_size),
shutting_down_(false),
mock_table_factory_(new mock::MockTableFactory()) {}
virtual ~FlushJobTestBase() {
if (getenv("KEEP_DB")) {
fprintf(stdout, "db is still in %s\n", dbname_.c_str());
} else {
// destroy versions_ to release all file handles
versions_.reset();
EXPECT_OK(DestroyDir(env_, dbname_));
}
}
void NewDB() {
ASSERT_OK(SetIdentityFile(env_, dbname_));
VersionEdit new_db;
new_db.SetLogNumber(0);
new_db.SetNextFile(2);
new_db.SetLastSequence(0);
autovector<VersionEdit> new_cfs;
SequenceNumber last_seq = 1;
uint32_t cf_id = 1;
for (size_t i = 1; i != column_family_names_.size(); ++i) {
VersionEdit new_cf;
new_cf.AddColumnFamily(column_family_names_[i]);
new_cf.SetColumnFamily(cf_id++);
new_cf.SetComparatorName(ucmp_->Name());
Support switching on / off UDT together with in-Memtable-only feature (#11623) Summary: Add support to allow enabling / disabling user-defined timestamps feature for an existing column family in combination with the in-Memtable only feature. To do this, this PR includes: 1) Log the `persist_user_defined_timestamps` option per column family in Manifest to facilitate detecting an attempt to enable / disable UDT. This entry is enforced to be logged in the same VersionEdit as the user comparator name entry. 2) User-defined timestamps related options are validated when re-opening a column family, including user comparator name and the `persist_user_defined_timestamps` flag. These type of settings and settings change are considered valid: a) no user comparator change and no effective `persist_user_defined_timestamp` flag change. b) switch user comparator to enable UDT provided the immediately effective `persist_user_defined_timestamps` flag is false. c) switch user comparator to disable UDT provided that the before-change `persist_user_defined_timestamps` is already false. 3) when an attempt to enable UDT is detected, we mark all its existing SST files as "having no UDT" by marking its `FileMetaData.user_defined_timestamps_persisted` flag to false and handle their file boundaries `FileMetaData.smallest`, `FileMetaData.largest` by padding a min timestamp. 4) while enabling / disabling UDT feature, timestamp size inconsistency in existing WAL logs are handled to make it compatible with the running user comparator. Pull Request resolved: https://github.com/facebook/rocksdb/pull/11623 Test Plan: ``` make all check ./db_with_timestamp_basic_test --gtest-filter="*EnableDisableUDT*" ./db_wal_test --gtest_filter="*EnableDisableUDT*" ``` Reviewed By: ltamasi Differential Revision: D47636862 Pulled By: jowlyzhang fbshipit-source-id: dcd19f67292da3c3cc9584c09ad00331c9ab9322
2023-07-27 03:16:32 +00:00
new_cf.SetPersistUserDefinedTimestamps(persist_udt_);
new_cf.SetLogNumber(0);
new_cf.SetNextFile(2);
new_cf.SetLastSequence(last_seq++);
new_cfs.emplace_back(new_cf);
}
const std::string manifest = DescriptorFileName(dbname_, 1);
const auto& fs = env_->GetFileSystem();
std::unique_ptr<WritableFileWriter> file_writer;
Status s = WritableFileWriter::Create(
fs, manifest, fs->OptimizeForManifestWrite(env_options_), &file_writer,
nullptr);
ASSERT_OK(s);
{
log::Writer log(std::move(file_writer), 0, false);
std::string record;
new_db.EncodeTo(&record);
s = log.AddRecord(record);
ASSERT_OK(s);
for (const auto& e : new_cfs) {
record.clear();
e.EncodeTo(&record);
s = log.AddRecord(record);
ASSERT_OK(s);
}
}
ASSERT_OK(s);
// Make "CURRENT" file that points to the new manifest file.
s = SetCurrentFile(fs_.get(), dbname_, 1, nullptr);
ASSERT_OK(s);
}
void SetUp() override {
EXPECT_OK(env_->CreateDirIfMissing(dbname_));
// TODO(icanadi) Remove this once we mock out VersionSet
NewDB();
db_options_.env = env_;
db_options_.fs = fs_;
db_options_.db_paths.emplace_back(dbname_,
std::numeric_limits<uint64_t>::max());
db_options_.statistics = CreateDBStatistics();
cf_options_.comparator = ucmp_;
Logically strip timestamp during flush (#11557) 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
2023-06-29 22:50:50 +00:00
cf_options_.persist_user_defined_timestamps = persist_udt_;
cf_options_.paranoid_file_checks = paranoid_file_checks_;
std::vector<ColumnFamilyDescriptor> column_families;
cf_options_.table_factory = mock_table_factory_;
for (const auto& cf_name : column_family_names_) {
column_families.emplace_back(cf_name, cf_options_);
}
versions_.reset(
new VersionSet(dbname_, &db_options_, env_options_, table_cache_.get(),
&write_buffer_manager_, &write_controller_,
/*block_cache_tracer=*/nullptr, /*io_tracer=*/nullptr,
/*db_id*/ "", /*db_session_id*/ ""));
EXPECT_OK(versions_->Recover(column_families, false));
}
Env* env_;
Introduce a new storage specific Env API (#5761) Summary: The current Env API encompasses both storage/file operations, as well as OS related operations. Most of the APIs return a Status, which does not have enough metadata about an error, such as whether its retry-able or not, scope (i.e fault domain) of the error etc., that may be required in order to properly handle a storage error. The file APIs also do not provide enough control over the IO SLA, such as timeout, prioritization, hinting about placement and redundancy etc. This PR separates out the file/storage APIs from Env into a new FileSystem class. The APIs are updated to return an IOStatus with metadata about the error, as well as to take an IOOptions structure as input in order to allow more control over the IO. The user can set both ```options.env``` and ```options.file_system``` to specify that RocksDB should use the former for OS related operations and the latter for storage operations. Internally, a ```CompositeEnvWrapper``` has been introduced that inherits from ```Env``` and redirects individual methods to either an ```Env``` implementation or the ```FileSystem``` as appropriate. When options are sanitized during ```DB::Open```, ```options.env``` is replaced with a newly allocated ```CompositeEnvWrapper``` instance if both env and file_system have been specified. This way, the rest of the RocksDB code can continue to function as before. This PR also ports PosixEnv to the new API by splitting it into two - PosixEnv and PosixFileSystem. PosixEnv is defined as a sub-class of CompositeEnvWrapper, and threading/time functions are overridden with Posix specific implementations in order to avoid an extra level of indirection. The ```CompositeEnvWrapper``` translates ```IOStatus``` return code to ```Status```, and sets the severity to ```kSoftError``` if the io_status is retryable. The error handling code in RocksDB can then recover the DB automatically. Pull Request resolved: https://github.com/facebook/rocksdb/pull/5761 Differential Revision: D18868376 Pulled By: anand1976 fbshipit-source-id: 39efe18a162ea746fabac6360ff529baba48486f
2019-12-13 22:47:08 +00:00
std::shared_ptr<FileSystem> fs_;
std::string dbname_;
const Comparator* const ucmp_;
EnvOptions env_options_;
Options options_;
ImmutableDBOptions db_options_;
const std::vector<std::string> column_family_names_;
std::shared_ptr<Cache> table_cache_;
WriteController write_controller_;
WriteBufferManager write_buffer_manager_;
ColumnFamilyOptions cf_options_;
std::unique_ptr<VersionSet> versions_;
InstrumentedMutex mutex_;
std::atomic<bool> shutting_down_;
std::shared_ptr<mock::MockTableFactory> mock_table_factory_;
Logically strip timestamp during flush (#11557) 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
2023-06-29 22:50:50 +00:00
bool persist_udt_ = true;
bool paranoid_file_checks_ = false;
SeqnoToTimeMapping empty_seqno_to_time_mapping_;
};
class FlushJobTest : public FlushJobTestBase {
public:
FlushJobTest()
: FlushJobTestBase(test::PerThreadDBPath("flush_job_test"),
BytewiseComparator()) {}
};
rocksdb: switch to gtest Summary: Our existing test notation is very similar to what is used in gtest. It makes it easy to adopt what is different. In this diff I modify existing [[ https://code.google.com/p/googletest/wiki/Primer#Test_Fixtures:_Using_the_Same_Data_Configuration_for_Multiple_Te | test fixture ]] classes to inherit from `testing::Test`. Also for unit tests that use fixture class, `TEST` is replaced with `TEST_F` as required in gtest. There are several custom `main` functions in our existing tests. To make this transition easier, I modify all `main` functions to fallow gtest notation. But eventually we can remove them and use implementation of `main` that gtest provides. ```lang=bash % cat ~/transform #!/bin/sh files=$(git ls-files '*test\.cc') for file in $files do if grep -q "rocksdb::test::RunAllTests()" $file then if grep -Eq '^class \w+Test {' $file then perl -pi -e 's/^(class \w+Test) {/${1}: public testing::Test {/g' $file perl -pi -e 's/^(TEST)/${1}_F/g' $file fi perl -pi -e 's/(int main.*\{)/${1}::testing::InitGoogleTest(&argc, argv);/g' $file perl -pi -e 's/rocksdb::test::RunAllTests/RUN_ALL_TESTS/g' $file fi done % sh ~/transform % make format ``` Second iteration of this diff contains only scripted changes. Third iteration contains manual changes to fix last errors and make it compilable. Test Plan: Build and notice no errors. ```lang=bash % USE_CLANG=1 make check -j55 ``` Tests are still testing. Reviewers: meyering, sdong, rven, igor Reviewed By: igor Subscribers: dhruba, leveldb Differential Revision: https://reviews.facebook.net/D35157
2015-03-17 21:08:00 +00:00
TEST_F(FlushJobTest, Empty) {
JobContext job_context(0);
auto cfd = versions_->GetColumnFamilySet()->GetDefault();
EventLogger event_logger(db_options_.info_log.get());
SnapshotChecker* snapshot_checker = nullptr; // not relavant
FlushJob flush_job(
dbname_, versions_->GetColumnFamilySet()->GetDefault(), db_options_,
*cfd->GetLatestMutableCFOptions(),
std::numeric_limits<uint64_t>::max() /* memtable_id */, env_options_,
versions_.get(), &mutex_, &shutting_down_, {}, kMaxSequenceNumber,
snapshot_checker, &job_context, FlushReason::kTest, nullptr, nullptr,
nullptr, kNoCompression, nullptr, &event_logger, false,
true /* sync_output_directory */, true /* write_manifest */,
Env::Priority::USER, nullptr /*IOTracer*/, empty_seqno_to_time_mapping_);
{
InstrumentedMutexLock l(&mutex_);
flush_job.PickMemTable();
ASSERT_OK(flush_job.Run());
}
2014-11-15 00:57:17 +00:00
job_context.Clean();
}
rocksdb: switch to gtest Summary: Our existing test notation is very similar to what is used in gtest. It makes it easy to adopt what is different. In this diff I modify existing [[ https://code.google.com/p/googletest/wiki/Primer#Test_Fixtures:_Using_the_Same_Data_Configuration_for_Multiple_Te | test fixture ]] classes to inherit from `testing::Test`. Also for unit tests that use fixture class, `TEST` is replaced with `TEST_F` as required in gtest. There are several custom `main` functions in our existing tests. To make this transition easier, I modify all `main` functions to fallow gtest notation. But eventually we can remove them and use implementation of `main` that gtest provides. ```lang=bash % cat ~/transform #!/bin/sh files=$(git ls-files '*test\.cc') for file in $files do if grep -q "rocksdb::test::RunAllTests()" $file then if grep -Eq '^class \w+Test {' $file then perl -pi -e 's/^(class \w+Test) {/${1}: public testing::Test {/g' $file perl -pi -e 's/^(TEST)/${1}_F/g' $file fi perl -pi -e 's/(int main.*\{)/${1}::testing::InitGoogleTest(&argc, argv);/g' $file perl -pi -e 's/rocksdb::test::RunAllTests/RUN_ALL_TESTS/g' $file fi done % sh ~/transform % make format ``` Second iteration of this diff contains only scripted changes. Third iteration contains manual changes to fix last errors and make it compilable. Test Plan: Build and notice no errors. ```lang=bash % USE_CLANG=1 make check -j55 ``` Tests are still testing. Reviewers: meyering, sdong, rven, igor Reviewed By: igor Subscribers: dhruba, leveldb Differential Revision: https://reviews.facebook.net/D35157
2015-03-17 21:08:00 +00:00
TEST_F(FlushJobTest, NonEmpty) {
JobContext job_context(0);
auto cfd = versions_->GetColumnFamilySet()->GetDefault();
auto new_mem = cfd->ConstructNewMemtable(*cfd->GetLatestMutableCFOptions(),
kMaxSequenceNumber);
new_mem->Ref();
auto inserted_keys = mock::MakeMockFile();
// Test data:
// seqno [ 1, 2 ... 8998, 8999, 9000, 9001, 9002 ... 9999 ]
// key [ 1001, 1002 ... 9998, 9999, 0, 1, 2 ... 999 ]
// range-delete "9995" -> "9999" at seqno 10000
// blob references with seqnos 10001..10006
for (int i = 1; i < 10000; ++i) {
std::string key(std::to_string((i + 1000) % 10000));
std::string value("value" + key);
Integrity protection for live updates to WriteBatch (#7748) Summary: This PR adds the foundation classes for key-value integrity protection and the first use case: protecting live updates from the source buffers added to `WriteBatch` through the destination buffer in `MemTable`. The width of the protection info is not yet configurable -- only eight bytes per key is supported. This PR allows users to enable protection by constructing `WriteBatch` with `protection_bytes_per_key == 8`. It does not yet expose a way for users to get integrity protection via other write APIs (e.g., `Put()`, `Merge()`, `Delete()`, etc.). The foundation classes (`ProtectionInfo.*`) embed the coverage info in their type, and provide `Protect.*()` and `Strip.*()` functions to navigate between types with different coverage. For making bytes per key configurable (for powers of two up to eight) in the future, these classes are templated on the unsigned integer type used to store the protection info. That integer contains the XOR'd result of hashes with independent seeds for all covered fields. For integer fields, the hash is computed on the raw unadjusted bytes, so the result is endian-dependent. The most significant bytes are truncated when the hash value (8 bytes) is wider than the protection integer. When `WriteBatch` is constructed with `protection_bytes_per_key == 8`, we hold a `ProtectionInfoKVOTC` (i.e., one that covers key, value, optype aka `ValueType`, timestamp, and CF ID) for each entry added to the batch. The protection info is generated from the original buffers passed by the user, as well as the original metadata generated internally. When writing to memtable, each entry is transformed to a `ProtectionInfoKVOTS` (i.e., dropping coverage of CF ID and adding coverage of sequence number), since at that point we know the sequence number, and have already selected a memtable corresponding to a particular CF. This protection info is verified once the entry is encoded in the `MemTable` buffer. Pull Request resolved: https://github.com/facebook/rocksdb/pull/7748 Test Plan: - an integration test to verify a wide variety of single-byte changes to the encoded `MemTable` buffer are caught - add to stress/crash test to verify it works in variety of configs/operations without intentional corruption - [deferred] unit tests for `ProtectionInfo.*` classes for edge cases like KV swap, `SliceParts` and `Slice` APIs are interchangeable, etc. Reviewed By: pdillinger Differential Revision: D25754492 Pulled By: ajkr fbshipit-source-id: e481bac6c03c2ab268be41359730f1ceb9964866
2021-01-29 20:17:17 +00:00
ASSERT_OK(new_mem->Add(SequenceNumber(i), kTypeValue, key, value,
nullptr /* kv_prot_info */));
if ((i + 1000) % 10000 < 9995) {
InternalKey internal_key(key, SequenceNumber(i), kTypeValue);
inserted_keys.push_back({internal_key.Encode().ToString(), value});
}
}
{
ASSERT_OK(new_mem->Add(SequenceNumber(10000), kTypeRangeDeletion, "9995",
Integrity protection for live updates to WriteBatch (#7748) Summary: This PR adds the foundation classes for key-value integrity protection and the first use case: protecting live updates from the source buffers added to `WriteBatch` through the destination buffer in `MemTable`. The width of the protection info is not yet configurable -- only eight bytes per key is supported. This PR allows users to enable protection by constructing `WriteBatch` with `protection_bytes_per_key == 8`. It does not yet expose a way for users to get integrity protection via other write APIs (e.g., `Put()`, `Merge()`, `Delete()`, etc.). The foundation classes (`ProtectionInfo.*`) embed the coverage info in their type, and provide `Protect.*()` and `Strip.*()` functions to navigate between types with different coverage. For making bytes per key configurable (for powers of two up to eight) in the future, these classes are templated on the unsigned integer type used to store the protection info. That integer contains the XOR'd result of hashes with independent seeds for all covered fields. For integer fields, the hash is computed on the raw unadjusted bytes, so the result is endian-dependent. The most significant bytes are truncated when the hash value (8 bytes) is wider than the protection integer. When `WriteBatch` is constructed with `protection_bytes_per_key == 8`, we hold a `ProtectionInfoKVOTC` (i.e., one that covers key, value, optype aka `ValueType`, timestamp, and CF ID) for each entry added to the batch. The protection info is generated from the original buffers passed by the user, as well as the original metadata generated internally. When writing to memtable, each entry is transformed to a `ProtectionInfoKVOTS` (i.e., dropping coverage of CF ID and adding coverage of sequence number), since at that point we know the sequence number, and have already selected a memtable corresponding to a particular CF. This protection info is verified once the entry is encoded in the `MemTable` buffer. Pull Request resolved: https://github.com/facebook/rocksdb/pull/7748 Test Plan: - an integration test to verify a wide variety of single-byte changes to the encoded `MemTable` buffer are caught - add to stress/crash test to verify it works in variety of configs/operations without intentional corruption - [deferred] unit tests for `ProtectionInfo.*` classes for edge cases like KV swap, `SliceParts` and `Slice` APIs are interchangeable, etc. Reviewed By: pdillinger Differential Revision: D25754492 Pulled By: ajkr fbshipit-source-id: e481bac6c03c2ab268be41359730f1ceb9964866
2021-01-29 20:17:17 +00:00
"9999a", nullptr /* kv_prot_info */));
InternalKey internal_key("9995", SequenceNumber(10000), kTypeRangeDeletion);
inserted_keys.push_back({internal_key.Encode().ToString(), "9999a"});
}
// Note: the first two blob references will not be considered when resolving
// the oldest blob file referenced (the first one is inlined TTL, while the
// second one is TTL and thus points to a TTL blob file).
constexpr std::array<uint64_t, 6> blob_file_numbers{
{kInvalidBlobFileNumber, 5, 103, 17, 102, 101}};
for (size_t i = 0; i < blob_file_numbers.size(); ++i) {
std::string key(std::to_string(i + 10001));
std::string blob_index;
if (i == 0) {
BlobIndex::EncodeInlinedTTL(&blob_index, /* expiration */ 1234567890ULL,
"foo");
} else if (i == 1) {
BlobIndex::EncodeBlobTTL(&blob_index, /* expiration */ 1234567890ULL,
blob_file_numbers[i], /* offset */ i << 10,
/* size */ i << 20, kNoCompression);
} else {
BlobIndex::EncodeBlob(&blob_index, blob_file_numbers[i],
/* offset */ i << 10, /* size */ i << 20,
kNoCompression);
}
const SequenceNumber seq(i + 10001);
Integrity protection for live updates to WriteBatch (#7748) Summary: This PR adds the foundation classes for key-value integrity protection and the first use case: protecting live updates from the source buffers added to `WriteBatch` through the destination buffer in `MemTable`. The width of the protection info is not yet configurable -- only eight bytes per key is supported. This PR allows users to enable protection by constructing `WriteBatch` with `protection_bytes_per_key == 8`. It does not yet expose a way for users to get integrity protection via other write APIs (e.g., `Put()`, `Merge()`, `Delete()`, etc.). The foundation classes (`ProtectionInfo.*`) embed the coverage info in their type, and provide `Protect.*()` and `Strip.*()` functions to navigate between types with different coverage. For making bytes per key configurable (for powers of two up to eight) in the future, these classes are templated on the unsigned integer type used to store the protection info. That integer contains the XOR'd result of hashes with independent seeds for all covered fields. For integer fields, the hash is computed on the raw unadjusted bytes, so the result is endian-dependent. The most significant bytes are truncated when the hash value (8 bytes) is wider than the protection integer. When `WriteBatch` is constructed with `protection_bytes_per_key == 8`, we hold a `ProtectionInfoKVOTC` (i.e., one that covers key, value, optype aka `ValueType`, timestamp, and CF ID) for each entry added to the batch. The protection info is generated from the original buffers passed by the user, as well as the original metadata generated internally. When writing to memtable, each entry is transformed to a `ProtectionInfoKVOTS` (i.e., dropping coverage of CF ID and adding coverage of sequence number), since at that point we know the sequence number, and have already selected a memtable corresponding to a particular CF. This protection info is verified once the entry is encoded in the `MemTable` buffer. Pull Request resolved: https://github.com/facebook/rocksdb/pull/7748 Test Plan: - an integration test to verify a wide variety of single-byte changes to the encoded `MemTable` buffer are caught - add to stress/crash test to verify it works in variety of configs/operations without intentional corruption - [deferred] unit tests for `ProtectionInfo.*` classes for edge cases like KV swap, `SliceParts` and `Slice` APIs are interchangeable, etc. Reviewed By: pdillinger Differential Revision: D25754492 Pulled By: ajkr fbshipit-source-id: e481bac6c03c2ab268be41359730f1ceb9964866
2021-01-29 20:17:17 +00:00
ASSERT_OK(new_mem->Add(seq, kTypeBlobIndex, key, blob_index,
nullptr /* kv_prot_info */));
InternalKey internal_key(key, seq, kTypeBlobIndex);
inserted_keys.push_back({internal_key.Encode().ToString(), blob_index});
}
mock::SortKVVector(&inserted_keys);
Support saving history in memtable_list Summary: For transactions, we are using the memtables to validate that there are no write conflicts. But after flushing, we don't have any memtables, and transactions could fail to commit. So we want to someone keep around some extra history to use for conflict checking. In addition, we want to provide a way to increase the size of this history if too many transactions fail to commit. After chatting with people, it seems like everyone prefers just using Memtables to store this history (instead of a separate history structure). It seems like the best place for this is abstracted inside the memtable_list. I decide to create a separate list in MemtableListVersion as using the same list complicated the flush/installalflushresults logic too much. This diff adds a new parameter to control how much memtable history to keep around after flushing. However, it sounds like people aren't too fond of adding new parameters. So I am making the default size of flushed+not-flushed memtables be set to max_write_buffers. This should not change the maximum amount of memory used, but make it more likely we're using closer the the limit. (We are now postponing deleting flushed memtables until the max_write_buffer limit is reached). So while we might use more memory on average, we are still obeying the limit set (and you could argue it's better to go ahead and use up memory now instead of waiting for a write stall to happen to test this limit). However, if people are opposed to this default behavior, we can easily set it to 0 and require this parameter be set in order to use transactions. Test Plan: Added a xfunc test to play around with setting different values of this parameter in all tests. Added testing in memtablelist_test and planning on adding more testing here. Reviewers: sdong, rven, igor Reviewed By: igor Subscribers: dhruba, leveldb Differential Revision: https://reviews.facebook.net/D37443
2015-05-28 23:34:24 +00:00
autovector<MemTable*> to_delete;
Fragment memtable range tombstone in the write path (#10380) Summary: - Right now each read fragments the memtable range tombstones https://github.com/facebook/rocksdb/issues/4808. This PR explores the idea of fragmenting memtable range tombstones in the write path and reads can just read this cached fragmented tombstone without any fragmenting cost. This PR only does the caching for immutable memtable, and does so right before a memtable is added to an immutable memtable list. The fragmentation is done without holding mutex to minimize its performance impact. - db_bench is updated to print out the number of range deletions executed if there is any. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10380 Test Plan: - CI, added asserts in various places to check whether a fragmented range tombstone list should have been constructed. - Benchmark: as this PR only optimizes immutable memtable path, the number of writes in the benchmark is chosen such an immutable memtable is created and range tombstones are in that memtable. ``` single thread: ./db_bench --benchmarks=fillrandom,readrandom --writes_per_range_tombstone=1 --max_write_buffer_number=100 --min_write_buffer_number_to_merge=100 --writes=500000 --reads=100000 --max_num_range_tombstones=100 multi_thread ./db_bench --benchmarks=fillrandom,readrandom --writes_per_range_tombstone=1 --max_write_buffer_number=100 --min_write_buffer_number_to_merge=100 --writes=15000 --reads=20000 --threads=32 --max_num_range_tombstones=100 ``` Commit 99cdf16464a057ca44de2f747541dedf651bae9e is included in benchmark result. It was an earlier attempt where tombstones are fragmented for each write operation. Reader threads share it using a shared_ptr which would slow down multi-thread read performance as seen in benchmark results. Results are averaged over 5 runs. Single thread result: | Max # tombstones | main fillrandom micros/op | 99cdf16464a057ca44de2f747541dedf651bae9e | Post PR | main readrandom micros/op | 99cdf16464a057ca44de2f747541dedf651bae9e | Post PR | | ------------- | ------------- |------------- |------------- |------------- |------------- |------------- | | 0 |6.68 |6.57 |6.72 |4.72 |4.79 |4.54 | | 1 |6.67 |6.58 |6.62 |5.41 |4.74 |4.72 | | 10 |6.59 |6.5 |6.56 |7.83 |4.69 |4.59 | | 100 |6.62 |6.75 |6.58 |29.57 |5.04 |5.09 | | 1000 |6.54 |6.82 |6.61 |320.33 |5.22 |5.21 | 32-thread result: note that "Max # tombstones" is per thread. | Max # tombstones | main fillrandom micros/op | 99cdf16464a057ca44de2f747541dedf651bae9e | Post PR | main readrandom micros/op | 99cdf16464a057ca44de2f747541dedf651bae9e | Post PR | | ------------- | ------------- |------------- |------------- |------------- |------------- |------------- | | 0 |234.52 |260.25 |239.42 |5.06 |5.38 |5.09 | | 1 |236.46 |262.0 |231.1 |19.57 |22.14 |5.45 | | 10 |236.95 |263.84 |251.49 |151.73 |21.61 |5.73 | | 100 |268.16 |296.8 |280.13 |2308.52 |22.27 |6.57 | Reviewed By: ajkr Differential Revision: D37916564 Pulled By: cbi42 fbshipit-source-id: 05d6d2e16df26c374c57ddcca13a5bfe9d5b731e
2022-08-05 19:02:33 +00:00
new_mem->ConstructFragmentedRangeTombstones();
Support saving history in memtable_list Summary: For transactions, we are using the memtables to validate that there are no write conflicts. But after flushing, we don't have any memtables, and transactions could fail to commit. So we want to someone keep around some extra history to use for conflict checking. In addition, we want to provide a way to increase the size of this history if too many transactions fail to commit. After chatting with people, it seems like everyone prefers just using Memtables to store this history (instead of a separate history structure). It seems like the best place for this is abstracted inside the memtable_list. I decide to create a separate list in MemtableListVersion as using the same list complicated the flush/installalflushresults logic too much. This diff adds a new parameter to control how much memtable history to keep around after flushing. However, it sounds like people aren't too fond of adding new parameters. So I am making the default size of flushed+not-flushed memtables be set to max_write_buffers. This should not change the maximum amount of memory used, but make it more likely we're using closer the the limit. (We are now postponing deleting flushed memtables until the max_write_buffer limit is reached). So while we might use more memory on average, we are still obeying the limit set (and you could argue it's better to go ahead and use up memory now instead of waiting for a write stall to happen to test this limit). However, if people are opposed to this default behavior, we can easily set it to 0 and require this parameter be set in order to use transactions. Test Plan: Added a xfunc test to play around with setting different values of this parameter in all tests. Added testing in memtablelist_test and planning on adding more testing here. Reviewers: sdong, rven, igor Reviewed By: igor Subscribers: dhruba, leveldb Differential Revision: https://reviews.facebook.net/D37443
2015-05-28 23:34:24 +00:00
cfd->imm()->Add(new_mem, &to_delete);
for (auto& m : to_delete) {
delete m;
}
EventLogger event_logger(db_options_.info_log.get());
SnapshotChecker* snapshot_checker = nullptr; // not relavant
FlushJob flush_job(
dbname_, versions_->GetColumnFamilySet()->GetDefault(), db_options_,
*cfd->GetLatestMutableCFOptions(),
std::numeric_limits<uint64_t>::max() /* memtable_id */, env_options_,
versions_.get(), &mutex_, &shutting_down_, {}, kMaxSequenceNumber,
snapshot_checker, &job_context, FlushReason::kTest, nullptr, nullptr,
nullptr, kNoCompression, db_options_.statistics.get(), &event_logger,
true, true /* sync_output_directory */, true /* write_manifest */,
Env::Priority::USER, nullptr /*IOTracer*/, empty_seqno_to_time_mapping_);
HistogramData hist;
FileMetaData file_meta;
mutex_.Lock();
flush_job.PickMemTable();
ASSERT_OK(flush_job.Run(nullptr, &file_meta));
mutex_.Unlock();
db_options_.statistics->histogramData(FLUSH_TIME, &hist);
ASSERT_GT(hist.average, 0.0);
ASSERT_EQ(std::to_string(0), file_meta.smallest.user_key().ToString());
ASSERT_EQ("9999a", file_meta.largest.user_key().ToString());
ASSERT_EQ(1, file_meta.fd.smallest_seqno);
ASSERT_EQ(10006, file_meta.fd.largest_seqno);
ASSERT_EQ(17, file_meta.oldest_blob_file_number);
mock_table_factory_->AssertSingleFile(inserted_keys);
job_context.Clean();
}
TEST_F(FlushJobTest, FlushMemTablesSingleColumnFamily) {
const size_t num_mems = 2;
const size_t num_mems_to_flush = 1;
const size_t num_keys_per_table = 100;
JobContext job_context(0);
ColumnFamilyData* cfd = versions_->GetColumnFamilySet()->GetDefault();
std::vector<uint64_t> memtable_ids;
std::vector<MemTable*> new_mems;
for (size_t i = 0; i != num_mems; ++i) {
MemTable* mem = cfd->ConstructNewMemtable(*cfd->GetLatestMutableCFOptions(),
kMaxSequenceNumber);
mem->SetID(i);
mem->Ref();
new_mems.emplace_back(mem);
memtable_ids.push_back(mem->GetID());
for (size_t j = 0; j < num_keys_per_table; ++j) {
std::string key(std::to_string(j + i * num_keys_per_table));
std::string value("value" + key);
ASSERT_OK(mem->Add(SequenceNumber(j + i * num_keys_per_table), kTypeValue,
Integrity protection for live updates to WriteBatch (#7748) Summary: This PR adds the foundation classes for key-value integrity protection and the first use case: protecting live updates from the source buffers added to `WriteBatch` through the destination buffer in `MemTable`. The width of the protection info is not yet configurable -- only eight bytes per key is supported. This PR allows users to enable protection by constructing `WriteBatch` with `protection_bytes_per_key == 8`. It does not yet expose a way for users to get integrity protection via other write APIs (e.g., `Put()`, `Merge()`, `Delete()`, etc.). The foundation classes (`ProtectionInfo.*`) embed the coverage info in their type, and provide `Protect.*()` and `Strip.*()` functions to navigate between types with different coverage. For making bytes per key configurable (for powers of two up to eight) in the future, these classes are templated on the unsigned integer type used to store the protection info. That integer contains the XOR'd result of hashes with independent seeds for all covered fields. For integer fields, the hash is computed on the raw unadjusted bytes, so the result is endian-dependent. The most significant bytes are truncated when the hash value (8 bytes) is wider than the protection integer. When `WriteBatch` is constructed with `protection_bytes_per_key == 8`, we hold a `ProtectionInfoKVOTC` (i.e., one that covers key, value, optype aka `ValueType`, timestamp, and CF ID) for each entry added to the batch. The protection info is generated from the original buffers passed by the user, as well as the original metadata generated internally. When writing to memtable, each entry is transformed to a `ProtectionInfoKVOTS` (i.e., dropping coverage of CF ID and adding coverage of sequence number), since at that point we know the sequence number, and have already selected a memtable corresponding to a particular CF. This protection info is verified once the entry is encoded in the `MemTable` buffer. Pull Request resolved: https://github.com/facebook/rocksdb/pull/7748 Test Plan: - an integration test to verify a wide variety of single-byte changes to the encoded `MemTable` buffer are caught - add to stress/crash test to verify it works in variety of configs/operations without intentional corruption - [deferred] unit tests for `ProtectionInfo.*` classes for edge cases like KV swap, `SliceParts` and `Slice` APIs are interchangeable, etc. Reviewed By: pdillinger Differential Revision: D25754492 Pulled By: ajkr fbshipit-source-id: e481bac6c03c2ab268be41359730f1ceb9964866
2021-01-29 20:17:17 +00:00
key, value, nullptr /* kv_prot_info */));
}
}
autovector<MemTable*> to_delete;
for (auto mem : new_mems) {
Fragment memtable range tombstone in the write path (#10380) Summary: - Right now each read fragments the memtable range tombstones https://github.com/facebook/rocksdb/issues/4808. This PR explores the idea of fragmenting memtable range tombstones in the write path and reads can just read this cached fragmented tombstone without any fragmenting cost. This PR only does the caching for immutable memtable, and does so right before a memtable is added to an immutable memtable list. The fragmentation is done without holding mutex to minimize its performance impact. - db_bench is updated to print out the number of range deletions executed if there is any. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10380 Test Plan: - CI, added asserts in various places to check whether a fragmented range tombstone list should have been constructed. - Benchmark: as this PR only optimizes immutable memtable path, the number of writes in the benchmark is chosen such an immutable memtable is created and range tombstones are in that memtable. ``` single thread: ./db_bench --benchmarks=fillrandom,readrandom --writes_per_range_tombstone=1 --max_write_buffer_number=100 --min_write_buffer_number_to_merge=100 --writes=500000 --reads=100000 --max_num_range_tombstones=100 multi_thread ./db_bench --benchmarks=fillrandom,readrandom --writes_per_range_tombstone=1 --max_write_buffer_number=100 --min_write_buffer_number_to_merge=100 --writes=15000 --reads=20000 --threads=32 --max_num_range_tombstones=100 ``` Commit 99cdf16464a057ca44de2f747541dedf651bae9e is included in benchmark result. It was an earlier attempt where tombstones are fragmented for each write operation. Reader threads share it using a shared_ptr which would slow down multi-thread read performance as seen in benchmark results. Results are averaged over 5 runs. Single thread result: | Max # tombstones | main fillrandom micros/op | 99cdf16464a057ca44de2f747541dedf651bae9e | Post PR | main readrandom micros/op | 99cdf16464a057ca44de2f747541dedf651bae9e | Post PR | | ------------- | ------------- |------------- |------------- |------------- |------------- |------------- | | 0 |6.68 |6.57 |6.72 |4.72 |4.79 |4.54 | | 1 |6.67 |6.58 |6.62 |5.41 |4.74 |4.72 | | 10 |6.59 |6.5 |6.56 |7.83 |4.69 |4.59 | | 100 |6.62 |6.75 |6.58 |29.57 |5.04 |5.09 | | 1000 |6.54 |6.82 |6.61 |320.33 |5.22 |5.21 | 32-thread result: note that "Max # tombstones" is per thread. | Max # tombstones | main fillrandom micros/op | 99cdf16464a057ca44de2f747541dedf651bae9e | Post PR | main readrandom micros/op | 99cdf16464a057ca44de2f747541dedf651bae9e | Post PR | | ------------- | ------------- |------------- |------------- |------------- |------------- |------------- | | 0 |234.52 |260.25 |239.42 |5.06 |5.38 |5.09 | | 1 |236.46 |262.0 |231.1 |19.57 |22.14 |5.45 | | 10 |236.95 |263.84 |251.49 |151.73 |21.61 |5.73 | | 100 |268.16 |296.8 |280.13 |2308.52 |22.27 |6.57 | Reviewed By: ajkr Differential Revision: D37916564 Pulled By: cbi42 fbshipit-source-id: 05d6d2e16df26c374c57ddcca13a5bfe9d5b731e
2022-08-05 19:02:33 +00:00
mem->ConstructFragmentedRangeTombstones();
cfd->imm()->Add(mem, &to_delete);
}
EventLogger event_logger(db_options_.info_log.get());
SnapshotChecker* snapshot_checker = nullptr; // not relavant
assert(memtable_ids.size() == num_mems);
uint64_t smallest_memtable_id = memtable_ids.front();
uint64_t flush_memtable_id = smallest_memtable_id + num_mems_to_flush - 1;
FlushJob flush_job(
dbname_, versions_->GetColumnFamilySet()->GetDefault(), db_options_,
*cfd->GetLatestMutableCFOptions(), flush_memtable_id, env_options_,
versions_.get(), &mutex_, &shutting_down_, {}, kMaxSequenceNumber,
snapshot_checker, &job_context, FlushReason::kTest, nullptr, nullptr,
nullptr, kNoCompression, db_options_.statistics.get(), &event_logger,
true, true /* sync_output_directory */, true /* write_manifest */,
Env::Priority::USER, nullptr /*IOTracer*/, empty_seqno_to_time_mapping_);
HistogramData hist;
FileMetaData file_meta;
mutex_.Lock();
flush_job.PickMemTable();
ASSERT_OK(flush_job.Run(nullptr /* prep_tracker */, &file_meta));
mutex_.Unlock();
db_options_.statistics->histogramData(FLUSH_TIME, &hist);
ASSERT_GT(hist.average, 0.0);
ASSERT_EQ(std::to_string(0), file_meta.smallest.user_key().ToString());
ASSERT_EQ("99", file_meta.largest.user_key().ToString());
ASSERT_EQ(0, file_meta.fd.smallest_seqno);
ASSERT_EQ(SequenceNumber(num_mems_to_flush * num_keys_per_table - 1),
file_meta.fd.largest_seqno);
ASSERT_EQ(kInvalidBlobFileNumber, file_meta.oldest_blob_file_number);
for (auto m : to_delete) {
delete m;
}
to_delete.clear();
job_context.Clean();
}
TEST_F(FlushJobTest, FlushMemtablesMultipleColumnFamilies) {
autovector<ColumnFamilyData*> all_cfds;
for (auto cfd : *versions_->GetColumnFamilySet()) {
all_cfds.push_back(cfd);
}
const std::vector<size_t> num_memtables = {2, 1, 3};
assert(num_memtables.size() == column_family_names_.size());
const size_t num_keys_per_memtable = 1000;
JobContext job_context(0);
std::vector<uint64_t> memtable_ids;
std::vector<SequenceNumber> smallest_seqs;
std::vector<SequenceNumber> largest_seqs;
autovector<MemTable*> to_delete;
SequenceNumber curr_seqno = 0;
size_t k = 0;
for (auto cfd : all_cfds) {
smallest_seqs.push_back(curr_seqno);
for (size_t i = 0; i != num_memtables[k]; ++i) {
MemTable* mem = cfd->ConstructNewMemtable(
*cfd->GetLatestMutableCFOptions(), kMaxSequenceNumber);
mem->SetID(i);
mem->Ref();
for (size_t j = 0; j != num_keys_per_memtable; ++j) {
std::string key(std::to_string(j + i * num_keys_per_memtable));
std::string value("value" + key);
Integrity protection for live updates to WriteBatch (#7748) Summary: This PR adds the foundation classes for key-value integrity protection and the first use case: protecting live updates from the source buffers added to `WriteBatch` through the destination buffer in `MemTable`. The width of the protection info is not yet configurable -- only eight bytes per key is supported. This PR allows users to enable protection by constructing `WriteBatch` with `protection_bytes_per_key == 8`. It does not yet expose a way for users to get integrity protection via other write APIs (e.g., `Put()`, `Merge()`, `Delete()`, etc.). The foundation classes (`ProtectionInfo.*`) embed the coverage info in their type, and provide `Protect.*()` and `Strip.*()` functions to navigate between types with different coverage. For making bytes per key configurable (for powers of two up to eight) in the future, these classes are templated on the unsigned integer type used to store the protection info. That integer contains the XOR'd result of hashes with independent seeds for all covered fields. For integer fields, the hash is computed on the raw unadjusted bytes, so the result is endian-dependent. The most significant bytes are truncated when the hash value (8 bytes) is wider than the protection integer. When `WriteBatch` is constructed with `protection_bytes_per_key == 8`, we hold a `ProtectionInfoKVOTC` (i.e., one that covers key, value, optype aka `ValueType`, timestamp, and CF ID) for each entry added to the batch. The protection info is generated from the original buffers passed by the user, as well as the original metadata generated internally. When writing to memtable, each entry is transformed to a `ProtectionInfoKVOTS` (i.e., dropping coverage of CF ID and adding coverage of sequence number), since at that point we know the sequence number, and have already selected a memtable corresponding to a particular CF. This protection info is verified once the entry is encoded in the `MemTable` buffer. Pull Request resolved: https://github.com/facebook/rocksdb/pull/7748 Test Plan: - an integration test to verify a wide variety of single-byte changes to the encoded `MemTable` buffer are caught - add to stress/crash test to verify it works in variety of configs/operations without intentional corruption - [deferred] unit tests for `ProtectionInfo.*` classes for edge cases like KV swap, `SliceParts` and `Slice` APIs are interchangeable, etc. Reviewed By: pdillinger Differential Revision: D25754492 Pulled By: ajkr fbshipit-source-id: e481bac6c03c2ab268be41359730f1ceb9964866
2021-01-29 20:17:17 +00:00
ASSERT_OK(mem->Add(curr_seqno++, kTypeValue, key, value,
nullptr /* kv_prot_info */));
}
Fragment memtable range tombstone in the write path (#10380) Summary: - Right now each read fragments the memtable range tombstones https://github.com/facebook/rocksdb/issues/4808. This PR explores the idea of fragmenting memtable range tombstones in the write path and reads can just read this cached fragmented tombstone without any fragmenting cost. This PR only does the caching for immutable memtable, and does so right before a memtable is added to an immutable memtable list. The fragmentation is done without holding mutex to minimize its performance impact. - db_bench is updated to print out the number of range deletions executed if there is any. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10380 Test Plan: - CI, added asserts in various places to check whether a fragmented range tombstone list should have been constructed. - Benchmark: as this PR only optimizes immutable memtable path, the number of writes in the benchmark is chosen such an immutable memtable is created and range tombstones are in that memtable. ``` single thread: ./db_bench --benchmarks=fillrandom,readrandom --writes_per_range_tombstone=1 --max_write_buffer_number=100 --min_write_buffer_number_to_merge=100 --writes=500000 --reads=100000 --max_num_range_tombstones=100 multi_thread ./db_bench --benchmarks=fillrandom,readrandom --writes_per_range_tombstone=1 --max_write_buffer_number=100 --min_write_buffer_number_to_merge=100 --writes=15000 --reads=20000 --threads=32 --max_num_range_tombstones=100 ``` Commit 99cdf16464a057ca44de2f747541dedf651bae9e is included in benchmark result. It was an earlier attempt where tombstones are fragmented for each write operation. Reader threads share it using a shared_ptr which would slow down multi-thread read performance as seen in benchmark results. Results are averaged over 5 runs. Single thread result: | Max # tombstones | main fillrandom micros/op | 99cdf16464a057ca44de2f747541dedf651bae9e | Post PR | main readrandom micros/op | 99cdf16464a057ca44de2f747541dedf651bae9e | Post PR | | ------------- | ------------- |------------- |------------- |------------- |------------- |------------- | | 0 |6.68 |6.57 |6.72 |4.72 |4.79 |4.54 | | 1 |6.67 |6.58 |6.62 |5.41 |4.74 |4.72 | | 10 |6.59 |6.5 |6.56 |7.83 |4.69 |4.59 | | 100 |6.62 |6.75 |6.58 |29.57 |5.04 |5.09 | | 1000 |6.54 |6.82 |6.61 |320.33 |5.22 |5.21 | 32-thread result: note that "Max # tombstones" is per thread. | Max # tombstones | main fillrandom micros/op | 99cdf16464a057ca44de2f747541dedf651bae9e | Post PR | main readrandom micros/op | 99cdf16464a057ca44de2f747541dedf651bae9e | Post PR | | ------------- | ------------- |------------- |------------- |------------- |------------- |------------- | | 0 |234.52 |260.25 |239.42 |5.06 |5.38 |5.09 | | 1 |236.46 |262.0 |231.1 |19.57 |22.14 |5.45 | | 10 |236.95 |263.84 |251.49 |151.73 |21.61 |5.73 | | 100 |268.16 |296.8 |280.13 |2308.52 |22.27 |6.57 | Reviewed By: ajkr Differential Revision: D37916564 Pulled By: cbi42 fbshipit-source-id: 05d6d2e16df26c374c57ddcca13a5bfe9d5b731e
2022-08-05 19:02:33 +00:00
mem->ConstructFragmentedRangeTombstones();
cfd->imm()->Add(mem, &to_delete);
}
largest_seqs.push_back(curr_seqno - 1);
memtable_ids.push_back(num_memtables[k++] - 1);
}
EventLogger event_logger(db_options_.info_log.get());
SnapshotChecker* snapshot_checker = nullptr; // not relevant
std::vector<std::unique_ptr<FlushJob>> flush_jobs;
k = 0;
for (auto cfd : all_cfds) {
std::vector<SequenceNumber> snapshot_seqs;
flush_jobs.emplace_back(new FlushJob(
dbname_, cfd, db_options_, *cfd->GetLatestMutableCFOptions(),
memtable_ids[k], env_options_, versions_.get(), &mutex_,
&shutting_down_, snapshot_seqs, kMaxSequenceNumber, snapshot_checker,
&job_context, FlushReason::kTest, nullptr, nullptr, nullptr,
kNoCompression, db_options_.statistics.get(), &event_logger, true,
false /* sync_output_directory */, false /* write_manifest */,
Env::Priority::USER, nullptr /*IOTracer*/,
empty_seqno_to_time_mapping_));
k++;
}
HistogramData hist;
std::vector<FileMetaData> file_metas;
// Call reserve to avoid auto-resizing
file_metas.reserve(flush_jobs.size());
mutex_.Lock();
for (auto& job : flush_jobs) {
job->PickMemTable();
}
for (auto& job : flush_jobs) {
FileMetaData meta;
// Run will release and re-acquire mutex
ASSERT_OK(job->Run(nullptr /**/, &meta));
file_metas.emplace_back(meta);
}
autovector<FileMetaData*> file_meta_ptrs;
for (auto& meta : file_metas) {
file_meta_ptrs.push_back(&meta);
}
autovector<const autovector<MemTable*>*> mems_list;
for (size_t i = 0; i != all_cfds.size(); ++i) {
const auto& mems = flush_jobs[i]->GetMemTables();
mems_list.push_back(&mems);
}
autovector<const MutableCFOptions*> mutable_cf_options_list;
for (auto cfd : all_cfds) {
mutable_cf_options_list.push_back(cfd->GetLatestMutableCFOptions());
}
autovector<std::list<std::unique_ptr<FlushJobInfo>>*>
committed_flush_jobs_info;
for (auto& job : flush_jobs) {
committed_flush_jobs_info.push_back(job->GetCommittedFlushJobsInfo());
}
Status s = InstallMemtableAtomicFlushResults(
nullptr /* imm_lists */, all_cfds, mutable_cf_options_list, mems_list,
versions_.get(), nullptr /* prep_tracker */, &mutex_, file_meta_ptrs,
committed_flush_jobs_info, &job_context.memtables_to_free,
nullptr /* db_directory */, nullptr /* log_buffer */);
ASSERT_OK(s);
mutex_.Unlock();
db_options_.statistics->histogramData(FLUSH_TIME, &hist);
ASSERT_GT(hist.average, 0.0);
k = 0;
for (const auto& file_meta : file_metas) {
ASSERT_EQ(std::to_string(0), file_meta.smallest.user_key().ToString());
ASSERT_EQ("999", file_meta.largest.user_key()
.ToString()); // max key by bytewise comparator
ASSERT_EQ(smallest_seqs[k], file_meta.fd.smallest_seqno);
ASSERT_EQ(largest_seqs[k], file_meta.fd.largest_seqno);
// Verify that imm is empty
ASSERT_EQ(std::numeric_limits<uint64_t>::max(),
all_cfds[k]->imm()->GetEarliestMemTableID());
Give retry flushes their own functions (#11903) Summary: Recovery triggers flushes for very different scenarios: (1) `FlushReason::kErrorRecoveryRetryFlush`: a flush failed (2) `FlushReason::kErrorRecovery`: a WAL may be corrupted (3) `FlushReason::kCatchUpAfterErrorRecovery`: immutable memtables may have accumulated The old code called called `FlushAllColumnFamilies()` in all cases, which uses manual flush functions: `AtomicFlushMemTables()` and `FlushMemTable()`. Forcing flushing the latest data on all CFs was useful for (2) because it ensures all CFs move past the corrupted WAL. However, those code paths were overkill for (1) and (3), where only already-immutable memtables need to be flushed. There were conditionals to exclude some of the extraneous logic but I found there was still too much happening. For example, both of the manual flush functions enter the write thread. Entering the write thread is inconvenient because then we can't allow stalled writes to wait on a retrying flush to finish. Instead of continuing down the path of adding more conditionals to the manual flush functions, this PR introduces a dedicated function for cases (1) and (3): `RetryFlushesForErrorRecovery()`. Also I cleaned up the manual flush functions to remove existing conditionals for these cases as they're no longer needed. Pull Request resolved: https://github.com/facebook/rocksdb/pull/11903 Reviewed By: cbi42 Differential Revision: D49693812 Pulled By: ajkr fbshipit-source-id: 7630ac539b9d6c92052c13a3cdce53256134d990
2023-10-02 23:26:24 +00:00
ASSERT_EQ(0, all_cfds[k]->imm()->GetLatestMemTableID(
false /* for_atomic_flush */));
++k;
}
for (auto m : to_delete) {
delete m;
}
to_delete.clear();
job_context.Clean();
}
TEST_F(FlushJobTest, Snapshots) {
JobContext job_context(0);
auto cfd = versions_->GetColumnFamilySet()->GetDefault();
auto new_mem = cfd->ConstructNewMemtable(*cfd->GetLatestMutableCFOptions(),
kMaxSequenceNumber);
std::set<SequenceNumber> snapshots_set;
int keys = 10000;
int max_inserts_per_keys = 8;
Random rnd(301);
for (int i = 0; i < keys / 2; ++i) {
snapshots_set.insert(rnd.Uniform(keys * (max_inserts_per_keys / 2)) + 1);
}
// set has already removed the duplicate snapshots
std::vector<SequenceNumber> snapshots(snapshots_set.begin(),
snapshots_set.end());
new_mem->Ref();
SequenceNumber current_seqno = 0;
auto inserted_keys = mock::MakeMockFile();
for (int i = 1; i < keys; ++i) {
std::string key(std::to_string(i));
int insertions = rnd.Uniform(max_inserts_per_keys);
for (int j = 0; j < insertions; ++j) {
std::string value(rnd.HumanReadableString(10));
auto seqno = ++current_seqno;
Integrity protection for live updates to WriteBatch (#7748) Summary: This PR adds the foundation classes for key-value integrity protection and the first use case: protecting live updates from the source buffers added to `WriteBatch` through the destination buffer in `MemTable`. The width of the protection info is not yet configurable -- only eight bytes per key is supported. This PR allows users to enable protection by constructing `WriteBatch` with `protection_bytes_per_key == 8`. It does not yet expose a way for users to get integrity protection via other write APIs (e.g., `Put()`, `Merge()`, `Delete()`, etc.). The foundation classes (`ProtectionInfo.*`) embed the coverage info in their type, and provide `Protect.*()` and `Strip.*()` functions to navigate between types with different coverage. For making bytes per key configurable (for powers of two up to eight) in the future, these classes are templated on the unsigned integer type used to store the protection info. That integer contains the XOR'd result of hashes with independent seeds for all covered fields. For integer fields, the hash is computed on the raw unadjusted bytes, so the result is endian-dependent. The most significant bytes are truncated when the hash value (8 bytes) is wider than the protection integer. When `WriteBatch` is constructed with `protection_bytes_per_key == 8`, we hold a `ProtectionInfoKVOTC` (i.e., one that covers key, value, optype aka `ValueType`, timestamp, and CF ID) for each entry added to the batch. The protection info is generated from the original buffers passed by the user, as well as the original metadata generated internally. When writing to memtable, each entry is transformed to a `ProtectionInfoKVOTS` (i.e., dropping coverage of CF ID and adding coverage of sequence number), since at that point we know the sequence number, and have already selected a memtable corresponding to a particular CF. This protection info is verified once the entry is encoded in the `MemTable` buffer. Pull Request resolved: https://github.com/facebook/rocksdb/pull/7748 Test Plan: - an integration test to verify a wide variety of single-byte changes to the encoded `MemTable` buffer are caught - add to stress/crash test to verify it works in variety of configs/operations without intentional corruption - [deferred] unit tests for `ProtectionInfo.*` classes for edge cases like KV swap, `SliceParts` and `Slice` APIs are interchangeable, etc. Reviewed By: pdillinger Differential Revision: D25754492 Pulled By: ajkr fbshipit-source-id: e481bac6c03c2ab268be41359730f1ceb9964866
2021-01-29 20:17:17 +00:00
ASSERT_OK(new_mem->Add(SequenceNumber(seqno), kTypeValue, key, value,
nullptr /* kv_prot_info */));
// a key is visible only if:
// 1. it's the last one written (j == insertions - 1)
// 2. there's a snapshot pointing at it
bool visible = (j == insertions - 1) ||
(snapshots_set.find(seqno) != snapshots_set.end());
if (visible) {
InternalKey internal_key(key, seqno, kTypeValue);
inserted_keys.push_back({internal_key.Encode().ToString(), value});
}
}
}
mock::SortKVVector(&inserted_keys);
autovector<MemTable*> to_delete;
Fragment memtable range tombstone in the write path (#10380) Summary: - Right now each read fragments the memtable range tombstones https://github.com/facebook/rocksdb/issues/4808. This PR explores the idea of fragmenting memtable range tombstones in the write path and reads can just read this cached fragmented tombstone without any fragmenting cost. This PR only does the caching for immutable memtable, and does so right before a memtable is added to an immutable memtable list. The fragmentation is done without holding mutex to minimize its performance impact. - db_bench is updated to print out the number of range deletions executed if there is any. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10380 Test Plan: - CI, added asserts in various places to check whether a fragmented range tombstone list should have been constructed. - Benchmark: as this PR only optimizes immutable memtable path, the number of writes in the benchmark is chosen such an immutable memtable is created and range tombstones are in that memtable. ``` single thread: ./db_bench --benchmarks=fillrandom,readrandom --writes_per_range_tombstone=1 --max_write_buffer_number=100 --min_write_buffer_number_to_merge=100 --writes=500000 --reads=100000 --max_num_range_tombstones=100 multi_thread ./db_bench --benchmarks=fillrandom,readrandom --writes_per_range_tombstone=1 --max_write_buffer_number=100 --min_write_buffer_number_to_merge=100 --writes=15000 --reads=20000 --threads=32 --max_num_range_tombstones=100 ``` Commit 99cdf16464a057ca44de2f747541dedf651bae9e is included in benchmark result. It was an earlier attempt where tombstones are fragmented for each write operation. Reader threads share it using a shared_ptr which would slow down multi-thread read performance as seen in benchmark results. Results are averaged over 5 runs. Single thread result: | Max # tombstones | main fillrandom micros/op | 99cdf16464a057ca44de2f747541dedf651bae9e | Post PR | main readrandom micros/op | 99cdf16464a057ca44de2f747541dedf651bae9e | Post PR | | ------------- | ------------- |------------- |------------- |------------- |------------- |------------- | | 0 |6.68 |6.57 |6.72 |4.72 |4.79 |4.54 | | 1 |6.67 |6.58 |6.62 |5.41 |4.74 |4.72 | | 10 |6.59 |6.5 |6.56 |7.83 |4.69 |4.59 | | 100 |6.62 |6.75 |6.58 |29.57 |5.04 |5.09 | | 1000 |6.54 |6.82 |6.61 |320.33 |5.22 |5.21 | 32-thread result: note that "Max # tombstones" is per thread. | Max # tombstones | main fillrandom micros/op | 99cdf16464a057ca44de2f747541dedf651bae9e | Post PR | main readrandom micros/op | 99cdf16464a057ca44de2f747541dedf651bae9e | Post PR | | ------------- | ------------- |------------- |------------- |------------- |------------- |------------- | | 0 |234.52 |260.25 |239.42 |5.06 |5.38 |5.09 | | 1 |236.46 |262.0 |231.1 |19.57 |22.14 |5.45 | | 10 |236.95 |263.84 |251.49 |151.73 |21.61 |5.73 | | 100 |268.16 |296.8 |280.13 |2308.52 |22.27 |6.57 | Reviewed By: ajkr Differential Revision: D37916564 Pulled By: cbi42 fbshipit-source-id: 05d6d2e16df26c374c57ddcca13a5bfe9d5b731e
2022-08-05 19:02:33 +00:00
new_mem->ConstructFragmentedRangeTombstones();
cfd->imm()->Add(new_mem, &to_delete);
for (auto& m : to_delete) {
delete m;
}
EventLogger event_logger(db_options_.info_log.get());
SnapshotChecker* snapshot_checker = nullptr; // not relavant
FlushJob flush_job(
dbname_, versions_->GetColumnFamilySet()->GetDefault(), db_options_,
*cfd->GetLatestMutableCFOptions(),
std::numeric_limits<uint64_t>::max() /* memtable_id */, env_options_,
versions_.get(), &mutex_, &shutting_down_, snapshots, kMaxSequenceNumber,
snapshot_checker, &job_context, FlushReason::kTest, nullptr, nullptr,
nullptr, kNoCompression, db_options_.statistics.get(), &event_logger,
true, true /* sync_output_directory */, true /* write_manifest */,
Env::Priority::USER, nullptr /*IOTracer*/, empty_seqno_to_time_mapping_);
mutex_.Lock();
flush_job.PickMemTable();
ASSERT_OK(flush_job.Run());
mutex_.Unlock();
mock_table_factory_->AssertSingleFile(inserted_keys);
HistogramData hist;
db_options_.statistics->histogramData(FLUSH_TIME, &hist);
ASSERT_GT(hist.average, 0.0);
2014-11-15 00:57:17 +00:00
job_context.Clean();
}
Set Write rate limiter priority dynamically and pass it to FS (#9988) Summary: ### Context: Background compactions and flush generate large reads and writes, and can be long running, especially for universal compaction. In some cases, this can impact foreground reads and writes by users. From the RocksDB perspective, there can be two kinds of rate limiters, the internal (native) one and the external one. - The internal (native) rate limiter is introduced in [the wiki](https://github.com/facebook/rocksdb/wiki/Rate-Limiter). Currently, only IO_LOW and IO_HIGH are used and they are set statically. - For the external rate limiter, in FSWritableFile functions, IOOptions is open for end users to set and get rate_limiter_priority for their own rate limiter. Currently, RocksDB doesn’t pass the rate_limiter_priority through IOOptions to the file system. ### Solution During the User Read, Flush write, Compaction read/write, the WriteController is used to determine whether DB writes are stalled or slowed down. The rate limiter priority (Env::IOPriority) can be determined accordingly. We decided to always pass the priority in IOOptions. What the file system does with it should be a contract between the user and the file system. We would like to set the rate limiter priority at file level, since the Flush/Compaction job level may be too coarse with multiple files and block IO level is too granular. **This PR is for the Write path.** The **Write:** dynamic priority for different state are listed as follows: | State | Normal | Delayed | Stalled | | ----- | ------ | ------- | ------- | | Flush | IO_HIGH | IO_USER | IO_USER | | Compaction | IO_LOW | IO_USER | IO_USER | Flush and Compaction writes share the same call path through BlockBaseTableWriter, WritableFileWriter, and FSWritableFile. When a new FSWritableFile object is created, its io_priority_ can be set dynamically based on the state of the WriteController. In WritableFileWriter, before the call sites of FSWritableFile functions, WritableFileWriter::DecideRateLimiterPriority() determines the rate_limiter_priority. The options (IOOptions) argument of FSWritableFile functions will be updated with the rate_limiter_priority. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9988 Test Plan: Add unit tests. Reviewed By: anand1976 Differential Revision: D36395159 Pulled By: gitbw95 fbshipit-source-id: a7c82fc29759139a1a07ec46c37dbf7e753474cf
2022-05-18 07:41:41 +00:00
TEST_F(FlushJobTest, GetRateLimiterPriorityForWrite) {
// Prepare a FlushJob that flush MemTables of Single Column Family.
const size_t num_mems = 2;
const size_t num_mems_to_flush = 1;
const size_t num_keys_per_table = 100;
JobContext job_context(0);
ColumnFamilyData* cfd = versions_->GetColumnFamilySet()->GetDefault();
std::vector<uint64_t> memtable_ids;
std::vector<MemTable*> new_mems;
for (size_t i = 0; i != num_mems; ++i) {
MemTable* mem = cfd->ConstructNewMemtable(*cfd->GetLatestMutableCFOptions(),
kMaxSequenceNumber);
mem->SetID(i);
mem->Ref();
new_mems.emplace_back(mem);
memtable_ids.push_back(mem->GetID());
for (size_t j = 0; j < num_keys_per_table; ++j) {
std::string key(std::to_string(j + i * num_keys_per_table));
std::string value("value" + key);
ASSERT_OK(mem->Add(SequenceNumber(j + i * num_keys_per_table), kTypeValue,
key, value, nullptr /* kv_prot_info */));
}
}
autovector<MemTable*> to_delete;
for (auto mem : new_mems) {
Fragment memtable range tombstone in the write path (#10380) Summary: - Right now each read fragments the memtable range tombstones https://github.com/facebook/rocksdb/issues/4808. This PR explores the idea of fragmenting memtable range tombstones in the write path and reads can just read this cached fragmented tombstone without any fragmenting cost. This PR only does the caching for immutable memtable, and does so right before a memtable is added to an immutable memtable list. The fragmentation is done without holding mutex to minimize its performance impact. - db_bench is updated to print out the number of range deletions executed if there is any. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10380 Test Plan: - CI, added asserts in various places to check whether a fragmented range tombstone list should have been constructed. - Benchmark: as this PR only optimizes immutable memtable path, the number of writes in the benchmark is chosen such an immutable memtable is created and range tombstones are in that memtable. ``` single thread: ./db_bench --benchmarks=fillrandom,readrandom --writes_per_range_tombstone=1 --max_write_buffer_number=100 --min_write_buffer_number_to_merge=100 --writes=500000 --reads=100000 --max_num_range_tombstones=100 multi_thread ./db_bench --benchmarks=fillrandom,readrandom --writes_per_range_tombstone=1 --max_write_buffer_number=100 --min_write_buffer_number_to_merge=100 --writes=15000 --reads=20000 --threads=32 --max_num_range_tombstones=100 ``` Commit 99cdf16464a057ca44de2f747541dedf651bae9e is included in benchmark result. It was an earlier attempt where tombstones are fragmented for each write operation. Reader threads share it using a shared_ptr which would slow down multi-thread read performance as seen in benchmark results. Results are averaged over 5 runs. Single thread result: | Max # tombstones | main fillrandom micros/op | 99cdf16464a057ca44de2f747541dedf651bae9e | Post PR | main readrandom micros/op | 99cdf16464a057ca44de2f747541dedf651bae9e | Post PR | | ------------- | ------------- |------------- |------------- |------------- |------------- |------------- | | 0 |6.68 |6.57 |6.72 |4.72 |4.79 |4.54 | | 1 |6.67 |6.58 |6.62 |5.41 |4.74 |4.72 | | 10 |6.59 |6.5 |6.56 |7.83 |4.69 |4.59 | | 100 |6.62 |6.75 |6.58 |29.57 |5.04 |5.09 | | 1000 |6.54 |6.82 |6.61 |320.33 |5.22 |5.21 | 32-thread result: note that "Max # tombstones" is per thread. | Max # tombstones | main fillrandom micros/op | 99cdf16464a057ca44de2f747541dedf651bae9e | Post PR | main readrandom micros/op | 99cdf16464a057ca44de2f747541dedf651bae9e | Post PR | | ------------- | ------------- |------------- |------------- |------------- |------------- |------------- | | 0 |234.52 |260.25 |239.42 |5.06 |5.38 |5.09 | | 1 |236.46 |262.0 |231.1 |19.57 |22.14 |5.45 | | 10 |236.95 |263.84 |251.49 |151.73 |21.61 |5.73 | | 100 |268.16 |296.8 |280.13 |2308.52 |22.27 |6.57 | Reviewed By: ajkr Differential Revision: D37916564 Pulled By: cbi42 fbshipit-source-id: 05d6d2e16df26c374c57ddcca13a5bfe9d5b731e
2022-08-05 19:02:33 +00:00
mem->ConstructFragmentedRangeTombstones();
Set Write rate limiter priority dynamically and pass it to FS (#9988) Summary: ### Context: Background compactions and flush generate large reads and writes, and can be long running, especially for universal compaction. In some cases, this can impact foreground reads and writes by users. From the RocksDB perspective, there can be two kinds of rate limiters, the internal (native) one and the external one. - The internal (native) rate limiter is introduced in [the wiki](https://github.com/facebook/rocksdb/wiki/Rate-Limiter). Currently, only IO_LOW and IO_HIGH are used and they are set statically. - For the external rate limiter, in FSWritableFile functions, IOOptions is open for end users to set and get rate_limiter_priority for their own rate limiter. Currently, RocksDB doesn’t pass the rate_limiter_priority through IOOptions to the file system. ### Solution During the User Read, Flush write, Compaction read/write, the WriteController is used to determine whether DB writes are stalled or slowed down. The rate limiter priority (Env::IOPriority) can be determined accordingly. We decided to always pass the priority in IOOptions. What the file system does with it should be a contract between the user and the file system. We would like to set the rate limiter priority at file level, since the Flush/Compaction job level may be too coarse with multiple files and block IO level is too granular. **This PR is for the Write path.** The **Write:** dynamic priority for different state are listed as follows: | State | Normal | Delayed | Stalled | | ----- | ------ | ------- | ------- | | Flush | IO_HIGH | IO_USER | IO_USER | | Compaction | IO_LOW | IO_USER | IO_USER | Flush and Compaction writes share the same call path through BlockBaseTableWriter, WritableFileWriter, and FSWritableFile. When a new FSWritableFile object is created, its io_priority_ can be set dynamically based on the state of the WriteController. In WritableFileWriter, before the call sites of FSWritableFile functions, WritableFileWriter::DecideRateLimiterPriority() determines the rate_limiter_priority. The options (IOOptions) argument of FSWritableFile functions will be updated with the rate_limiter_priority. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9988 Test Plan: Add unit tests. Reviewed By: anand1976 Differential Revision: D36395159 Pulled By: gitbw95 fbshipit-source-id: a7c82fc29759139a1a07ec46c37dbf7e753474cf
2022-05-18 07:41:41 +00:00
cfd->imm()->Add(mem, &to_delete);
}
EventLogger event_logger(db_options_.info_log.get());
SnapshotChecker* snapshot_checker = nullptr; // not relavant
assert(memtable_ids.size() == num_mems);
uint64_t smallest_memtable_id = memtable_ids.front();
uint64_t flush_memtable_id = smallest_memtable_id + num_mems_to_flush - 1;
FlushJob flush_job(
dbname_, versions_->GetColumnFamilySet()->GetDefault(), db_options_,
*cfd->GetLatestMutableCFOptions(), flush_memtable_id, env_options_,
versions_.get(), &mutex_, &shutting_down_, {}, kMaxSequenceNumber,
snapshot_checker, &job_context, FlushReason::kTest, nullptr, nullptr,
nullptr, kNoCompression, db_options_.statistics.get(), &event_logger,
true, true /* sync_output_directory */, true /* write_manifest */,
Env::Priority::USER, nullptr /*IOTracer*/, empty_seqno_to_time_mapping_);
Set Write rate limiter priority dynamically and pass it to FS (#9988) Summary: ### Context: Background compactions and flush generate large reads and writes, and can be long running, especially for universal compaction. In some cases, this can impact foreground reads and writes by users. From the RocksDB perspective, there can be two kinds of rate limiters, the internal (native) one and the external one. - The internal (native) rate limiter is introduced in [the wiki](https://github.com/facebook/rocksdb/wiki/Rate-Limiter). Currently, only IO_LOW and IO_HIGH are used and they are set statically. - For the external rate limiter, in FSWritableFile functions, IOOptions is open for end users to set and get rate_limiter_priority for their own rate limiter. Currently, RocksDB doesn’t pass the rate_limiter_priority through IOOptions to the file system. ### Solution During the User Read, Flush write, Compaction read/write, the WriteController is used to determine whether DB writes are stalled or slowed down. The rate limiter priority (Env::IOPriority) can be determined accordingly. We decided to always pass the priority in IOOptions. What the file system does with it should be a contract between the user and the file system. We would like to set the rate limiter priority at file level, since the Flush/Compaction job level may be too coarse with multiple files and block IO level is too granular. **This PR is for the Write path.** The **Write:** dynamic priority for different state are listed as follows: | State | Normal | Delayed | Stalled | | ----- | ------ | ------- | ------- | | Flush | IO_HIGH | IO_USER | IO_USER | | Compaction | IO_LOW | IO_USER | IO_USER | Flush and Compaction writes share the same call path through BlockBaseTableWriter, WritableFileWriter, and FSWritableFile. When a new FSWritableFile object is created, its io_priority_ can be set dynamically based on the state of the WriteController. In WritableFileWriter, before the call sites of FSWritableFile functions, WritableFileWriter::DecideRateLimiterPriority() determines the rate_limiter_priority. The options (IOOptions) argument of FSWritableFile functions will be updated with the rate_limiter_priority. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9988 Test Plan: Add unit tests. Reviewed By: anand1976 Differential Revision: D36395159 Pulled By: gitbw95 fbshipit-source-id: a7c82fc29759139a1a07ec46c37dbf7e753474cf
2022-05-18 07:41:41 +00:00
// When the state from WriteController is normal.
ASSERT_EQ(flush_job.GetRateLimiterPriorityForWrite(), Env::IO_HIGH);
WriteController* write_controller =
flush_job.versions_->GetColumnFamilySet()->write_controller();
{
// When the state from WriteController is Delayed.
std::unique_ptr<WriteControllerToken> delay_token =
write_controller->GetDelayToken(1000000);
ASSERT_EQ(flush_job.GetRateLimiterPriorityForWrite(), Env::IO_USER);
}
{
// When the state from WriteController is Stopped.
std::unique_ptr<WriteControllerToken> stop_token =
write_controller->GetStopToken();
ASSERT_EQ(flush_job.GetRateLimiterPriorityForWrite(), Env::IO_USER);
}
}
Logically strip timestamp during flush (#11557) 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
2023-06-29 22:50:50 +00:00
// Test parameters:
// param 0): paranoid file check
// param 1): user-defined timestamp test mode
class FlushJobTimestampTest
: public FlushJobTestBase,
public testing::WithParamInterface<
std::tuple<bool, test::UserDefinedTimestampTestMode>> {
public:
FlushJobTimestampTest()
: FlushJobTestBase(test::PerThreadDBPath("flush_job_ts_gc_test"),
test::BytewiseComparatorWithU64TsWrapper()) {}
void AddKeyValueToMemtable(MemTable* memtable, std::string key, uint64_t ts,
SequenceNumber seq, ValueType value_type,
Slice value) {
std::string key_str(std::move(key));
PutFixed64(&key_str, ts);
Integrity protection for live updates to WriteBatch (#7748) Summary: This PR adds the foundation classes for key-value integrity protection and the first use case: protecting live updates from the source buffers added to `WriteBatch` through the destination buffer in `MemTable`. The width of the protection info is not yet configurable -- only eight bytes per key is supported. This PR allows users to enable protection by constructing `WriteBatch` with `protection_bytes_per_key == 8`. It does not yet expose a way for users to get integrity protection via other write APIs (e.g., `Put()`, `Merge()`, `Delete()`, etc.). The foundation classes (`ProtectionInfo.*`) embed the coverage info in their type, and provide `Protect.*()` and `Strip.*()` functions to navigate between types with different coverage. For making bytes per key configurable (for powers of two up to eight) in the future, these classes are templated on the unsigned integer type used to store the protection info. That integer contains the XOR'd result of hashes with independent seeds for all covered fields. For integer fields, the hash is computed on the raw unadjusted bytes, so the result is endian-dependent. The most significant bytes are truncated when the hash value (8 bytes) is wider than the protection integer. When `WriteBatch` is constructed with `protection_bytes_per_key == 8`, we hold a `ProtectionInfoKVOTC` (i.e., one that covers key, value, optype aka `ValueType`, timestamp, and CF ID) for each entry added to the batch. The protection info is generated from the original buffers passed by the user, as well as the original metadata generated internally. When writing to memtable, each entry is transformed to a `ProtectionInfoKVOTS` (i.e., dropping coverage of CF ID and adding coverage of sequence number), since at that point we know the sequence number, and have already selected a memtable corresponding to a particular CF. This protection info is verified once the entry is encoded in the `MemTable` buffer. Pull Request resolved: https://github.com/facebook/rocksdb/pull/7748 Test Plan: - an integration test to verify a wide variety of single-byte changes to the encoded `MemTable` buffer are caught - add to stress/crash test to verify it works in variety of configs/operations without intentional corruption - [deferred] unit tests for `ProtectionInfo.*` classes for edge cases like KV swap, `SliceParts` and `Slice` APIs are interchangeable, etc. Reviewed By: pdillinger Differential Revision: D25754492 Pulled By: ajkr fbshipit-source-id: e481bac6c03c2ab268be41359730f1ceb9964866
2021-01-29 20:17:17 +00:00
ASSERT_OK(memtable->Add(seq, value_type, key_str, value,
nullptr /* kv_prot_info */));
}
protected:
Logically strip timestamp during flush (#11557) 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
2023-06-29 22:50:50 +00:00
void SetUp() override {
paranoid_file_checks_ = std::get<0>(GetParam());
auto udt_test_mode = std::get<1>(GetParam());
persist_udt_ = test::ShouldPersistUDT(udt_test_mode);
FlushJobTestBase::SetUp();
}
static constexpr uint64_t kStartTs = 10;
static constexpr SequenceNumber kStartSeq = 0;
SequenceNumber curr_seq_{kStartSeq};
std::atomic<uint64_t> curr_ts_{kStartTs};
Logically strip timestamp during flush (#11557) 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
2023-06-29 22:50:50 +00:00
void CheckFileMetaData(ColumnFamilyData* cfd,
const InternalKey& expected_smallest,
const InternalKey& expected_largest,
const FileMetaData* meta_from_flush) const {
ASSERT_EQ(expected_smallest.Encode(), meta_from_flush->smallest.Encode());
ASSERT_EQ(expected_largest.Encode(), meta_from_flush->largest.Encode());
const VersionStorageInfo* storage_info = cfd->current()->storage_info();
const std::vector<FileMetaData*>& l0_files = storage_info->LevelFiles(0);
ASSERT_EQ(l0_files.size(), 1);
auto installed_file_meta = l0_files[0];
ASSERT_EQ(expected_smallest.Encode(),
installed_file_meta->smallest.Encode());
ASSERT_EQ(expected_largest.Encode(), installed_file_meta->largest.Encode());
}
Respect cutoff timestamp during flush (#11599) Summary: Make flush respect the cutoff timestamp `full_history_ts_low` as much as possible for the user-defined timestamps in Memtables only feature. We achieve this by not proceeding with the actual flushing but instead reschedule the same `FlushRequest` so a follow up flush job can continue with the check after some interval. This approach doesn't work well for atomic flush, so this feature currently is not supported in combination with atomic flush. Furthermore, this approach also requires a customized method to get the next immediately bigger user-defined timestamp. So currently it's limited to comparator that use uint64_t as the user-defined timestamp format. This support can be extended when we add such a customized method to `AdvancedColumnFamilyOptions`. For non atomic flush request, at any single time, a column family can only have as many as one FlushRequest for it in the `flush_queue_`. There is deduplication done at `FlushRequest` enqueueing(`SchedulePendingFlush`) and dequeueing time (`PopFirstFromFlushQueue`). We hold the db mutex between when a `FlushRequest` is popped from the queue and the same FlushRequest get rescheduled, so no other `FlushRequest` with a higher `max_memtable_id` can be added to the `flush_queue_` blocking us from re-enqueueing the same `FlushRequest`. Flush is continued nevertheless if there is risk of entering write stall mode had the flush being postponed, e.g. due to accumulation of write buffers, exceeding the `max_write_buffer_number` setting. When this happens, the newest user-defined timestamp in the involved Memtables need to be tracked and we use it to increase the `full_history_ts_low`, which is an inclusive cutoff timestamp for which RocksDB promises to keep all user-defined timestamps equal to and newer than it. Tet plan: ``` ./column_family_test --gtest_filter="*RetainUDT*" ./memtable_list_test --gtest_filter="*WithTimestamp*" ./flush_job_test --gtest_filter="*WithTimestamp*" ``` Pull Request resolved: https://github.com/facebook/rocksdb/pull/11599 Reviewed By: ajkr Differential Revision: D47561586 Pulled By: jowlyzhang fbshipit-source-id: 9400445f983dd6eac489e9dd0fb5d9b99637fe89
2023-07-26 23:25:06 +00:00
void CheckFullHistoryTsLow(ColumnFamilyData* cfd,
const std::string& expected_full_history_ts_low) {
ASSERT_EQ(expected_full_history_ts_low, cfd->GetFullHistoryTsLow());
}
};
Logically strip timestamp during flush (#11557) 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
2023-06-29 22:50:50 +00:00
TEST_P(FlushJobTimestampTest, AllKeysExpired) {
ColumnFamilyData* cfd = versions_->GetColumnFamilySet()->GetDefault();
autovector<MemTable*> to_delete;
{
MemTable* new_mem = cfd->ConstructNewMemtable(
*cfd->GetLatestMutableCFOptions(), kMaxSequenceNumber);
new_mem->Ref();
for (int i = 0; i < 100; ++i) {
uint64_t ts = curr_ts_.fetch_add(1);
SequenceNumber seq = (curr_seq_++);
AddKeyValueToMemtable(new_mem, test::EncodeInt(0), ts, seq,
ValueType::kTypeValue, "0_value");
}
uint64_t ts = curr_ts_.fetch_add(1);
SequenceNumber seq = (curr_seq_++);
AddKeyValueToMemtable(new_mem, test::EncodeInt(0), ts, seq,
ValueType::kTypeDeletionWithTimestamp, "");
Fragment memtable range tombstone in the write path (#10380) Summary: - Right now each read fragments the memtable range tombstones https://github.com/facebook/rocksdb/issues/4808. This PR explores the idea of fragmenting memtable range tombstones in the write path and reads can just read this cached fragmented tombstone without any fragmenting cost. This PR only does the caching for immutable memtable, and does so right before a memtable is added to an immutable memtable list. The fragmentation is done without holding mutex to minimize its performance impact. - db_bench is updated to print out the number of range deletions executed if there is any. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10380 Test Plan: - CI, added asserts in various places to check whether a fragmented range tombstone list should have been constructed. - Benchmark: as this PR only optimizes immutable memtable path, the number of writes in the benchmark is chosen such an immutable memtable is created and range tombstones are in that memtable. ``` single thread: ./db_bench --benchmarks=fillrandom,readrandom --writes_per_range_tombstone=1 --max_write_buffer_number=100 --min_write_buffer_number_to_merge=100 --writes=500000 --reads=100000 --max_num_range_tombstones=100 multi_thread ./db_bench --benchmarks=fillrandom,readrandom --writes_per_range_tombstone=1 --max_write_buffer_number=100 --min_write_buffer_number_to_merge=100 --writes=15000 --reads=20000 --threads=32 --max_num_range_tombstones=100 ``` Commit 99cdf16464a057ca44de2f747541dedf651bae9e is included in benchmark result. It was an earlier attempt where tombstones are fragmented for each write operation. Reader threads share it using a shared_ptr which would slow down multi-thread read performance as seen in benchmark results. Results are averaged over 5 runs. Single thread result: | Max # tombstones | main fillrandom micros/op | 99cdf16464a057ca44de2f747541dedf651bae9e | Post PR | main readrandom micros/op | 99cdf16464a057ca44de2f747541dedf651bae9e | Post PR | | ------------- | ------------- |------------- |------------- |------------- |------------- |------------- | | 0 |6.68 |6.57 |6.72 |4.72 |4.79 |4.54 | | 1 |6.67 |6.58 |6.62 |5.41 |4.74 |4.72 | | 10 |6.59 |6.5 |6.56 |7.83 |4.69 |4.59 | | 100 |6.62 |6.75 |6.58 |29.57 |5.04 |5.09 | | 1000 |6.54 |6.82 |6.61 |320.33 |5.22 |5.21 | 32-thread result: note that "Max # tombstones" is per thread. | Max # tombstones | main fillrandom micros/op | 99cdf16464a057ca44de2f747541dedf651bae9e | Post PR | main readrandom micros/op | 99cdf16464a057ca44de2f747541dedf651bae9e | Post PR | | ------------- | ------------- |------------- |------------- |------------- |------------- |------------- | | 0 |234.52 |260.25 |239.42 |5.06 |5.38 |5.09 | | 1 |236.46 |262.0 |231.1 |19.57 |22.14 |5.45 | | 10 |236.95 |263.84 |251.49 |151.73 |21.61 |5.73 | | 100 |268.16 |296.8 |280.13 |2308.52 |22.27 |6.57 | Reviewed By: ajkr Differential Revision: D37916564 Pulled By: cbi42 fbshipit-source-id: 05d6d2e16df26c374c57ddcca13a5bfe9d5b731e
2022-08-05 19:02:33 +00:00
new_mem->ConstructFragmentedRangeTombstones();
cfd->imm()->Add(new_mem, &to_delete);
}
std::vector<SequenceNumber> snapshots;
constexpr SnapshotChecker* const snapshot_checker = nullptr;
JobContext job_context(0);
EventLogger event_logger(db_options_.info_log.get());
std::string full_history_ts_low;
PutFixed64(&full_history_ts_low, std::numeric_limits<uint64_t>::max());
Respect cutoff timestamp during flush (#11599) Summary: Make flush respect the cutoff timestamp `full_history_ts_low` as much as possible for the user-defined timestamps in Memtables only feature. We achieve this by not proceeding with the actual flushing but instead reschedule the same `FlushRequest` so a follow up flush job can continue with the check after some interval. This approach doesn't work well for atomic flush, so this feature currently is not supported in combination with atomic flush. Furthermore, this approach also requires a customized method to get the next immediately bigger user-defined timestamp. So currently it's limited to comparator that use uint64_t as the user-defined timestamp format. This support can be extended when we add such a customized method to `AdvancedColumnFamilyOptions`. For non atomic flush request, at any single time, a column family can only have as many as one FlushRequest for it in the `flush_queue_`. There is deduplication done at `FlushRequest` enqueueing(`SchedulePendingFlush`) and dequeueing time (`PopFirstFromFlushQueue`). We hold the db mutex between when a `FlushRequest` is popped from the queue and the same FlushRequest get rescheduled, so no other `FlushRequest` with a higher `max_memtable_id` can be added to the `flush_queue_` blocking us from re-enqueueing the same `FlushRequest`. Flush is continued nevertheless if there is risk of entering write stall mode had the flush being postponed, e.g. due to accumulation of write buffers, exceeding the `max_write_buffer_number` setting. When this happens, the newest user-defined timestamp in the involved Memtables need to be tracked and we use it to increase the `full_history_ts_low`, which is an inclusive cutoff timestamp for which RocksDB promises to keep all user-defined timestamps equal to and newer than it. Tet plan: ``` ./column_family_test --gtest_filter="*RetainUDT*" ./memtable_list_test --gtest_filter="*WithTimestamp*" ./flush_job_test --gtest_filter="*WithTimestamp*" ``` Pull Request resolved: https://github.com/facebook/rocksdb/pull/11599 Reviewed By: ajkr Differential Revision: D47561586 Pulled By: jowlyzhang fbshipit-source-id: 9400445f983dd6eac489e9dd0fb5d9b99637fe89
2023-07-26 23:25:06 +00:00
cfd->SetFullHistoryTsLow(full_history_ts_low);
FlushJob flush_job(
dbname_, cfd, db_options_, *cfd->GetLatestMutableCFOptions(),
std::numeric_limits<uint64_t>::max() /* memtable_id */, env_options_,
versions_.get(), &mutex_, &shutting_down_, snapshots, kMaxSequenceNumber,
snapshot_checker, &job_context, FlushReason::kTest, nullptr, nullptr,
nullptr, kNoCompression, db_options_.statistics.get(), &event_logger,
true, true /* sync_output_directory */, true /* write_manifest */,
Env::Priority::USER, nullptr /*IOTracer*/, empty_seqno_to_time_mapping_,
/*db_id=*/"",
/*db_session_id=*/"", full_history_ts_low);
FileMetaData fmeta;
mutex_.Lock();
flush_job.PickMemTable();
ASSERT_OK(flush_job.Run(/*prep_tracker=*/nullptr, &fmeta));
mutex_.Unlock();
{
std::string key = test::EncodeInt(0);
Logically strip timestamp during flush (#11557) 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
2023-06-29 22:50:50 +00:00
if (!persist_udt_) {
// When `AdvancedColumnFamilyOptions.persist_user_defined_timestamps` flag
// is set to false. The user-defined timestamp is stripped from user key
// during flush, making the user key logically containing the minimum
// timestamp.
key.append(test::EncodeInt(0));
} else {
key.append(test::EncodeInt(curr_ts_.load(std::memory_order_relaxed) - 1));
}
InternalKey ikey(key, curr_seq_ - 1, ValueType::kTypeDeletionWithTimestamp);
Logically strip timestamp during flush (#11557) 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
2023-06-29 22:50:50 +00:00
CheckFileMetaData(cfd, ikey, ikey, &fmeta);
Respect cutoff timestamp during flush (#11599) Summary: Make flush respect the cutoff timestamp `full_history_ts_low` as much as possible for the user-defined timestamps in Memtables only feature. We achieve this by not proceeding with the actual flushing but instead reschedule the same `FlushRequest` so a follow up flush job can continue with the check after some interval. This approach doesn't work well for atomic flush, so this feature currently is not supported in combination with atomic flush. Furthermore, this approach also requires a customized method to get the next immediately bigger user-defined timestamp. So currently it's limited to comparator that use uint64_t as the user-defined timestamp format. This support can be extended when we add such a customized method to `AdvancedColumnFamilyOptions`. For non atomic flush request, at any single time, a column family can only have as many as one FlushRequest for it in the `flush_queue_`. There is deduplication done at `FlushRequest` enqueueing(`SchedulePendingFlush`) and dequeueing time (`PopFirstFromFlushQueue`). We hold the db mutex between when a `FlushRequest` is popped from the queue and the same FlushRequest get rescheduled, so no other `FlushRequest` with a higher `max_memtable_id` can be added to the `flush_queue_` blocking us from re-enqueueing the same `FlushRequest`. Flush is continued nevertheless if there is risk of entering write stall mode had the flush being postponed, e.g. due to accumulation of write buffers, exceeding the `max_write_buffer_number` setting. When this happens, the newest user-defined timestamp in the involved Memtables need to be tracked and we use it to increase the `full_history_ts_low`, which is an inclusive cutoff timestamp for which RocksDB promises to keep all user-defined timestamps equal to and newer than it. Tet plan: ``` ./column_family_test --gtest_filter="*RetainUDT*" ./memtable_list_test --gtest_filter="*WithTimestamp*" ./flush_job_test --gtest_filter="*WithTimestamp*" ``` Pull Request resolved: https://github.com/facebook/rocksdb/pull/11599 Reviewed By: ajkr Differential Revision: D47561586 Pulled By: jowlyzhang fbshipit-source-id: 9400445f983dd6eac489e9dd0fb5d9b99637fe89
2023-07-26 23:25:06 +00:00
CheckFullHistoryTsLow(cfd, full_history_ts_low);
}
job_context.Clean();
ASSERT_TRUE(to_delete.empty());
}
Logically strip timestamp during flush (#11557) 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
2023-06-29 22:50:50 +00:00
TEST_P(FlushJobTimestampTest, NoKeyExpired) {
ColumnFamilyData* cfd = versions_->GetColumnFamilySet()->GetDefault();
autovector<MemTable*> to_delete;
{
MemTable* new_mem = cfd->ConstructNewMemtable(
*cfd->GetLatestMutableCFOptions(), kMaxSequenceNumber);
new_mem->Ref();
for (int i = 0; i < 100; ++i) {
uint64_t ts = curr_ts_.fetch_add(1);
SequenceNumber seq = (curr_seq_++);
AddKeyValueToMemtable(new_mem, test::EncodeInt(0), ts, seq,
ValueType::kTypeValue, "0_value");
}
Fragment memtable range tombstone in the write path (#10380) Summary: - Right now each read fragments the memtable range tombstones https://github.com/facebook/rocksdb/issues/4808. This PR explores the idea of fragmenting memtable range tombstones in the write path and reads can just read this cached fragmented tombstone without any fragmenting cost. This PR only does the caching for immutable memtable, and does so right before a memtable is added to an immutable memtable list. The fragmentation is done without holding mutex to minimize its performance impact. - db_bench is updated to print out the number of range deletions executed if there is any. Pull Request resolved: https://github.com/facebook/rocksdb/pull/10380 Test Plan: - CI, added asserts in various places to check whether a fragmented range tombstone list should have been constructed. - Benchmark: as this PR only optimizes immutable memtable path, the number of writes in the benchmark is chosen such an immutable memtable is created and range tombstones are in that memtable. ``` single thread: ./db_bench --benchmarks=fillrandom,readrandom --writes_per_range_tombstone=1 --max_write_buffer_number=100 --min_write_buffer_number_to_merge=100 --writes=500000 --reads=100000 --max_num_range_tombstones=100 multi_thread ./db_bench --benchmarks=fillrandom,readrandom --writes_per_range_tombstone=1 --max_write_buffer_number=100 --min_write_buffer_number_to_merge=100 --writes=15000 --reads=20000 --threads=32 --max_num_range_tombstones=100 ``` Commit 99cdf16464a057ca44de2f747541dedf651bae9e is included in benchmark result. It was an earlier attempt where tombstones are fragmented for each write operation. Reader threads share it using a shared_ptr which would slow down multi-thread read performance as seen in benchmark results. Results are averaged over 5 runs. Single thread result: | Max # tombstones | main fillrandom micros/op | 99cdf16464a057ca44de2f747541dedf651bae9e | Post PR | main readrandom micros/op | 99cdf16464a057ca44de2f747541dedf651bae9e | Post PR | | ------------- | ------------- |------------- |------------- |------------- |------------- |------------- | | 0 |6.68 |6.57 |6.72 |4.72 |4.79 |4.54 | | 1 |6.67 |6.58 |6.62 |5.41 |4.74 |4.72 | | 10 |6.59 |6.5 |6.56 |7.83 |4.69 |4.59 | | 100 |6.62 |6.75 |6.58 |29.57 |5.04 |5.09 | | 1000 |6.54 |6.82 |6.61 |320.33 |5.22 |5.21 | 32-thread result: note that "Max # tombstones" is per thread. | Max # tombstones | main fillrandom micros/op | 99cdf16464a057ca44de2f747541dedf651bae9e | Post PR | main readrandom micros/op | 99cdf16464a057ca44de2f747541dedf651bae9e | Post PR | | ------------- | ------------- |------------- |------------- |------------- |------------- |------------- | | 0 |234.52 |260.25 |239.42 |5.06 |5.38 |5.09 | | 1 |236.46 |262.0 |231.1 |19.57 |22.14 |5.45 | | 10 |236.95 |263.84 |251.49 |151.73 |21.61 |5.73 | | 100 |268.16 |296.8 |280.13 |2308.52 |22.27 |6.57 | Reviewed By: ajkr Differential Revision: D37916564 Pulled By: cbi42 fbshipit-source-id: 05d6d2e16df26c374c57ddcca13a5bfe9d5b731e
2022-08-05 19:02:33 +00:00
new_mem->ConstructFragmentedRangeTombstones();
cfd->imm()->Add(new_mem, &to_delete);
}
std::vector<SequenceNumber> snapshots;
SnapshotChecker* const snapshot_checker = nullptr;
JobContext job_context(0);
EventLogger event_logger(db_options_.info_log.get());
std::string full_history_ts_low;
PutFixed64(&full_history_ts_low, 0);
Respect cutoff timestamp during flush (#11599) Summary: Make flush respect the cutoff timestamp `full_history_ts_low` as much as possible for the user-defined timestamps in Memtables only feature. We achieve this by not proceeding with the actual flushing but instead reschedule the same `FlushRequest` so a follow up flush job can continue with the check after some interval. This approach doesn't work well for atomic flush, so this feature currently is not supported in combination with atomic flush. Furthermore, this approach also requires a customized method to get the next immediately bigger user-defined timestamp. So currently it's limited to comparator that use uint64_t as the user-defined timestamp format. This support can be extended when we add such a customized method to `AdvancedColumnFamilyOptions`. For non atomic flush request, at any single time, a column family can only have as many as one FlushRequest for it in the `flush_queue_`. There is deduplication done at `FlushRequest` enqueueing(`SchedulePendingFlush`) and dequeueing time (`PopFirstFromFlushQueue`). We hold the db mutex between when a `FlushRequest` is popped from the queue and the same FlushRequest get rescheduled, so no other `FlushRequest` with a higher `max_memtable_id` can be added to the `flush_queue_` blocking us from re-enqueueing the same `FlushRequest`. Flush is continued nevertheless if there is risk of entering write stall mode had the flush being postponed, e.g. due to accumulation of write buffers, exceeding the `max_write_buffer_number` setting. When this happens, the newest user-defined timestamp in the involved Memtables need to be tracked and we use it to increase the `full_history_ts_low`, which is an inclusive cutoff timestamp for which RocksDB promises to keep all user-defined timestamps equal to and newer than it. Tet plan: ``` ./column_family_test --gtest_filter="*RetainUDT*" ./memtable_list_test --gtest_filter="*WithTimestamp*" ./flush_job_test --gtest_filter="*WithTimestamp*" ``` Pull Request resolved: https://github.com/facebook/rocksdb/pull/11599 Reviewed By: ajkr Differential Revision: D47561586 Pulled By: jowlyzhang fbshipit-source-id: 9400445f983dd6eac489e9dd0fb5d9b99637fe89
2023-07-26 23:25:06 +00:00
cfd->SetFullHistoryTsLow(full_history_ts_low);
FlushJob flush_job(
dbname_, cfd, db_options_, *cfd->GetLatestMutableCFOptions(),
std::numeric_limits<uint64_t>::max() /* memtable_id */, env_options_,
versions_.get(), &mutex_, &shutting_down_, snapshots, kMaxSequenceNumber,
snapshot_checker, &job_context, FlushReason::kTest, nullptr, nullptr,
nullptr, kNoCompression, db_options_.statistics.get(), &event_logger,
true, true /* sync_output_directory */, true /* write_manifest */,
Env::Priority::USER, nullptr /*IOTracer*/, empty_seqno_to_time_mapping_,
/*db_id=*/"",
/*db_session_id=*/"", full_history_ts_low);
FileMetaData fmeta;
mutex_.Lock();
flush_job.PickMemTable();
ASSERT_OK(flush_job.Run(/*prep_tracker=*/nullptr, &fmeta));
mutex_.Unlock();
{
std::string ukey = test::EncodeInt(0);
Logically strip timestamp during flush (#11557) 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
2023-06-29 22:50:50 +00:00
std::string smallest_key;
std::string largest_key;
Respect cutoff timestamp during flush (#11599) Summary: Make flush respect the cutoff timestamp `full_history_ts_low` as much as possible for the user-defined timestamps in Memtables only feature. We achieve this by not proceeding with the actual flushing but instead reschedule the same `FlushRequest` so a follow up flush job can continue with the check after some interval. This approach doesn't work well for atomic flush, so this feature currently is not supported in combination with atomic flush. Furthermore, this approach also requires a customized method to get the next immediately bigger user-defined timestamp. So currently it's limited to comparator that use uint64_t as the user-defined timestamp format. This support can be extended when we add such a customized method to `AdvancedColumnFamilyOptions`. For non atomic flush request, at any single time, a column family can only have as many as one FlushRequest for it in the `flush_queue_`. There is deduplication done at `FlushRequest` enqueueing(`SchedulePendingFlush`) and dequeueing time (`PopFirstFromFlushQueue`). We hold the db mutex between when a `FlushRequest` is popped from the queue and the same FlushRequest get rescheduled, so no other `FlushRequest` with a higher `max_memtable_id` can be added to the `flush_queue_` blocking us from re-enqueueing the same `FlushRequest`. Flush is continued nevertheless if there is risk of entering write stall mode had the flush being postponed, e.g. due to accumulation of write buffers, exceeding the `max_write_buffer_number` setting. When this happens, the newest user-defined timestamp in the involved Memtables need to be tracked and we use it to increase the `full_history_ts_low`, which is an inclusive cutoff timestamp for which RocksDB promises to keep all user-defined timestamps equal to and newer than it. Tet plan: ``` ./column_family_test --gtest_filter="*RetainUDT*" ./memtable_list_test --gtest_filter="*WithTimestamp*" ./flush_job_test --gtest_filter="*WithTimestamp*" ``` Pull Request resolved: https://github.com/facebook/rocksdb/pull/11599 Reviewed By: ajkr Differential Revision: D47561586 Pulled By: jowlyzhang fbshipit-source-id: 9400445f983dd6eac489e9dd0fb5d9b99637fe89
2023-07-26 23:25:06 +00:00
std::string expected_full_history_ts_low;
Logically strip timestamp during flush (#11557) 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
2023-06-29 22:50:50 +00:00
if (!persist_udt_) {
// When `AdvancedColumnFamilyOptions.persist_user_defined_timestamps` flag
// is set to false. The user-defined timestamp is stripped from user key
// during flush, making the user key logically containing the minimum
// timestamp, which is hardcoded to be all zeros for now.
smallest_key = ukey + test::EncodeInt(0);
largest_key = ukey + test::EncodeInt(0);
Respect cutoff timestamp during flush (#11599) Summary: Make flush respect the cutoff timestamp `full_history_ts_low` as much as possible for the user-defined timestamps in Memtables only feature. We achieve this by not proceeding with the actual flushing but instead reschedule the same `FlushRequest` so a follow up flush job can continue with the check after some interval. This approach doesn't work well for atomic flush, so this feature currently is not supported in combination with atomic flush. Furthermore, this approach also requires a customized method to get the next immediately bigger user-defined timestamp. So currently it's limited to comparator that use uint64_t as the user-defined timestamp format. This support can be extended when we add such a customized method to `AdvancedColumnFamilyOptions`. For non atomic flush request, at any single time, a column family can only have as many as one FlushRequest for it in the `flush_queue_`. There is deduplication done at `FlushRequest` enqueueing(`SchedulePendingFlush`) and dequeueing time (`PopFirstFromFlushQueue`). We hold the db mutex between when a `FlushRequest` is popped from the queue and the same FlushRequest get rescheduled, so no other `FlushRequest` with a higher `max_memtable_id` can be added to the `flush_queue_` blocking us from re-enqueueing the same `FlushRequest`. Flush is continued nevertheless if there is risk of entering write stall mode had the flush being postponed, e.g. due to accumulation of write buffers, exceeding the `max_write_buffer_number` setting. When this happens, the newest user-defined timestamp in the involved Memtables need to be tracked and we use it to increase the `full_history_ts_low`, which is an inclusive cutoff timestamp for which RocksDB promises to keep all user-defined timestamps equal to and newer than it. Tet plan: ``` ./column_family_test --gtest_filter="*RetainUDT*" ./memtable_list_test --gtest_filter="*WithTimestamp*" ./flush_job_test --gtest_filter="*WithTimestamp*" ``` Pull Request resolved: https://github.com/facebook/rocksdb/pull/11599 Reviewed By: ajkr Differential Revision: D47561586 Pulled By: jowlyzhang fbshipit-source-id: 9400445f983dd6eac489e9dd0fb5d9b99637fe89
2023-07-26 23:25:06 +00:00
// When not all keys have expired and `persist_user_defined_timestamps` is
// false. UDTs will be removed during flush, `full_history_ts_low` should
// be automatically increased to above the effective cutoff UDT in the
// flush.
PutFixed64(&expected_full_history_ts_low, curr_ts_.fetch_add(1));
Logically strip timestamp during flush (#11557) 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
2023-06-29 22:50:50 +00:00
} else {
smallest_key =
ukey + test::EncodeInt(curr_ts_.load(std::memory_order_relaxed) - 1);
largest_key = ukey + test::EncodeInt(kStartTs);
Respect cutoff timestamp during flush (#11599) Summary: Make flush respect the cutoff timestamp `full_history_ts_low` as much as possible for the user-defined timestamps in Memtables only feature. We achieve this by not proceeding with the actual flushing but instead reschedule the same `FlushRequest` so a follow up flush job can continue with the check after some interval. This approach doesn't work well for atomic flush, so this feature currently is not supported in combination with atomic flush. Furthermore, this approach also requires a customized method to get the next immediately bigger user-defined timestamp. So currently it's limited to comparator that use uint64_t as the user-defined timestamp format. This support can be extended when we add such a customized method to `AdvancedColumnFamilyOptions`. For non atomic flush request, at any single time, a column family can only have as many as one FlushRequest for it in the `flush_queue_`. There is deduplication done at `FlushRequest` enqueueing(`SchedulePendingFlush`) and dequeueing time (`PopFirstFromFlushQueue`). We hold the db mutex between when a `FlushRequest` is popped from the queue and the same FlushRequest get rescheduled, so no other `FlushRequest` with a higher `max_memtable_id` can be added to the `flush_queue_` blocking us from re-enqueueing the same `FlushRequest`. Flush is continued nevertheless if there is risk of entering write stall mode had the flush being postponed, e.g. due to accumulation of write buffers, exceeding the `max_write_buffer_number` setting. When this happens, the newest user-defined timestamp in the involved Memtables need to be tracked and we use it to increase the `full_history_ts_low`, which is an inclusive cutoff timestamp for which RocksDB promises to keep all user-defined timestamps equal to and newer than it. Tet plan: ``` ./column_family_test --gtest_filter="*RetainUDT*" ./memtable_list_test --gtest_filter="*WithTimestamp*" ./flush_job_test --gtest_filter="*WithTimestamp*" ``` Pull Request resolved: https://github.com/facebook/rocksdb/pull/11599 Reviewed By: ajkr Differential Revision: D47561586 Pulled By: jowlyzhang fbshipit-source-id: 9400445f983dd6eac489e9dd0fb5d9b99637fe89
2023-07-26 23:25:06 +00:00
expected_full_history_ts_low = full_history_ts_low;
Logically strip timestamp during flush (#11557) 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
2023-06-29 22:50:50 +00:00
}
InternalKey smallest(smallest_key, curr_seq_ - 1, ValueType::kTypeValue);
InternalKey largest(largest_key, kStartSeq, ValueType::kTypeValue);
Logically strip timestamp during flush (#11557) 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
2023-06-29 22:50:50 +00:00
CheckFileMetaData(cfd, smallest, largest, &fmeta);
Respect cutoff timestamp during flush (#11599) Summary: Make flush respect the cutoff timestamp `full_history_ts_low` as much as possible for the user-defined timestamps in Memtables only feature. We achieve this by not proceeding with the actual flushing but instead reschedule the same `FlushRequest` so a follow up flush job can continue with the check after some interval. This approach doesn't work well for atomic flush, so this feature currently is not supported in combination with atomic flush. Furthermore, this approach also requires a customized method to get the next immediately bigger user-defined timestamp. So currently it's limited to comparator that use uint64_t as the user-defined timestamp format. This support can be extended when we add such a customized method to `AdvancedColumnFamilyOptions`. For non atomic flush request, at any single time, a column family can only have as many as one FlushRequest for it in the `flush_queue_`. There is deduplication done at `FlushRequest` enqueueing(`SchedulePendingFlush`) and dequeueing time (`PopFirstFromFlushQueue`). We hold the db mutex between when a `FlushRequest` is popped from the queue and the same FlushRequest get rescheduled, so no other `FlushRequest` with a higher `max_memtable_id` can be added to the `flush_queue_` blocking us from re-enqueueing the same `FlushRequest`. Flush is continued nevertheless if there is risk of entering write stall mode had the flush being postponed, e.g. due to accumulation of write buffers, exceeding the `max_write_buffer_number` setting. When this happens, the newest user-defined timestamp in the involved Memtables need to be tracked and we use it to increase the `full_history_ts_low`, which is an inclusive cutoff timestamp for which RocksDB promises to keep all user-defined timestamps equal to and newer than it. Tet plan: ``` ./column_family_test --gtest_filter="*RetainUDT*" ./memtable_list_test --gtest_filter="*WithTimestamp*" ./flush_job_test --gtest_filter="*WithTimestamp*" ``` Pull Request resolved: https://github.com/facebook/rocksdb/pull/11599 Reviewed By: ajkr Differential Revision: D47561586 Pulled By: jowlyzhang fbshipit-source-id: 9400445f983dd6eac489e9dd0fb5d9b99637fe89
2023-07-26 23:25:06 +00:00
CheckFullHistoryTsLow(cfd, expected_full_history_ts_low);
}
job_context.Clean();
ASSERT_TRUE(to_delete.empty());
}
Logically strip timestamp during flush (#11557) 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
2023-06-29 22:50:50 +00:00
// Param 0: paranoid file check
// Param 1: test mode for the user-defined timestamp feature
INSTANTIATE_TEST_CASE_P(
FlushJobTimestampTest, FlushJobTimestampTest,
::testing::Combine(
::testing::Bool(),
::testing::Values(
test::UserDefinedTimestampTestMode::kStripUserDefinedTimestamp,
test::UserDefinedTimestampTestMode::kNormal)));
} // namespace ROCKSDB_NAMESPACE
rocksdb: switch to gtest Summary: Our existing test notation is very similar to what is used in gtest. It makes it easy to adopt what is different. In this diff I modify existing [[ https://code.google.com/p/googletest/wiki/Primer#Test_Fixtures:_Using_the_Same_Data_Configuration_for_Multiple_Te | test fixture ]] classes to inherit from `testing::Test`. Also for unit tests that use fixture class, `TEST` is replaced with `TEST_F` as required in gtest. There are several custom `main` functions in our existing tests. To make this transition easier, I modify all `main` functions to fallow gtest notation. But eventually we can remove them and use implementation of `main` that gtest provides. ```lang=bash % cat ~/transform #!/bin/sh files=$(git ls-files '*test\.cc') for file in $files do if grep -q "rocksdb::test::RunAllTests()" $file then if grep -Eq '^class \w+Test {' $file then perl -pi -e 's/^(class \w+Test) {/${1}: public testing::Test {/g' $file perl -pi -e 's/^(TEST)/${1}_F/g' $file fi perl -pi -e 's/(int main.*\{)/${1}::testing::InitGoogleTest(&argc, argv);/g' $file perl -pi -e 's/rocksdb::test::RunAllTests/RUN_ALL_TESTS/g' $file fi done % sh ~/transform % make format ``` Second iteration of this diff contains only scripted changes. Third iteration contains manual changes to fix last errors and make it compilable. Test Plan: Build and notice no errors. ```lang=bash % USE_CLANG=1 make check -j55 ``` Tests are still testing. Reviewers: meyering, sdong, rven, igor Reviewed By: igor Subscribers: dhruba, leveldb Differential Revision: https://reviews.facebook.net/D35157
2015-03-17 21:08:00 +00:00
int main(int argc, char** argv) {
ROCKSDB_NAMESPACE::port::InstallStackTraceHandler();
rocksdb: switch to gtest Summary: Our existing test notation is very similar to what is used in gtest. It makes it easy to adopt what is different. In this diff I modify existing [[ https://code.google.com/p/googletest/wiki/Primer#Test_Fixtures:_Using_the_Same_Data_Configuration_for_Multiple_Te | test fixture ]] classes to inherit from `testing::Test`. Also for unit tests that use fixture class, `TEST` is replaced with `TEST_F` as required in gtest. There are several custom `main` functions in our existing tests. To make this transition easier, I modify all `main` functions to fallow gtest notation. But eventually we can remove them and use implementation of `main` that gtest provides. ```lang=bash % cat ~/transform #!/bin/sh files=$(git ls-files '*test\.cc') for file in $files do if grep -q "rocksdb::test::RunAllTests()" $file then if grep -Eq '^class \w+Test {' $file then perl -pi -e 's/^(class \w+Test) {/${1}: public testing::Test {/g' $file perl -pi -e 's/^(TEST)/${1}_F/g' $file fi perl -pi -e 's/(int main.*\{)/${1}::testing::InitGoogleTest(&argc, argv);/g' $file perl -pi -e 's/rocksdb::test::RunAllTests/RUN_ALL_TESTS/g' $file fi done % sh ~/transform % make format ``` Second iteration of this diff contains only scripted changes. Third iteration contains manual changes to fix last errors and make it compilable. Test Plan: Build and notice no errors. ```lang=bash % USE_CLANG=1 make check -j55 ``` Tests are still testing. Reviewers: meyering, sdong, rven, igor Reviewed By: igor Subscribers: dhruba, leveldb Differential Revision: https://reviews.facebook.net/D35157
2015-03-17 21:08:00 +00:00
::testing::InitGoogleTest(&argc, argv);
return RUN_ALL_TESTS();
}