rocksdb/db/plain_table_db_test.cc

1349 lines
47 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).
//
// Copyright (c) 2011 The LevelDB Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file. See the AUTHORS file for names of contributors.
#include <algorithm>
#include <set>
#include "db/db_impl/db_impl.h"
#include "db/version_set.h"
#include "db/write_batch_internal.h"
#include "file/filename.h"
#include "rocksdb/cache.h"
#include "rocksdb/compaction_filter.h"
#include "rocksdb/db.h"
#include "rocksdb/env.h"
#include "rocksdb/filter_policy.h"
#include "rocksdb/slice_transform.h"
#include "rocksdb/table.h"
#include "table/meta_blocks.h"
#include "table/plain/plain_table_bloom.h"
#include "table/plain/plain_table_factory.h"
#include "table/plain/plain_table_key_coding.h"
#include "table/plain/plain_table_reader.h"
#include "table/table_builder.h"
#include "test_util/testharness.h"
#include "test_util/testutil.h"
#include "util/cast_util.h"
#include "util/hash.h"
#include "util/mutexlock.h"
#include "util/random.h"
#include "util/string_util.h"
#include "utilities/merge_operators.h"
namespace ROCKSDB_NAMESPACE {
class PlainTableKeyDecoderTest : public testing::Test {};
TEST_F(PlainTableKeyDecoderTest, ReadNonMmap) {
Random rnd(301);
const uint32_t kLength = 2222;
std::string tmp = rnd.RandomString(kLength);
Slice contents(tmp);
test::StringSource* string_source =
new test::StringSource(contents, 0, false);
std::unique_ptr<FSRandomAccessFile> holder(string_source);
std::unique_ptr<RandomAccessFileReader> file_reader(
new RandomAccessFileReader(std::move(holder), "test"));
std::unique_ptr<PlainTableReaderFileInfo> file_info(
new PlainTableReaderFileInfo(std::move(file_reader), EnvOptions(),
kLength));
{
PlainTableFileReader reader(file_info.get());
const uint32_t kReadSize = 77;
for (uint32_t pos = 0; pos < kLength; pos += kReadSize) {
uint32_t read_size = std::min(kLength - pos, kReadSize);
Slice out;
ASSERT_TRUE(reader.Read(pos, read_size, &out));
ASSERT_EQ(0, out.compare(tmp.substr(pos, read_size)));
}
ASSERT_LT(uint32_t(string_source->total_reads()), kLength / kReadSize / 2);
}
std::vector<std::vector<std::pair<uint32_t, uint32_t>>> reads = {
{{600, 30}, {590, 30}, {600, 20}, {600, 40}},
{{800, 20}, {100, 20}, {500, 20}, {1500, 20}, {100, 20}, {80, 20}},
{{1000, 20}, {500, 20}, {1000, 50}},
{{1000, 20}, {500, 20}, {500, 20}},
{{1000, 20}, {500, 20}, {200, 20}, {500, 20}},
{{1000, 20}, {500, 20}, {200, 20}, {1000, 50}},
{{600, 500}, {610, 20}, {100, 20}},
{{500, 100}, {490, 100}, {550, 50}},
};
std::vector<int> num_file_reads = {2, 6, 2, 2, 4, 3, 2, 2};
for (size_t i = 0; i < reads.size(); i++) {
string_source->set_total_reads(0);
PlainTableFileReader reader(file_info.get());
for (auto p : reads[i]) {
Slice out;
ASSERT_TRUE(reader.Read(p.first, p.second, &out));
ASSERT_EQ(0, out.compare(tmp.substr(p.first, p.second)));
}
ASSERT_EQ(num_file_reads[i], string_source->total_reads());
}
}
class PlainTableDBTest : public testing::Test,
public testing::WithParamInterface<bool> {
2014-01-03 18:53:21 +00:00
protected:
private:
std::string dbname_;
Env* env_;
DB* db_;
bool mmap_mode_;
Options last_options_;
2014-01-03 18:53:21 +00:00
public:
PlainTableDBTest() : env_(Env::Default()) {}
~PlainTableDBTest() override {
delete db_;
rocksdb: Replace ASSERT* with EXPECT* in functions that does not return void value Summary: gtest does not use exceptions to fail a unit test by design, and `ASSERT*`s are implemented using `return`. As a consequence we cannot use `ASSERT*` in a function that does not return `void` value ([[ https://code.google.com/p/googletest/wiki/AdvancedGuide#Assertion_Placement | 1]]), and have to fix our existing code. This diff does this in a generic way, with no manual changes. In order to detect all existing `ASSERT*` that are used in functions that doesn't return void value, I change the code to generate compile errors for such cases. In `util/testharness.h` I defined `EXPECT*` assertions, the same way as `ASSERT*`, and redefined `ASSERT*` to return `void`. Then executed: ```lang=bash % USE_CLANG=1 make all -j55 -k 2> build.log % perl -naF: -e 'print "-- -number=".$F[1]." ".$F[0]."\n" if /: error:/' \ build.log | xargs -L 1 perl -spi -e 's/ASSERT/EXPECT/g if $. == $number' % make format ``` After that I reverted back change to `ASSERT*` in `util/testharness.h`. But preserved introduced `EXPECT*`, which is the same as `ASSERT*`. This will be deleted once switched to gtest. This diff is independent and contains manual changes only in `util/testharness.h`. Test Plan: Make sure all tests are passing. ```lang=bash % USE_CLANG=1 make check ``` Reviewers: igor, lgalanis, sdong, yufei.zhu, rven, meyering Reviewed By: meyering Subscribers: dhruba, leveldb Differential Revision: https://reviews.facebook.net/D33333
2015-03-17 03:52:32 +00:00
EXPECT_OK(DestroyDB(dbname_, Options()));
}
void SetUp() override {
mmap_mode_ = GetParam();
dbname_ = test::PerThreadDBPath("plain_table_db_test");
EXPECT_OK(DestroyDB(dbname_, Options()));
db_ = nullptr;
Reopen();
}
// Return the current option configuration.
Options CurrentOptions() {
Options options;
Make option `level_compaction_dynamic_level_bytes` true by default (#11525) Summary: after https://github.com/facebook/rocksdb/issues/11321 and https://github.com/facebook/rocksdb/issues/11340 (both included in RocksDB v8.2), migration from `level_compaction_dynamic_level_bytes=false` to `level_compaction_dynamic_level_bytes=true` is automatic by RocksDB and requires no manual compaction from user. Making the option true by default as it has several advantages: 1. better space amplification guarantee (a more stable LSM shape). 2. compaction is more adaptive to write traffic. 3. automatic draining of unneeded levels. Wiki is updated with more detail: https://github.com/facebook/rocksdb/wiki/Leveled-Compaction#option-level_compaction_dynamic_level_bytes-and-levels-target-size. The PR mostly contains fixes for unit tests as they assumed `level_compaction_dynamic_level_bytes=false`. Most notable change is commit https://github.com/facebook/rocksdb/pull/11525/commits/f742be330ca1a7abc33107b00df99818f71c387b and https://github.com/facebook/rocksdb/pull/11525/commits/b1928e42b34c0d4f1cc5d5239149870c6dc7a737 which override the default option in DBTestBase to still set `level_compaction_dynamic_level_bytes=false` by default. This helps to reduce the change needed for unit tests. I think this default option override in unit tests is okay since the behavior of `level_compaction_dynamic_level_bytes=true` is tested by explicitly setting this option. Also, `level_compaction_dynamic_level_bytes=false` may be more desired in unit tests as it makes it easier to create a desired LSM shape. Comment for option `level_compaction_dynamic_level_bytes` is updated to reflect this change and change made in https://github.com/facebook/rocksdb/issues/10057. Pull Request resolved: https://github.com/facebook/rocksdb/pull/11525 Test Plan: `make -j32 J=32 check` several times to try to catch flaky tests due to this option change. Reviewed By: ajkr Differential Revision: D46654256 Pulled By: cbi42 fbshipit-source-id: 6b5827dae124f6f1fdc8cca2ac6f6fcd878830e1
2023-06-16 04:12:39 +00:00
options.level_compaction_dynamic_level_bytes = false;
PlainTableOptions plain_table_options;
plain_table_options.user_key_len = 0;
plain_table_options.bloom_bits_per_key = 2;
plain_table_options.hash_table_ratio = 0.8;
plain_table_options.index_sparseness = 3;
plain_table_options.huge_page_tlb_size = 0;
plain_table_options.encoding_type = kPrefix;
plain_table_options.full_scan_mode = false;
plain_table_options.store_index_in_file = false;
options.table_factory.reset(NewPlainTableFactory(plain_table_options));
options.memtable_factory.reset(NewHashLinkListRepFactory(4, 0, 3, true));
options.prefix_extractor.reset(NewFixedPrefixTransform(8));
options.allow_mmap_reads = mmap_mode_;
options.allow_concurrent_memtable_write = false;
Unordered Writes (#5218) Summary: Performing unordered writes in rocksdb when unordered_write option is set to true. When enabled the writes to memtable are done without joining any write thread. This offers much higher write throughput since the upcoming writes would not have to wait for the slowest memtable write to finish. The tradeoff is that the writes visible to a snapshot might change over time. If the application cannot tolerate that, it should implement its own mechanisms to work around that. Using TransactionDB with WRITE_PREPARED write policy is one way to achieve that. Doing so increases the max throughput by 2.2x without however compromising the snapshot guarantees. The patch is prepared based on an original by siying Existing unit tests are extended to include unordered_write option. Benchmark Results: ``` TEST_TMPDIR=/dev/shm/ ./db_bench_unordered --benchmarks=fillrandom --threads=32 --num=10000000 -max_write_buffer_number=16 --max_background_jobs=64 --batch_size=8 --writes=3000000 -level0_file_num_compaction_trigger=99999 --level0_slowdown_writes_trigger=99999 --level0_stop_writes_trigger=99999 -enable_pipelined_write=false -disable_auto_compactions --unordered_write=1 ``` With WAL - Vanilla RocksDB: 78.6 MB/s - WRITER_PREPARED with unordered_write: 177.8 MB/s (2.2x) - unordered_write: 368.9 MB/s (4.7x with relaxed snapshot guarantees) Without WAL - Vanilla RocksDB: 111.3 MB/s - WRITER_PREPARED with unordered_write: 259.3 MB/s MB/s (2.3x) - unordered_write: 645.6 MB/s (5.8x with relaxed snapshot guarantees) - WRITER_PREPARED with unordered_write disable concurrency control: 185.3 MB/s MB/s (2.35x) Limitations: - The feature is not yet extended to `max_successive_merges` > 0. The feature is also incompatible with `enable_pipelined_write` = true as well as with `allow_concurrent_memtable_write` = false. Pull Request resolved: https://github.com/facebook/rocksdb/pull/5218 Differential Revision: D15219029 Pulled By: maysamyabandeh fbshipit-source-id: 38f2abc4af8780148c6128acdba2b3227bc81759
2019-05-14 00:43:47 +00:00
options.unordered_write = false;
return options;
}
DBImpl* dbfull() { return static_cast_with_check<DBImpl>(db_); }
void Reopen(Options* options = nullptr) { ASSERT_OK(TryReopen(options)); }
void Close() {
delete db_;
db_ = nullptr;
}
bool mmap_mode() const { return mmap_mode_; }
void DestroyAndReopen(Options* options = nullptr) {
// Destroy using last options
Destroy(&last_options_);
ASSERT_OK(TryReopen(options));
}
void Destroy(Options* options) {
delete db_;
db_ = nullptr;
ASSERT_OK(DestroyDB(dbname_, *options));
}
Status PureReopen(Options* options, DB** db) {
return DB::Open(*options, dbname_, db);
}
Status ReopenForReadOnly(Options* options) {
delete db_;
db_ = nullptr;
return DB::OpenForReadOnly(*options, dbname_, &db_);
}
Status TryReopen(Options* options = nullptr) {
delete db_;
db_ = nullptr;
Options opts;
if (options != nullptr) {
opts = *options;
} else {
opts = CurrentOptions();
opts.create_if_missing = true;
}
last_options_ = opts;
return DB::Open(opts, dbname_, &db_);
}
Status Put(const Slice& k, const Slice& v) {
return db_->Put(WriteOptions(), k, v);
}
Status Delete(const std::string& k) { return db_->Delete(WriteOptions(), k); }
std::string Get(const std::string& k, const Snapshot* snapshot = nullptr) {
ReadOptions options;
options.snapshot = snapshot;
std::string result;
Status s = db_->Get(options, k, &result);
if (s.IsNotFound()) {
result = "NOT_FOUND";
} else if (!s.ok()) {
result = s.ToString();
}
return result;
}
int NumTableFilesAtLevel(int level) {
std::string property;
EXPECT_TRUE(db_->GetProperty(
"rocksdb.num-files-at-level" + std::to_string(level), &property));
return atoi(property.c_str());
}
// Return spread of files per level
std::string FilesPerLevel() {
std::string result;
size_t last_non_zero_offset = 0;
for (int level = 0; level < db_->NumberLevels(); level++) {
int f = NumTableFilesAtLevel(level);
char buf[100];
snprintf(buf, sizeof(buf), "%s%d", (level ? "," : ""), f);
result += buf;
if (f > 0) {
last_non_zero_offset = result.size();
}
}
result.resize(last_non_zero_offset);
return result;
}
std::string IterStatus(Iterator* iter) {
std::string result;
if (iter->Valid()) {
result = iter->key().ToString() + "->" + iter->value().ToString();
} else {
result = "(invalid)";
}
return result;
}
};
TEST_P(PlainTableDBTest, Empty) {
2014-01-03 18:53:21 +00:00
ASSERT_TRUE(dbfull() != nullptr);
ASSERT_EQ("NOT_FOUND", Get("0000000000000foo"));
}
class TestPlainTableReader : public PlainTableReader {
public:
TestPlainTableReader(
const EnvOptions& env_options, const InternalKeyComparator& icomparator,
EncodingType encoding_type, uint64_t file_size, int bloom_bits_per_key,
double hash_table_ratio, size_t index_sparseness,
Improve / clean up meta block code & integrity (#9163) Summary: * Checksums are now checked on meta blocks unless specifically suppressed or not applicable (e.g. plain table). (Was other way around.) This means a number of cases that were not checking checksums now are, including direct read TableProperties in Version::GetTableProperties (fixed in meta_blocks ReadTableProperties), reading any block from PersistentCache (fixed in BlockFetcher), read TableProperties in SstFileDumper (ldb/sst_dump/BackupEngine) before table reader open, maybe more. * For that to work, I moved the global_seqno+TableProperties checksum logic to the shared table/ code, because that is used by many utilies such as SstFileDumper. * Also for that to work, we have to know when we're dealing with a block that has a checksum (trailer), so added that capability to Footer based on magic number, and from there BlockFetcher. * Knowledge of trailer presence has also fixed a problem where other table formats were reading blocks including bytes for a non-existant trailer--and awkwardly kind-of not using them, e.g. no shared code checking checksums. (BlockFetcher compression type was populated incorrectly.) Now we only read what is needed. * Minimized code duplication and differing/incompatible/awkward abstractions in meta_blocks.{cc,h} (e.g. SeekTo in metaindex block without parsing block handle) * Moved some meta block handling code from table_properties*.* * Moved some code specific to block-based table from shared table/ code to BlockBasedTable class. The checksum stuff means we can't completely separate it, but things that don't need to be in shared table/ code should not be. * Use unique_ptr rather than raw ptr in more places. (Note: you can std::move from unique_ptr to shared_ptr.) Without enhancements to GetPropertiesOfAllTablesTest (see below), net reduction of roughly 100 lines of code. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9163 Test Plan: existing tests and * Enhanced DBTablePropertiesTest.GetPropertiesOfAllTablesTest to verify that checksums are now checked on direct read of table properties by TableCache (new test would fail before this change) * Also enhanced DBTablePropertiesTest.GetPropertiesOfAllTablesTest to test putting table properties under old meta name * Also generally enhanced that same test to actually test what it was supposed to be testing already, by kicking things out of table cache when we don't want them there. Reviewed By: ajkr, mrambacher Differential Revision: D32514757 Pulled By: pdillinger fbshipit-source-id: 507964b9311d186ae8d1131182290cbd97a99fa9
2021-11-18 19:42:12 +00:00
std::unique_ptr<TableProperties>&& props,
std::unique_ptr<RandomAccessFileReader>&& file,
const ImmutableOptions& ioptions, const SliceTransform* prefix_extractor,
bool* expect_bloom_not_match, bool store_index_in_file,
uint32_t column_family_id, const std::string& column_family_name)
: PlainTableReader(ioptions, std::move(file), env_options, icomparator,
Improve / clean up meta block code & integrity (#9163) Summary: * Checksums are now checked on meta blocks unless specifically suppressed or not applicable (e.g. plain table). (Was other way around.) This means a number of cases that were not checking checksums now are, including direct read TableProperties in Version::GetTableProperties (fixed in meta_blocks ReadTableProperties), reading any block from PersistentCache (fixed in BlockFetcher), read TableProperties in SstFileDumper (ldb/sst_dump/BackupEngine) before table reader open, maybe more. * For that to work, I moved the global_seqno+TableProperties checksum logic to the shared table/ code, because that is used by many utilies such as SstFileDumper. * Also for that to work, we have to know when we're dealing with a block that has a checksum (trailer), so added that capability to Footer based on magic number, and from there BlockFetcher. * Knowledge of trailer presence has also fixed a problem where other table formats were reading blocks including bytes for a non-existant trailer--and awkwardly kind-of not using them, e.g. no shared code checking checksums. (BlockFetcher compression type was populated incorrectly.) Now we only read what is needed. * Minimized code duplication and differing/incompatible/awkward abstractions in meta_blocks.{cc,h} (e.g. SeekTo in metaindex block without parsing block handle) * Moved some meta block handling code from table_properties*.* * Moved some code specific to block-based table from shared table/ code to BlockBasedTable class. The checksum stuff means we can't completely separate it, but things that don't need to be in shared table/ code should not be. * Use unique_ptr rather than raw ptr in more places. (Note: you can std::move from unique_ptr to shared_ptr.) Without enhancements to GetPropertiesOfAllTablesTest (see below), net reduction of roughly 100 lines of code. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9163 Test Plan: existing tests and * Enhanced DBTablePropertiesTest.GetPropertiesOfAllTablesTest to verify that checksums are now checked on direct read of table properties by TableCache (new test would fail before this change) * Also enhanced DBTablePropertiesTest.GetPropertiesOfAllTablesTest to test putting table properties under old meta name * Also generally enhanced that same test to actually test what it was supposed to be testing already, by kicking things out of table cache when we don't want them there. Reviewed By: ajkr, mrambacher Differential Revision: D32514757 Pulled By: pdillinger fbshipit-source-id: 507964b9311d186ae8d1131182290cbd97a99fa9
2021-11-18 19:42:12 +00:00
encoding_type, file_size, props.get(),
prefix_extractor),
expect_bloom_not_match_(expect_bloom_not_match) {
Status s = MmapDataIfNeeded();
rocksdb: Replace ASSERT* with EXPECT* in functions that does not return void value Summary: gtest does not use exceptions to fail a unit test by design, and `ASSERT*`s are implemented using `return`. As a consequence we cannot use `ASSERT*` in a function that does not return `void` value ([[ https://code.google.com/p/googletest/wiki/AdvancedGuide#Assertion_Placement | 1]]), and have to fix our existing code. This diff does this in a generic way, with no manual changes. In order to detect all existing `ASSERT*` that are used in functions that doesn't return void value, I change the code to generate compile errors for such cases. In `util/testharness.h` I defined `EXPECT*` assertions, the same way as `ASSERT*`, and redefined `ASSERT*` to return `void`. Then executed: ```lang=bash % USE_CLANG=1 make all -j55 -k 2> build.log % perl -naF: -e 'print "-- -number=".$F[1]." ".$F[0]."\n" if /: error:/' \ build.log | xargs -L 1 perl -spi -e 's/ASSERT/EXPECT/g if $. == $number' % make format ``` After that I reverted back change to `ASSERT*` in `util/testharness.h`. But preserved introduced `EXPECT*`, which is the same as `ASSERT*`. This will be deleted once switched to gtest. This diff is independent and contains manual changes only in `util/testharness.h`. Test Plan: Make sure all tests are passing. ```lang=bash % USE_CLANG=1 make check ``` Reviewers: igor, lgalanis, sdong, yufei.zhu, rven, meyering Reviewed By: meyering Subscribers: dhruba, leveldb Differential Revision: https://reviews.facebook.net/D33333
2015-03-17 03:52:32 +00:00
EXPECT_TRUE(s.ok());
Improve / clean up meta block code & integrity (#9163) Summary: * Checksums are now checked on meta blocks unless specifically suppressed or not applicable (e.g. plain table). (Was other way around.) This means a number of cases that were not checking checksums now are, including direct read TableProperties in Version::GetTableProperties (fixed in meta_blocks ReadTableProperties), reading any block from PersistentCache (fixed in BlockFetcher), read TableProperties in SstFileDumper (ldb/sst_dump/BackupEngine) before table reader open, maybe more. * For that to work, I moved the global_seqno+TableProperties checksum logic to the shared table/ code, because that is used by many utilies such as SstFileDumper. * Also for that to work, we have to know when we're dealing with a block that has a checksum (trailer), so added that capability to Footer based on magic number, and from there BlockFetcher. * Knowledge of trailer presence has also fixed a problem where other table formats were reading blocks including bytes for a non-existant trailer--and awkwardly kind-of not using them, e.g. no shared code checking checksums. (BlockFetcher compression type was populated incorrectly.) Now we only read what is needed. * Minimized code duplication and differing/incompatible/awkward abstractions in meta_blocks.{cc,h} (e.g. SeekTo in metaindex block without parsing block handle) * Moved some meta block handling code from table_properties*.* * Moved some code specific to block-based table from shared table/ code to BlockBasedTable class. The checksum stuff means we can't completely separate it, but things that don't need to be in shared table/ code should not be. * Use unique_ptr rather than raw ptr in more places. (Note: you can std::move from unique_ptr to shared_ptr.) Without enhancements to GetPropertiesOfAllTablesTest (see below), net reduction of roughly 100 lines of code. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9163 Test Plan: existing tests and * Enhanced DBTablePropertiesTest.GetPropertiesOfAllTablesTest to verify that checksums are now checked on direct read of table properties by TableCache (new test would fail before this change) * Also enhanced DBTablePropertiesTest.GetPropertiesOfAllTablesTest to test putting table properties under old meta name * Also generally enhanced that same test to actually test what it was supposed to be testing already, by kicking things out of table cache when we don't want them there. Reviewed By: ajkr, mrambacher Differential Revision: D32514757 Pulled By: pdillinger fbshipit-source-id: 507964b9311d186ae8d1131182290cbd97a99fa9
2021-11-18 19:42:12 +00:00
s = PopulateIndex(props.get(), bloom_bits_per_key, hash_table_ratio,
index_sparseness, 2 * 1024 * 1024);
rocksdb: Replace ASSERT* with EXPECT* in functions that does not return void value Summary: gtest does not use exceptions to fail a unit test by design, and `ASSERT*`s are implemented using `return`. As a consequence we cannot use `ASSERT*` in a function that does not return `void` value ([[ https://code.google.com/p/googletest/wiki/AdvancedGuide#Assertion_Placement | 1]]), and have to fix our existing code. This diff does this in a generic way, with no manual changes. In order to detect all existing `ASSERT*` that are used in functions that doesn't return void value, I change the code to generate compile errors for such cases. In `util/testharness.h` I defined `EXPECT*` assertions, the same way as `ASSERT*`, and redefined `ASSERT*` to return `void`. Then executed: ```lang=bash % USE_CLANG=1 make all -j55 -k 2> build.log % perl -naF: -e 'print "-- -number=".$F[1]." ".$F[0]."\n" if /: error:/' \ build.log | xargs -L 1 perl -spi -e 's/ASSERT/EXPECT/g if $. == $number' % make format ``` After that I reverted back change to `ASSERT*` in `util/testharness.h`. But preserved introduced `EXPECT*`, which is the same as `ASSERT*`. This will be deleted once switched to gtest. This diff is independent and contains manual changes only in `util/testharness.h`. Test Plan: Make sure all tests are passing. ```lang=bash % USE_CLANG=1 make check ``` Reviewers: igor, lgalanis, sdong, yufei.zhu, rven, meyering Reviewed By: meyering Subscribers: dhruba, leveldb Differential Revision: https://reviews.facebook.net/D33333
2015-03-17 03:52:32 +00:00
EXPECT_TRUE(s.ok());
EXPECT_EQ(column_family_id, static_cast<uint32_t>(props->column_family_id));
EXPECT_EQ(column_family_name, props->column_family_name);
if (store_index_in_file) {
auto bloom_version_ptr = props->user_collected_properties.find(
PlainTablePropertyNames::kBloomVersion);
EXPECT_TRUE(bloom_version_ptr != props->user_collected_properties.end());
EXPECT_EQ(bloom_version_ptr->second, std::string("1"));
if (ioptions.bloom_locality > 0) {
auto num_blocks_ptr = props->user_collected_properties.find(
PlainTablePropertyNames::kNumBloomBlocks);
EXPECT_TRUE(num_blocks_ptr != props->user_collected_properties.end());
}
}
Improve / clean up meta block code & integrity (#9163) Summary: * Checksums are now checked on meta blocks unless specifically suppressed or not applicable (e.g. plain table). (Was other way around.) This means a number of cases that were not checking checksums now are, including direct read TableProperties in Version::GetTableProperties (fixed in meta_blocks ReadTableProperties), reading any block from PersistentCache (fixed in BlockFetcher), read TableProperties in SstFileDumper (ldb/sst_dump/BackupEngine) before table reader open, maybe more. * For that to work, I moved the global_seqno+TableProperties checksum logic to the shared table/ code, because that is used by many utilies such as SstFileDumper. * Also for that to work, we have to know when we're dealing with a block that has a checksum (trailer), so added that capability to Footer based on magic number, and from there BlockFetcher. * Knowledge of trailer presence has also fixed a problem where other table formats were reading blocks including bytes for a non-existant trailer--and awkwardly kind-of not using them, e.g. no shared code checking checksums. (BlockFetcher compression type was populated incorrectly.) Now we only read what is needed. * Minimized code duplication and differing/incompatible/awkward abstractions in meta_blocks.{cc,h} (e.g. SeekTo in metaindex block without parsing block handle) * Moved some meta block handling code from table_properties*.* * Moved some code specific to block-based table from shared table/ code to BlockBasedTable class. The checksum stuff means we can't completely separate it, but things that don't need to be in shared table/ code should not be. * Use unique_ptr rather than raw ptr in more places. (Note: you can std::move from unique_ptr to shared_ptr.) Without enhancements to GetPropertiesOfAllTablesTest (see below), net reduction of roughly 100 lines of code. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9163 Test Plan: existing tests and * Enhanced DBTablePropertiesTest.GetPropertiesOfAllTablesTest to verify that checksums are now checked on direct read of table properties by TableCache (new test would fail before this change) * Also enhanced DBTablePropertiesTest.GetPropertiesOfAllTablesTest to test putting table properties under old meta name * Also generally enhanced that same test to actually test what it was supposed to be testing already, by kicking things out of table cache when we don't want them there. Reviewed By: ajkr, mrambacher Differential Revision: D32514757 Pulled By: pdillinger fbshipit-source-id: 507964b9311d186ae8d1131182290cbd97a99fa9
2021-11-18 19:42:12 +00:00
table_properties_ = std::move(props);
}
~TestPlainTableReader() override = default;
private:
bool MatchBloom(uint32_t hash) const override {
bool ret = PlainTableReader::MatchBloom(hash);
if (*expect_bloom_not_match_) {
rocksdb: Replace ASSERT* with EXPECT* in functions that does not return void value Summary: gtest does not use exceptions to fail a unit test by design, and `ASSERT*`s are implemented using `return`. As a consequence we cannot use `ASSERT*` in a function that does not return `void` value ([[ https://code.google.com/p/googletest/wiki/AdvancedGuide#Assertion_Placement | 1]]), and have to fix our existing code. This diff does this in a generic way, with no manual changes. In order to detect all existing `ASSERT*` that are used in functions that doesn't return void value, I change the code to generate compile errors for such cases. In `util/testharness.h` I defined `EXPECT*` assertions, the same way as `ASSERT*`, and redefined `ASSERT*` to return `void`. Then executed: ```lang=bash % USE_CLANG=1 make all -j55 -k 2> build.log % perl -naF: -e 'print "-- -number=".$F[1]." ".$F[0]."\n" if /: error:/' \ build.log | xargs -L 1 perl -spi -e 's/ASSERT/EXPECT/g if $. == $number' % make format ``` After that I reverted back change to `ASSERT*` in `util/testharness.h`. But preserved introduced `EXPECT*`, which is the same as `ASSERT*`. This will be deleted once switched to gtest. This diff is independent and contains manual changes only in `util/testharness.h`. Test Plan: Make sure all tests are passing. ```lang=bash % USE_CLANG=1 make check ``` Reviewers: igor, lgalanis, sdong, yufei.zhu, rven, meyering Reviewed By: meyering Subscribers: dhruba, leveldb Differential Revision: https://reviews.facebook.net/D33333
2015-03-17 03:52:32 +00:00
EXPECT_TRUE(!ret);
} else {
rocksdb: Replace ASSERT* with EXPECT* in functions that does not return void value Summary: gtest does not use exceptions to fail a unit test by design, and `ASSERT*`s are implemented using `return`. As a consequence we cannot use `ASSERT*` in a function that does not return `void` value ([[ https://code.google.com/p/googletest/wiki/AdvancedGuide#Assertion_Placement | 1]]), and have to fix our existing code. This diff does this in a generic way, with no manual changes. In order to detect all existing `ASSERT*` that are used in functions that doesn't return void value, I change the code to generate compile errors for such cases. In `util/testharness.h` I defined `EXPECT*` assertions, the same way as `ASSERT*`, and redefined `ASSERT*` to return `void`. Then executed: ```lang=bash % USE_CLANG=1 make all -j55 -k 2> build.log % perl -naF: -e 'print "-- -number=".$F[1]." ".$F[0]."\n" if /: error:/' \ build.log | xargs -L 1 perl -spi -e 's/ASSERT/EXPECT/g if $. == $number' % make format ``` After that I reverted back change to `ASSERT*` in `util/testharness.h`. But preserved introduced `EXPECT*`, which is the same as `ASSERT*`. This will be deleted once switched to gtest. This diff is independent and contains manual changes only in `util/testharness.h`. Test Plan: Make sure all tests are passing. ```lang=bash % USE_CLANG=1 make check ``` Reviewers: igor, lgalanis, sdong, yufei.zhu, rven, meyering Reviewed By: meyering Subscribers: dhruba, leveldb Differential Revision: https://reviews.facebook.net/D33333
2015-03-17 03:52:32 +00:00
EXPECT_TRUE(ret);
}
return ret;
}
bool* expect_bloom_not_match_;
};
class TestPlainTableFactory : public PlainTableFactory {
public:
explicit TestPlainTableFactory(bool* expect_bloom_not_match,
const PlainTableOptions& options,
uint32_t column_family_id,
std::string column_family_name)
: PlainTableFactory(options),
bloom_bits_per_key_(options.bloom_bits_per_key),
hash_table_ratio_(options.hash_table_ratio),
index_sparseness_(options.index_sparseness),
store_index_in_file_(options.store_index_in_file),
expect_bloom_not_match_(expect_bloom_not_match),
column_family_id_(column_family_id),
column_family_name_(std::move(column_family_name)) {}
using PlainTableFactory::NewTableReader;
Status NewTableReader(
const ReadOptions& /*ro*/, const TableReaderOptions& table_reader_options,
std::unique_ptr<RandomAccessFileReader>&& file, uint64_t file_size,
std::unique_ptr<TableReader>* table,
bool /*prefetch_index_and_filter_in_cache*/) const override {
Improve / clean up meta block code & integrity (#9163) Summary: * Checksums are now checked on meta blocks unless specifically suppressed or not applicable (e.g. plain table). (Was other way around.) This means a number of cases that were not checking checksums now are, including direct read TableProperties in Version::GetTableProperties (fixed in meta_blocks ReadTableProperties), reading any block from PersistentCache (fixed in BlockFetcher), read TableProperties in SstFileDumper (ldb/sst_dump/BackupEngine) before table reader open, maybe more. * For that to work, I moved the global_seqno+TableProperties checksum logic to the shared table/ code, because that is used by many utilies such as SstFileDumper. * Also for that to work, we have to know when we're dealing with a block that has a checksum (trailer), so added that capability to Footer based on magic number, and from there BlockFetcher. * Knowledge of trailer presence has also fixed a problem where other table formats were reading blocks including bytes for a non-existant trailer--and awkwardly kind-of not using them, e.g. no shared code checking checksums. (BlockFetcher compression type was populated incorrectly.) Now we only read what is needed. * Minimized code duplication and differing/incompatible/awkward abstractions in meta_blocks.{cc,h} (e.g. SeekTo in metaindex block without parsing block handle) * Moved some meta block handling code from table_properties*.* * Moved some code specific to block-based table from shared table/ code to BlockBasedTable class. The checksum stuff means we can't completely separate it, but things that don't need to be in shared table/ code should not be. * Use unique_ptr rather than raw ptr in more places. (Note: you can std::move from unique_ptr to shared_ptr.) Without enhancements to GetPropertiesOfAllTablesTest (see below), net reduction of roughly 100 lines of code. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9163 Test Plan: existing tests and * Enhanced DBTablePropertiesTest.GetPropertiesOfAllTablesTest to verify that checksums are now checked on direct read of table properties by TableCache (new test would fail before this change) * Also enhanced DBTablePropertiesTest.GetPropertiesOfAllTablesTest to test putting table properties under old meta name * Also generally enhanced that same test to actually test what it was supposed to be testing already, by kicking things out of table cache when we don't want them there. Reviewed By: ajkr, mrambacher Differential Revision: D32514757 Pulled By: pdillinger fbshipit-source-id: 507964b9311d186ae8d1131182290cbd97a99fa9
2021-11-18 19:42:12 +00:00
std::unique_ptr<TableProperties> props;
Group rocksdb.sst.read.micros stat by IOActivity flush and compaction (#11288) Summary: **Context:** The existing stat rocksdb.sst.read.micros does not reflect each of compaction and flush cases but aggregate them, which is not so helpful for us to understand IO read behavior of each of them. **Summary** - Update `StopWatch` and `RandomAccessFileReader` to record `rocksdb.sst.read.micros` and `rocksdb.file.{flush/compaction}.read.micros` - Fixed the default histogram in `RandomAccessFileReader` - New field `ReadOptions/IOOptions::io_activity`; Pass `ReadOptions` through paths under db open, flush and compaction to where we can prepare `IOOptions` and pass it to `RandomAccessFileReader` - Use `thread_status_util` for assertion in `DbStressFSWrapper` for continuous testing on we are passing correct `io_activity` under db open, flush and compaction Pull Request resolved: https://github.com/facebook/rocksdb/pull/11288 Test Plan: - **Stress test** - **Db bench 1: rocksdb.sst.read.micros COUNT ≈ sum of rocksdb.file.read.flush.micros's and rocksdb.file.read.compaction.micros's.** (without blob) - May not be exactly the same due to `HistogramStat::Add` only guarantees atomic not accuracy across threads. ``` ./db_bench -db=/dev/shm/testdb/ -statistics=true -benchmarks="fillseq" -key_size=32 -value_size=512 -num=50000 -write_buffer_size=655 -target_file_size_base=655 -disable_auto_compactions=false -compression_type=none -bloom_bits=3 (-use_plain_table=1 -prefix_size=10) ``` ``` // BlockBasedTable rocksdb.sst.read.micros P50 : 2.009374 P95 : 4.968548 P99 : 8.110362 P100 : 43.000000 COUNT : 40456 SUM : 114805 rocksdb.file.read.flush.micros P50 : 1.871841 P95 : 3.872407 P99 : 5.540541 P100 : 43.000000 COUNT : 2250 SUM : 6116 rocksdb.file.read.compaction.micros P50 : 2.023109 P95 : 5.029149 P99 : 8.196910 P100 : 26.000000 COUNT : 38206 SUM : 108689 // PlainTable Does not apply ``` - **Db bench 2: performance** **Read** SETUP: db with 900 files ``` ./db_bench -db=/dev/shm/testdb/ -benchmarks="fillseq" -key_size=32 -value_size=512 -num=50000 -write_buffer_size=655 -disable_auto_compactions=true -target_file_size_base=655 -compression_type=none ```run till convergence ``` ./db_bench -seed=1678564177044286 -use_existing_db=true -db=/dev/shm/testdb -benchmarks=readrandom[-X60] -statistics=true -num=1000000 -disable_auto_compactions=true -compression_type=none -bloom_bits=3 ``` Pre-change `readrandom [AVG 60 runs] : 21568 (± 248) ops/sec` Post-change (no regression, -0.3%) `readrandom [AVG 60 runs] : 21486 (± 236) ops/sec` **Compaction/Flush**run till convergence ``` ./db_bench -db=/dev/shm/testdb2/ -seed=1678564177044286 -benchmarks="fillseq[-X60]" -key_size=32 -value_size=512 -num=50000 -write_buffer_size=655 -disable_auto_compactions=false -target_file_size_base=655 -compression_type=none rocksdb.sst.read.micros COUNT : 33820 rocksdb.sst.read.flush.micros COUNT : 1800 rocksdb.sst.read.compaction.micros COUNT : 32020 ``` Pre-change `fillseq [AVG 46 runs] : 1391 (± 214) ops/sec; 0.7 (± 0.1) MB/sec` Post-change (no regression, ~-0.4%) `fillseq [AVG 46 runs] : 1385 (± 216) ops/sec; 0.7 (± 0.1) MB/sec` Reviewed By: ajkr Differential Revision: D44007011 Pulled By: hx235 fbshipit-source-id: a54c89e4846dfc9a135389edf3f3eedfea257132
2023-04-21 16:07:18 +00:00
const ReadOptions read_options;
Improve / clean up meta block code & integrity (#9163) Summary: * Checksums are now checked on meta blocks unless specifically suppressed or not applicable (e.g. plain table). (Was other way around.) This means a number of cases that were not checking checksums now are, including direct read TableProperties in Version::GetTableProperties (fixed in meta_blocks ReadTableProperties), reading any block from PersistentCache (fixed in BlockFetcher), read TableProperties in SstFileDumper (ldb/sst_dump/BackupEngine) before table reader open, maybe more. * For that to work, I moved the global_seqno+TableProperties checksum logic to the shared table/ code, because that is used by many utilies such as SstFileDumper. * Also for that to work, we have to know when we're dealing with a block that has a checksum (trailer), so added that capability to Footer based on magic number, and from there BlockFetcher. * Knowledge of trailer presence has also fixed a problem where other table formats were reading blocks including bytes for a non-existant trailer--and awkwardly kind-of not using them, e.g. no shared code checking checksums. (BlockFetcher compression type was populated incorrectly.) Now we only read what is needed. * Minimized code duplication and differing/incompatible/awkward abstractions in meta_blocks.{cc,h} (e.g. SeekTo in metaindex block without parsing block handle) * Moved some meta block handling code from table_properties*.* * Moved some code specific to block-based table from shared table/ code to BlockBasedTable class. The checksum stuff means we can't completely separate it, but things that don't need to be in shared table/ code should not be. * Use unique_ptr rather than raw ptr in more places. (Note: you can std::move from unique_ptr to shared_ptr.) Without enhancements to GetPropertiesOfAllTablesTest (see below), net reduction of roughly 100 lines of code. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9163 Test Plan: existing tests and * Enhanced DBTablePropertiesTest.GetPropertiesOfAllTablesTest to verify that checksums are now checked on direct read of table properties by TableCache (new test would fail before this change) * Also enhanced DBTablePropertiesTest.GetPropertiesOfAllTablesTest to test putting table properties under old meta name * Also generally enhanced that same test to actually test what it was supposed to be testing already, by kicking things out of table cache when we don't want them there. Reviewed By: ajkr, mrambacher Differential Revision: D32514757 Pulled By: pdillinger fbshipit-source-id: 507964b9311d186ae8d1131182290cbd97a99fa9
2021-11-18 19:42:12 +00:00
auto s = ReadTableProperties(file.get(), file_size, kPlainTableMagicNumber,
Group rocksdb.sst.read.micros stat by IOActivity flush and compaction (#11288) Summary: **Context:** The existing stat rocksdb.sst.read.micros does not reflect each of compaction and flush cases but aggregate them, which is not so helpful for us to understand IO read behavior of each of them. **Summary** - Update `StopWatch` and `RandomAccessFileReader` to record `rocksdb.sst.read.micros` and `rocksdb.file.{flush/compaction}.read.micros` - Fixed the default histogram in `RandomAccessFileReader` - New field `ReadOptions/IOOptions::io_activity`; Pass `ReadOptions` through paths under db open, flush and compaction to where we can prepare `IOOptions` and pass it to `RandomAccessFileReader` - Use `thread_status_util` for assertion in `DbStressFSWrapper` for continuous testing on we are passing correct `io_activity` under db open, flush and compaction Pull Request resolved: https://github.com/facebook/rocksdb/pull/11288 Test Plan: - **Stress test** - **Db bench 1: rocksdb.sst.read.micros COUNT ≈ sum of rocksdb.file.read.flush.micros's and rocksdb.file.read.compaction.micros's.** (without blob) - May not be exactly the same due to `HistogramStat::Add` only guarantees atomic not accuracy across threads. ``` ./db_bench -db=/dev/shm/testdb/ -statistics=true -benchmarks="fillseq" -key_size=32 -value_size=512 -num=50000 -write_buffer_size=655 -target_file_size_base=655 -disable_auto_compactions=false -compression_type=none -bloom_bits=3 (-use_plain_table=1 -prefix_size=10) ``` ``` // BlockBasedTable rocksdb.sst.read.micros P50 : 2.009374 P95 : 4.968548 P99 : 8.110362 P100 : 43.000000 COUNT : 40456 SUM : 114805 rocksdb.file.read.flush.micros P50 : 1.871841 P95 : 3.872407 P99 : 5.540541 P100 : 43.000000 COUNT : 2250 SUM : 6116 rocksdb.file.read.compaction.micros P50 : 2.023109 P95 : 5.029149 P99 : 8.196910 P100 : 26.000000 COUNT : 38206 SUM : 108689 // PlainTable Does not apply ``` - **Db bench 2: performance** **Read** SETUP: db with 900 files ``` ./db_bench -db=/dev/shm/testdb/ -benchmarks="fillseq" -key_size=32 -value_size=512 -num=50000 -write_buffer_size=655 -disable_auto_compactions=true -target_file_size_base=655 -compression_type=none ```run till convergence ``` ./db_bench -seed=1678564177044286 -use_existing_db=true -db=/dev/shm/testdb -benchmarks=readrandom[-X60] -statistics=true -num=1000000 -disable_auto_compactions=true -compression_type=none -bloom_bits=3 ``` Pre-change `readrandom [AVG 60 runs] : 21568 (± 248) ops/sec` Post-change (no regression, -0.3%) `readrandom [AVG 60 runs] : 21486 (± 236) ops/sec` **Compaction/Flush**run till convergence ``` ./db_bench -db=/dev/shm/testdb2/ -seed=1678564177044286 -benchmarks="fillseq[-X60]" -key_size=32 -value_size=512 -num=50000 -write_buffer_size=655 -disable_auto_compactions=false -target_file_size_base=655 -compression_type=none rocksdb.sst.read.micros COUNT : 33820 rocksdb.sst.read.flush.micros COUNT : 1800 rocksdb.sst.read.compaction.micros COUNT : 32020 ``` Pre-change `fillseq [AVG 46 runs] : 1391 (± 214) ops/sec; 0.7 (± 0.1) MB/sec` Post-change (no regression, ~-0.4%) `fillseq [AVG 46 runs] : 1385 (± 216) ops/sec; 0.7 (± 0.1) MB/sec` Reviewed By: ajkr Differential Revision: D44007011 Pulled By: hx235 fbshipit-source-id: a54c89e4846dfc9a135389edf3f3eedfea257132
2023-04-21 16:07:18 +00:00
table_reader_options.ioptions, read_options,
&props);
rocksdb: Replace ASSERT* with EXPECT* in functions that does not return void value Summary: gtest does not use exceptions to fail a unit test by design, and `ASSERT*`s are implemented using `return`. As a consequence we cannot use `ASSERT*` in a function that does not return `void` value ([[ https://code.google.com/p/googletest/wiki/AdvancedGuide#Assertion_Placement | 1]]), and have to fix our existing code. This diff does this in a generic way, with no manual changes. In order to detect all existing `ASSERT*` that are used in functions that doesn't return void value, I change the code to generate compile errors for such cases. In `util/testharness.h` I defined `EXPECT*` assertions, the same way as `ASSERT*`, and redefined `ASSERT*` to return `void`. Then executed: ```lang=bash % USE_CLANG=1 make all -j55 -k 2> build.log % perl -naF: -e 'print "-- -number=".$F[1]." ".$F[0]."\n" if /: error:/' \ build.log | xargs -L 1 perl -spi -e 's/ASSERT/EXPECT/g if $. == $number' % make format ``` After that I reverted back change to `ASSERT*` in `util/testharness.h`. But preserved introduced `EXPECT*`, which is the same as `ASSERT*`. This will be deleted once switched to gtest. This diff is independent and contains manual changes only in `util/testharness.h`. Test Plan: Make sure all tests are passing. ```lang=bash % USE_CLANG=1 make check ``` Reviewers: igor, lgalanis, sdong, yufei.zhu, rven, meyering Reviewed By: meyering Subscribers: dhruba, leveldb Differential Revision: https://reviews.facebook.net/D33333
2015-03-17 03:52:32 +00:00
EXPECT_TRUE(s.ok());
if (store_index_in_file_) {
BlockHandle bloom_block_handle;
Improve / clean up meta block code & integrity (#9163) Summary: * Checksums are now checked on meta blocks unless specifically suppressed or not applicable (e.g. plain table). (Was other way around.) This means a number of cases that were not checking checksums now are, including direct read TableProperties in Version::GetTableProperties (fixed in meta_blocks ReadTableProperties), reading any block from PersistentCache (fixed in BlockFetcher), read TableProperties in SstFileDumper (ldb/sst_dump/BackupEngine) before table reader open, maybe more. * For that to work, I moved the global_seqno+TableProperties checksum logic to the shared table/ code, because that is used by many utilies such as SstFileDumper. * Also for that to work, we have to know when we're dealing with a block that has a checksum (trailer), so added that capability to Footer based on magic number, and from there BlockFetcher. * Knowledge of trailer presence has also fixed a problem where other table formats were reading blocks including bytes for a non-existant trailer--and awkwardly kind-of not using them, e.g. no shared code checking checksums. (BlockFetcher compression type was populated incorrectly.) Now we only read what is needed. * Minimized code duplication and differing/incompatible/awkward abstractions in meta_blocks.{cc,h} (e.g. SeekTo in metaindex block without parsing block handle) * Moved some meta block handling code from table_properties*.* * Moved some code specific to block-based table from shared table/ code to BlockBasedTable class. The checksum stuff means we can't completely separate it, but things that don't need to be in shared table/ code should not be. * Use unique_ptr rather than raw ptr in more places. (Note: you can std::move from unique_ptr to shared_ptr.) Without enhancements to GetPropertiesOfAllTablesTest (see below), net reduction of roughly 100 lines of code. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9163 Test Plan: existing tests and * Enhanced DBTablePropertiesTest.GetPropertiesOfAllTablesTest to verify that checksums are now checked on direct read of table properties by TableCache (new test would fail before this change) * Also enhanced DBTablePropertiesTest.GetPropertiesOfAllTablesTest to test putting table properties under old meta name * Also generally enhanced that same test to actually test what it was supposed to be testing already, by kicking things out of table cache when we don't want them there. Reviewed By: ajkr, mrambacher Differential Revision: D32514757 Pulled By: pdillinger fbshipit-source-id: 507964b9311d186ae8d1131182290cbd97a99fa9
2021-11-18 19:42:12 +00:00
s = FindMetaBlockInFile(file.get(), file_size, kPlainTableMagicNumber,
Group rocksdb.sst.read.micros stat by IOActivity flush and compaction (#11288) Summary: **Context:** The existing stat rocksdb.sst.read.micros does not reflect each of compaction and flush cases but aggregate them, which is not so helpful for us to understand IO read behavior of each of them. **Summary** - Update `StopWatch` and `RandomAccessFileReader` to record `rocksdb.sst.read.micros` and `rocksdb.file.{flush/compaction}.read.micros` - Fixed the default histogram in `RandomAccessFileReader` - New field `ReadOptions/IOOptions::io_activity`; Pass `ReadOptions` through paths under db open, flush and compaction to where we can prepare `IOOptions` and pass it to `RandomAccessFileReader` - Use `thread_status_util` for assertion in `DbStressFSWrapper` for continuous testing on we are passing correct `io_activity` under db open, flush and compaction Pull Request resolved: https://github.com/facebook/rocksdb/pull/11288 Test Plan: - **Stress test** - **Db bench 1: rocksdb.sst.read.micros COUNT ≈ sum of rocksdb.file.read.flush.micros's and rocksdb.file.read.compaction.micros's.** (without blob) - May not be exactly the same due to `HistogramStat::Add` only guarantees atomic not accuracy across threads. ``` ./db_bench -db=/dev/shm/testdb/ -statistics=true -benchmarks="fillseq" -key_size=32 -value_size=512 -num=50000 -write_buffer_size=655 -target_file_size_base=655 -disable_auto_compactions=false -compression_type=none -bloom_bits=3 (-use_plain_table=1 -prefix_size=10) ``` ``` // BlockBasedTable rocksdb.sst.read.micros P50 : 2.009374 P95 : 4.968548 P99 : 8.110362 P100 : 43.000000 COUNT : 40456 SUM : 114805 rocksdb.file.read.flush.micros P50 : 1.871841 P95 : 3.872407 P99 : 5.540541 P100 : 43.000000 COUNT : 2250 SUM : 6116 rocksdb.file.read.compaction.micros P50 : 2.023109 P95 : 5.029149 P99 : 8.196910 P100 : 26.000000 COUNT : 38206 SUM : 108689 // PlainTable Does not apply ``` - **Db bench 2: performance** **Read** SETUP: db with 900 files ``` ./db_bench -db=/dev/shm/testdb/ -benchmarks="fillseq" -key_size=32 -value_size=512 -num=50000 -write_buffer_size=655 -disable_auto_compactions=true -target_file_size_base=655 -compression_type=none ```run till convergence ``` ./db_bench -seed=1678564177044286 -use_existing_db=true -db=/dev/shm/testdb -benchmarks=readrandom[-X60] -statistics=true -num=1000000 -disable_auto_compactions=true -compression_type=none -bloom_bits=3 ``` Pre-change `readrandom [AVG 60 runs] : 21568 (± 248) ops/sec` Post-change (no regression, -0.3%) `readrandom [AVG 60 runs] : 21486 (± 236) ops/sec` **Compaction/Flush**run till convergence ``` ./db_bench -db=/dev/shm/testdb2/ -seed=1678564177044286 -benchmarks="fillseq[-X60]" -key_size=32 -value_size=512 -num=50000 -write_buffer_size=655 -disable_auto_compactions=false -target_file_size_base=655 -compression_type=none rocksdb.sst.read.micros COUNT : 33820 rocksdb.sst.read.flush.micros COUNT : 1800 rocksdb.sst.read.compaction.micros COUNT : 32020 ``` Pre-change `fillseq [AVG 46 runs] : 1391 (± 214) ops/sec; 0.7 (± 0.1) MB/sec` Post-change (no regression, ~-0.4%) `fillseq [AVG 46 runs] : 1385 (± 216) ops/sec; 0.7 (± 0.1) MB/sec` Reviewed By: ajkr Differential Revision: D44007011 Pulled By: hx235 fbshipit-source-id: a54c89e4846dfc9a135389edf3f3eedfea257132
2023-04-21 16:07:18 +00:00
table_reader_options.ioptions, read_options,
Improve / clean up meta block code & integrity (#9163) Summary: * Checksums are now checked on meta blocks unless specifically suppressed or not applicable (e.g. plain table). (Was other way around.) This means a number of cases that were not checking checksums now are, including direct read TableProperties in Version::GetTableProperties (fixed in meta_blocks ReadTableProperties), reading any block from PersistentCache (fixed in BlockFetcher), read TableProperties in SstFileDumper (ldb/sst_dump/BackupEngine) before table reader open, maybe more. * For that to work, I moved the global_seqno+TableProperties checksum logic to the shared table/ code, because that is used by many utilies such as SstFileDumper. * Also for that to work, we have to know when we're dealing with a block that has a checksum (trailer), so added that capability to Footer based on magic number, and from there BlockFetcher. * Knowledge of trailer presence has also fixed a problem where other table formats were reading blocks including bytes for a non-existant trailer--and awkwardly kind-of not using them, e.g. no shared code checking checksums. (BlockFetcher compression type was populated incorrectly.) Now we only read what is needed. * Minimized code duplication and differing/incompatible/awkward abstractions in meta_blocks.{cc,h} (e.g. SeekTo in metaindex block without parsing block handle) * Moved some meta block handling code from table_properties*.* * Moved some code specific to block-based table from shared table/ code to BlockBasedTable class. The checksum stuff means we can't completely separate it, but things that don't need to be in shared table/ code should not be. * Use unique_ptr rather than raw ptr in more places. (Note: you can std::move from unique_ptr to shared_ptr.) Without enhancements to GetPropertiesOfAllTablesTest (see below), net reduction of roughly 100 lines of code. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9163 Test Plan: existing tests and * Enhanced DBTablePropertiesTest.GetPropertiesOfAllTablesTest to verify that checksums are now checked on direct read of table properties by TableCache (new test would fail before this change) * Also enhanced DBTablePropertiesTest.GetPropertiesOfAllTablesTest to test putting table properties under old meta name * Also generally enhanced that same test to actually test what it was supposed to be testing already, by kicking things out of table cache when we don't want them there. Reviewed By: ajkr, mrambacher Differential Revision: D32514757 Pulled By: pdillinger fbshipit-source-id: 507964b9311d186ae8d1131182290cbd97a99fa9
2021-11-18 19:42:12 +00:00
BloomBlockBuilder::kBloomBlock,
&bloom_block_handle);
EXPECT_TRUE(s.ok());
BlockHandle index_block_handle;
Improve / clean up meta block code & integrity (#9163) Summary: * Checksums are now checked on meta blocks unless specifically suppressed or not applicable (e.g. plain table). (Was other way around.) This means a number of cases that were not checking checksums now are, including direct read TableProperties in Version::GetTableProperties (fixed in meta_blocks ReadTableProperties), reading any block from PersistentCache (fixed in BlockFetcher), read TableProperties in SstFileDumper (ldb/sst_dump/BackupEngine) before table reader open, maybe more. * For that to work, I moved the global_seqno+TableProperties checksum logic to the shared table/ code, because that is used by many utilies such as SstFileDumper. * Also for that to work, we have to know when we're dealing with a block that has a checksum (trailer), so added that capability to Footer based on magic number, and from there BlockFetcher. * Knowledge of trailer presence has also fixed a problem where other table formats were reading blocks including bytes for a non-existant trailer--and awkwardly kind-of not using them, e.g. no shared code checking checksums. (BlockFetcher compression type was populated incorrectly.) Now we only read what is needed. * Minimized code duplication and differing/incompatible/awkward abstractions in meta_blocks.{cc,h} (e.g. SeekTo in metaindex block without parsing block handle) * Moved some meta block handling code from table_properties*.* * Moved some code specific to block-based table from shared table/ code to BlockBasedTable class. The checksum stuff means we can't completely separate it, but things that don't need to be in shared table/ code should not be. * Use unique_ptr rather than raw ptr in more places. (Note: you can std::move from unique_ptr to shared_ptr.) Without enhancements to GetPropertiesOfAllTablesTest (see below), net reduction of roughly 100 lines of code. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9163 Test Plan: existing tests and * Enhanced DBTablePropertiesTest.GetPropertiesOfAllTablesTest to verify that checksums are now checked on direct read of table properties by TableCache (new test would fail before this change) * Also enhanced DBTablePropertiesTest.GetPropertiesOfAllTablesTest to test putting table properties under old meta name * Also generally enhanced that same test to actually test what it was supposed to be testing already, by kicking things out of table cache when we don't want them there. Reviewed By: ajkr, mrambacher Differential Revision: D32514757 Pulled By: pdillinger fbshipit-source-id: 507964b9311d186ae8d1131182290cbd97a99fa9
2021-11-18 19:42:12 +00:00
s = FindMetaBlockInFile(file.get(), file_size, kPlainTableMagicNumber,
Group rocksdb.sst.read.micros stat by IOActivity flush and compaction (#11288) Summary: **Context:** The existing stat rocksdb.sst.read.micros does not reflect each of compaction and flush cases but aggregate them, which is not so helpful for us to understand IO read behavior of each of them. **Summary** - Update `StopWatch` and `RandomAccessFileReader` to record `rocksdb.sst.read.micros` and `rocksdb.file.{flush/compaction}.read.micros` - Fixed the default histogram in `RandomAccessFileReader` - New field `ReadOptions/IOOptions::io_activity`; Pass `ReadOptions` through paths under db open, flush and compaction to where we can prepare `IOOptions` and pass it to `RandomAccessFileReader` - Use `thread_status_util` for assertion in `DbStressFSWrapper` for continuous testing on we are passing correct `io_activity` under db open, flush and compaction Pull Request resolved: https://github.com/facebook/rocksdb/pull/11288 Test Plan: - **Stress test** - **Db bench 1: rocksdb.sst.read.micros COUNT ≈ sum of rocksdb.file.read.flush.micros's and rocksdb.file.read.compaction.micros's.** (without blob) - May not be exactly the same due to `HistogramStat::Add` only guarantees atomic not accuracy across threads. ``` ./db_bench -db=/dev/shm/testdb/ -statistics=true -benchmarks="fillseq" -key_size=32 -value_size=512 -num=50000 -write_buffer_size=655 -target_file_size_base=655 -disable_auto_compactions=false -compression_type=none -bloom_bits=3 (-use_plain_table=1 -prefix_size=10) ``` ``` // BlockBasedTable rocksdb.sst.read.micros P50 : 2.009374 P95 : 4.968548 P99 : 8.110362 P100 : 43.000000 COUNT : 40456 SUM : 114805 rocksdb.file.read.flush.micros P50 : 1.871841 P95 : 3.872407 P99 : 5.540541 P100 : 43.000000 COUNT : 2250 SUM : 6116 rocksdb.file.read.compaction.micros P50 : 2.023109 P95 : 5.029149 P99 : 8.196910 P100 : 26.000000 COUNT : 38206 SUM : 108689 // PlainTable Does not apply ``` - **Db bench 2: performance** **Read** SETUP: db with 900 files ``` ./db_bench -db=/dev/shm/testdb/ -benchmarks="fillseq" -key_size=32 -value_size=512 -num=50000 -write_buffer_size=655 -disable_auto_compactions=true -target_file_size_base=655 -compression_type=none ```run till convergence ``` ./db_bench -seed=1678564177044286 -use_existing_db=true -db=/dev/shm/testdb -benchmarks=readrandom[-X60] -statistics=true -num=1000000 -disable_auto_compactions=true -compression_type=none -bloom_bits=3 ``` Pre-change `readrandom [AVG 60 runs] : 21568 (± 248) ops/sec` Post-change (no regression, -0.3%) `readrandom [AVG 60 runs] : 21486 (± 236) ops/sec` **Compaction/Flush**run till convergence ``` ./db_bench -db=/dev/shm/testdb2/ -seed=1678564177044286 -benchmarks="fillseq[-X60]" -key_size=32 -value_size=512 -num=50000 -write_buffer_size=655 -disable_auto_compactions=false -target_file_size_base=655 -compression_type=none rocksdb.sst.read.micros COUNT : 33820 rocksdb.sst.read.flush.micros COUNT : 1800 rocksdb.sst.read.compaction.micros COUNT : 32020 ``` Pre-change `fillseq [AVG 46 runs] : 1391 (± 214) ops/sec; 0.7 (± 0.1) MB/sec` Post-change (no regression, ~-0.4%) `fillseq [AVG 46 runs] : 1385 (± 216) ops/sec; 0.7 (± 0.1) MB/sec` Reviewed By: ajkr Differential Revision: D44007011 Pulled By: hx235 fbshipit-source-id: a54c89e4846dfc9a135389edf3f3eedfea257132
2023-04-21 16:07:18 +00:00
table_reader_options.ioptions, read_options,
Improve / clean up meta block code & integrity (#9163) Summary: * Checksums are now checked on meta blocks unless specifically suppressed or not applicable (e.g. plain table). (Was other way around.) This means a number of cases that were not checking checksums now are, including direct read TableProperties in Version::GetTableProperties (fixed in meta_blocks ReadTableProperties), reading any block from PersistentCache (fixed in BlockFetcher), read TableProperties in SstFileDumper (ldb/sst_dump/BackupEngine) before table reader open, maybe more. * For that to work, I moved the global_seqno+TableProperties checksum logic to the shared table/ code, because that is used by many utilies such as SstFileDumper. * Also for that to work, we have to know when we're dealing with a block that has a checksum (trailer), so added that capability to Footer based on magic number, and from there BlockFetcher. * Knowledge of trailer presence has also fixed a problem where other table formats were reading blocks including bytes for a non-existant trailer--and awkwardly kind-of not using them, e.g. no shared code checking checksums. (BlockFetcher compression type was populated incorrectly.) Now we only read what is needed. * Minimized code duplication and differing/incompatible/awkward abstractions in meta_blocks.{cc,h} (e.g. SeekTo in metaindex block without parsing block handle) * Moved some meta block handling code from table_properties*.* * Moved some code specific to block-based table from shared table/ code to BlockBasedTable class. The checksum stuff means we can't completely separate it, but things that don't need to be in shared table/ code should not be. * Use unique_ptr rather than raw ptr in more places. (Note: you can std::move from unique_ptr to shared_ptr.) Without enhancements to GetPropertiesOfAllTablesTest (see below), net reduction of roughly 100 lines of code. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9163 Test Plan: existing tests and * Enhanced DBTablePropertiesTest.GetPropertiesOfAllTablesTest to verify that checksums are now checked on direct read of table properties by TableCache (new test would fail before this change) * Also enhanced DBTablePropertiesTest.GetPropertiesOfAllTablesTest to test putting table properties under old meta name * Also generally enhanced that same test to actually test what it was supposed to be testing already, by kicking things out of table cache when we don't want them there. Reviewed By: ajkr, mrambacher Differential Revision: D32514757 Pulled By: pdillinger fbshipit-source-id: 507964b9311d186ae8d1131182290cbd97a99fa9
2021-11-18 19:42:12 +00:00
PlainTableIndexBuilder::kPlainTableIndexBlock,
&index_block_handle);
EXPECT_TRUE(s.ok());
}
auto& user_props = props->user_collected_properties;
auto encoding_type_prop =
user_props.find(PlainTablePropertyNames::kEncodingType);
assert(encoding_type_prop != user_props.end());
EncodingType encoding_type = static_cast<EncodingType>(
DecodeFixed32(encoding_type_prop->second.c_str()));
std::unique_ptr<PlainTableReader> new_reader(new TestPlainTableReader(
table_reader_options.env_options,
table_reader_options.internal_comparator, encoding_type, file_size,
Improve / clean up meta block code & integrity (#9163) Summary: * Checksums are now checked on meta blocks unless specifically suppressed or not applicable (e.g. plain table). (Was other way around.) This means a number of cases that were not checking checksums now are, including direct read TableProperties in Version::GetTableProperties (fixed in meta_blocks ReadTableProperties), reading any block from PersistentCache (fixed in BlockFetcher), read TableProperties in SstFileDumper (ldb/sst_dump/BackupEngine) before table reader open, maybe more. * For that to work, I moved the global_seqno+TableProperties checksum logic to the shared table/ code, because that is used by many utilies such as SstFileDumper. * Also for that to work, we have to know when we're dealing with a block that has a checksum (trailer), so added that capability to Footer based on magic number, and from there BlockFetcher. * Knowledge of trailer presence has also fixed a problem where other table formats were reading blocks including bytes for a non-existant trailer--and awkwardly kind-of not using them, e.g. no shared code checking checksums. (BlockFetcher compression type was populated incorrectly.) Now we only read what is needed. * Minimized code duplication and differing/incompatible/awkward abstractions in meta_blocks.{cc,h} (e.g. SeekTo in metaindex block without parsing block handle) * Moved some meta block handling code from table_properties*.* * Moved some code specific to block-based table from shared table/ code to BlockBasedTable class. The checksum stuff means we can't completely separate it, but things that don't need to be in shared table/ code should not be. * Use unique_ptr rather than raw ptr in more places. (Note: you can std::move from unique_ptr to shared_ptr.) Without enhancements to GetPropertiesOfAllTablesTest (see below), net reduction of roughly 100 lines of code. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9163 Test Plan: existing tests and * Enhanced DBTablePropertiesTest.GetPropertiesOfAllTablesTest to verify that checksums are now checked on direct read of table properties by TableCache (new test would fail before this change) * Also enhanced DBTablePropertiesTest.GetPropertiesOfAllTablesTest to test putting table properties under old meta name * Also generally enhanced that same test to actually test what it was supposed to be testing already, by kicking things out of table cache when we don't want them there. Reviewed By: ajkr, mrambacher Differential Revision: D32514757 Pulled By: pdillinger fbshipit-source-id: 507964b9311d186ae8d1131182290cbd97a99fa9
2021-11-18 19:42:12 +00:00
bloom_bits_per_key_, hash_table_ratio_, index_sparseness_,
std::move(props), std::move(file), table_reader_options.ioptions,
Fast path for detecting unchanged prefix_extractor (#9407) Summary: Fixes a major performance regression in 6.26, where extra CPU is spent in SliceTransform::AsString when reads involve a prefix_extractor (Get, MultiGet, Seek). Common case performance is now better than 6.25. This change creates a "fast path" for verifying that the current prefix extractor is unchanged and compatible with what was used to generate a table file. This fast path detects the common case by pointer comparison on the current prefix_extractor and a "known good" prefix extractor (if applicable) that is saved at the time the table reader is opened. The "known good" prefix extractor is saved as another shared_ptr copy (in an existing field, however) to ensure the pointer is not recycled. When the prefix_extractor has changed to a different instance but same compatible configuration (rare, odd), performance is still a regression compared to 6.25, but this is likely acceptable because of the oddity of such a case. The performance of incompatible prefix_extractor is essentially unchanged. Also fixed a minor case (ForwardIterator) where a prefix_extractor could be used via a raw pointer after being freed as a shared_ptr, if replaced via SetOptions. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9407 Test Plan: ## Performance Populate DB with `TEST_TMPDIR=/dev/shm/rocksdb ./db_bench -benchmarks=fillrandom -num=10000000 -disable_wal=1 -write_buffer_size=10000000 -bloom_bits=16 -compaction_style=2 -fifo_compaction_max_table_files_size_mb=10000 -fifo_compaction_allow_compaction=0 -prefix_size=12` Running head-to-head comparisons simultaneously with `TEST_TMPDIR=/dev/shm/rocksdb ./db_bench -use_existing_db -readonly -benchmarks=seekrandom -num=10000000 -duration=20 -disable_wal=1 -bloom_bits=16 -compaction_style=2 -fifo_compaction_max_table_files_size_mb=10000 -fifo_compaction_allow_compaction=0 -prefix_size=12` Below each is compared by ops/sec vs. baseline which is version 6.25 (multiple baseline runs because of variable machine load) v6.26: 4833 vs. 6698 (<- major regression!) v6.27: 4737 vs. 6397 (still) New: 6704 vs. 6461 (better than baseline in common case) Disabled fastpath: 4843 vs. 6389 (e.g. if prefix extractor instance changes but is still compatible) Changed prefix size (no usable filter) in new: 787 vs. 5927 Changed prefix size (no usable filter) in new & baseline: 773 vs. 784 Reviewed By: mrambacher Differential Revision: D33677812 Pulled By: pdillinger fbshipit-source-id: 571d9711c461fb97f957378a061b7e7dbc4d6a76
2022-01-21 19:36:36 +00:00
table_reader_options.prefix_extractor.get(), expect_bloom_not_match_,
store_index_in_file_, column_family_id_, column_family_name_));
*table = std::move(new_reader);
return s;
}
private:
int bloom_bits_per_key_;
double hash_table_ratio_;
size_t index_sparseness_;
bool store_index_in_file_;
bool* expect_bloom_not_match_;
const uint32_t column_family_id_;
const std::string column_family_name_;
};
TEST_P(PlainTableDBTest, BadOptions1) {
// Build with a prefix extractor
ASSERT_OK(Put("1000000000000foo", "v1"));
ASSERT_OK(dbfull()->TEST_FlushMemTable());
// Bad attempt to re-open without a prefix extractor
Options options = CurrentOptions();
options.prefix_extractor.reset();
ASSERT_EQ(
"Invalid argument: Prefix extractor is missing when opening a PlainTable "
"built using a prefix extractor",
TryReopen(&options).ToString());
// Bad attempt to re-open with different prefix extractor
options.prefix_extractor.reset(NewFixedPrefixTransform(6));
ASSERT_EQ(
"Invalid argument: Prefix extractor given doesn't match the one used to "
"build PlainTable",
TryReopen(&options).ToString());
// Correct prefix extractor
options.prefix_extractor.reset(NewFixedPrefixTransform(8));
Reopen(&options);
ASSERT_EQ("v1", Get("1000000000000foo"));
}
TEST_P(PlainTableDBTest, BadOptions2) {
Options options = CurrentOptions();
options.prefix_extractor.reset();
options.create_if_missing = true;
DestroyAndReopen(&options);
// Build without a prefix extractor
// (apparently works even if hash_table_ratio > 0)
ASSERT_OK(Put("1000000000000foo", "v1"));
// Build without a prefix extractor, this call will fail and returns the
// status for this bad attempt.
ASSERT_NOK(dbfull()->TEST_FlushMemTable());
// Bad attempt to re-open with hash_table_ratio > 0 and no prefix extractor
Status s = TryReopen(&options);
ASSERT_EQ(
"Not implemented: PlainTable requires a prefix extractor enable prefix "
"hash mode.",
s.ToString());
// OK to open with hash_table_ratio == 0 and no prefix extractor
PlainTableOptions plain_table_options;
plain_table_options.hash_table_ratio = 0;
options.table_factory.reset(NewPlainTableFactory(plain_table_options));
Reopen(&options);
ASSERT_EQ("v1", Get("1000000000000foo"));
// OK to open newly with a prefix_extractor and hash table; builds index
// in memory.
options = CurrentOptions();
Reopen(&options);
ASSERT_EQ("v1", Get("1000000000000foo"));
}
TEST_P(PlainTableDBTest, Flush) {
for (size_t huge_page_tlb_size = 0; huge_page_tlb_size <= 2 * 1024 * 1024;
huge_page_tlb_size += 2 * 1024 * 1024) {
for (EncodingType encoding_type : {kPlain, kPrefix}) {
for (int bloom = -1; bloom <= 117; bloom += 117) {
const int bloom_bits = std::max(bloom, 0);
const bool full_scan_mode = bloom < 0;
for (int total_order = 0; total_order <= 1; total_order++) {
for (int store_index_in_file = 0; store_index_in_file <= 1;
++store_index_in_file) {
Options options = CurrentOptions();
options.create_if_missing = true;
// Set only one bucket to force bucket conflict.
// Test index interval for the same prefix to be 1, 2 and 4
if (total_order) {
options.prefix_extractor.reset();
PlainTableOptions plain_table_options;
plain_table_options.user_key_len = 0;
plain_table_options.bloom_bits_per_key = bloom_bits;
plain_table_options.hash_table_ratio = 0;
plain_table_options.index_sparseness = 2;
plain_table_options.huge_page_tlb_size = huge_page_tlb_size;
plain_table_options.encoding_type = encoding_type;
plain_table_options.full_scan_mode = full_scan_mode;
plain_table_options.store_index_in_file = store_index_in_file;
options.table_factory.reset(
NewPlainTableFactory(plain_table_options));
} else {
PlainTableOptions plain_table_options;
plain_table_options.user_key_len = 0;
plain_table_options.bloom_bits_per_key = bloom_bits;
plain_table_options.hash_table_ratio = 0.75;
plain_table_options.index_sparseness = 16;
plain_table_options.huge_page_tlb_size = huge_page_tlb_size;
plain_table_options.encoding_type = encoding_type;
plain_table_options.full_scan_mode = full_scan_mode;
plain_table_options.store_index_in_file = store_index_in_file;
options.table_factory.reset(
NewPlainTableFactory(plain_table_options));
}
DestroyAndReopen(&options);
uint64_t int_num;
ASSERT_TRUE(dbfull()->GetIntProperty(
"rocksdb.estimate-table-readers-mem", &int_num));
ASSERT_EQ(int_num, 0U);
ASSERT_OK(Put("1000000000000foo", "v1"));
ASSERT_OK(Put("0000000000000bar", "v2"));
ASSERT_OK(Put("1000000000000foo", "v3"));
ASSERT_OK(dbfull()->TEST_FlushMemTable());
ASSERT_TRUE(dbfull()->GetIntProperty(
"rocksdb.estimate-table-readers-mem", &int_num));
ASSERT_GT(int_num, 0U);
TablePropertiesCollection ptc;
Prefer static_cast in place of most reinterpret_cast (#12308) Summary: The following are risks associated with pointer-to-pointer reinterpret_cast: * Can produce the "wrong result" (crash or memory corruption). IIRC, in theory this can happen for any up-cast or down-cast for a non-standard-layout type, though in practice would only happen for multiple inheritance cases (where the base class pointer might be "inside" the derived object). We don't use multiple inheritance a lot, but we do. * Can mask useful compiler errors upon code change, including converting between unrelated pointer types that you are expecting to be related, and converting between pointer and scalar types unintentionally. I can only think of some obscure cases where static_cast could be troublesome when it compiles as a replacement: * Going through `void*` could plausibly cause unnecessary or broken pointer arithmetic. Suppose we have `struct Derived: public Base1, public Base2`. If we have `Derived*` -> `void*` -> `Base2*` -> `Derived*` through reinterpret casts, this could plausibly work (though technical UB) assuming the `Base2*` is not dereferenced. Changing to static cast could introduce breaking pointer arithmetic. * Unnecessary (but safe) pointer arithmetic could arise in a case like `Derived*` -> `Base2*` -> `Derived*` where before the Base2 pointer might not have been dereferenced. This could potentially affect performance. With some light scripting, I tried replacing pointer-to-pointer reinterpret_casts with static_cast and kept the cases that still compile. Most occurrences of reinterpret_cast have successfully been changed (except for java/ and third-party/). 294 changed, 257 remain. A couple of related interventions included here: * Previously Cache::Handle was not actually derived from in the implementations and just used as a `void*` stand-in with reinterpret_cast. Now there is a relationship to allow static_cast. In theory, this could introduce pointer arithmetic (as described above) but is unlikely without multiple inheritance AND non-empty Cache::Handle. * Remove some unnecessary casts to void* as this is allowed to be implicit (for better or worse). Most of the remaining reinterpret_casts are for converting to/from raw bytes of objects. We could consider better idioms for these patterns in follow-up work. I wish there were a way to implement a template variant of static_cast that would only compile if no pointer arithmetic is generated, but best I can tell, this is not possible. AFAIK the best you could do is a dynamic check that the void* conversion after the static cast is unchanged. Pull Request resolved: https://github.com/facebook/rocksdb/pull/12308 Test Plan: existing tests, CI Reviewed By: ltamasi Differential Revision: D53204947 Pulled By: pdillinger fbshipit-source-id: 9de23e618263b0d5b9820f4e15966876888a16e2
2024-02-07 18:44:11 +00:00
ASSERT_OK(
static_cast<DB*>(dbfull())->GetPropertiesOfAllTables(&ptc));
ASSERT_EQ(1U, ptc.size());
auto row = ptc.begin();
auto tp = row->second;
if (full_scan_mode) {
// Does not support Get/Seek
std::unique_ptr<Iterator> iter(
dbfull()->NewIterator(ReadOptions()));
iter->SeekToFirst();
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("0000000000000bar", iter->key().ToString());
ASSERT_EQ("v2", iter->value().ToString());
iter->Next();
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("1000000000000foo", iter->key().ToString());
ASSERT_EQ("v3", iter->value().ToString());
iter->Next();
ASSERT_TRUE(!iter->Valid());
ASSERT_TRUE(iter->status().ok());
} else {
if (!store_index_in_file) {
ASSERT_EQ(total_order ? "4" : "12",
(tp->user_collected_properties)
.at("plain_table_hash_table_size"));
ASSERT_EQ("0", (tp->user_collected_properties)
.at("plain_table_sub_index_size"));
} else {
ASSERT_EQ("0", (tp->user_collected_properties)
.at("plain_table_hash_table_size"));
ASSERT_EQ("0", (tp->user_collected_properties)
.at("plain_table_sub_index_size"));
}
ASSERT_EQ("v3", Get("1000000000000foo"));
ASSERT_EQ("v2", Get("0000000000000bar"));
}
}
}
}
}
}
}
TEST_P(PlainTableDBTest, Flush2) {
for (size_t huge_page_tlb_size = 0; huge_page_tlb_size <= 2 * 1024 * 1024;
huge_page_tlb_size += 2 * 1024 * 1024) {
for (EncodingType encoding_type : {kPlain, kPrefix}) {
for (int bloom_bits = 0; bloom_bits <= 117; bloom_bits += 117) {
for (int total_order = 0; total_order <= 1; total_order++) {
for (int store_index_in_file = 0; store_index_in_file <= 1;
++store_index_in_file) {
if (encoding_type == kPrefix && total_order) {
continue;
}
if (!bloom_bits && store_index_in_file) {
continue;
}
if (total_order && store_index_in_file) {
continue;
}
bool expect_bloom_not_match = false;
Options options = CurrentOptions();
options.create_if_missing = true;
// Set only one bucket to force bucket conflict.
// Test index interval for the same prefix to be 1, 2 and 4
PlainTableOptions plain_table_options;
if (total_order) {
options.prefix_extractor = nullptr;
plain_table_options.hash_table_ratio = 0;
plain_table_options.index_sparseness = 2;
} else {
plain_table_options.hash_table_ratio = 0.75;
plain_table_options.index_sparseness = 16;
}
plain_table_options.user_key_len = kPlainTableVariableLength;
plain_table_options.bloom_bits_per_key = bloom_bits;
plain_table_options.huge_page_tlb_size = huge_page_tlb_size;
plain_table_options.encoding_type = encoding_type;
plain_table_options.store_index_in_file = store_index_in_file;
options.table_factory.reset(new TestPlainTableFactory(
&expect_bloom_not_match, plain_table_options,
0 /* column_family_id */, kDefaultColumnFamilyName));
DestroyAndReopen(&options);
ASSERT_OK(Put("0000000000000bar", "b"));
ASSERT_OK(Put("1000000000000foo", "v1"));
ASSERT_OK(dbfull()->TEST_FlushMemTable());
ASSERT_OK(Put("1000000000000foo", "v2"));
ASSERT_OK(dbfull()->TEST_FlushMemTable());
ASSERT_EQ("v2", Get("1000000000000foo"));
ASSERT_OK(Put("0000000000000eee", "v3"));
ASSERT_OK(dbfull()->TEST_FlushMemTable());
ASSERT_EQ("v3", Get("0000000000000eee"));
ASSERT_OK(Delete("0000000000000bar"));
ASSERT_OK(dbfull()->TEST_FlushMemTable());
ASSERT_EQ("NOT_FOUND", Get("0000000000000bar"));
ASSERT_OK(Put("0000000000000eee", "v5"));
ASSERT_OK(Put("9000000000000eee", "v5"));
ASSERT_OK(dbfull()->TEST_FlushMemTable());
ASSERT_EQ("v5", Get("0000000000000eee"));
// Test Bloom Filter
if (bloom_bits > 0) {
// Neither key nor value should exist.
expect_bloom_not_match = true;
ASSERT_EQ("NOT_FOUND", Get("5_not00000000bar"));
// Key doesn't exist any more but prefix exists.
if (total_order) {
ASSERT_EQ("NOT_FOUND", Get("1000000000000not"));
ASSERT_EQ("NOT_FOUND", Get("0000000000000not"));
}
expect_bloom_not_match = false;
}
}
}
}
}
}
}
TEST_P(PlainTableDBTest, Immortal) {
for (EncodingType encoding_type : {kPlain, kPrefix}) {
Options options = CurrentOptions();
options.create_if_missing = true;
options.max_open_files = -1;
// Set only one bucket to force bucket conflict.
// Test index interval for the same prefix to be 1, 2 and 4
PlainTableOptions plain_table_options;
plain_table_options.hash_table_ratio = 0.75;
plain_table_options.index_sparseness = 16;
plain_table_options.user_key_len = kPlainTableVariableLength;
plain_table_options.bloom_bits_per_key = 10;
plain_table_options.encoding_type = encoding_type;
options.table_factory.reset(NewPlainTableFactory(plain_table_options));
DestroyAndReopen(&options);
ASSERT_OK(Put("0000000000000bar", "b"));
ASSERT_OK(Put("1000000000000foo", "v1"));
ASSERT_OK(dbfull()->TEST_FlushMemTable());
int copied = 0;
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
"GetContext::SaveValue::PinSelf", [&](void* /*arg*/) { copied++; });
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
ASSERT_EQ("b", Get("0000000000000bar"));
ASSERT_EQ("v1", Get("1000000000000foo"));
ASSERT_EQ(2, copied);
copied = 0;
Close();
ASSERT_OK(ReopenForReadOnly(&options));
ASSERT_EQ("b", Get("0000000000000bar"));
ASSERT_EQ("v1", Get("1000000000000foo"));
ASSERT_EQ("NOT_FOUND", Get("1000000000000bar"));
if (mmap_mode()) {
ASSERT_EQ(0, copied);
} else {
ASSERT_EQ(2, copied);
}
ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
}
}
TEST_P(PlainTableDBTest, Iterator) {
for (size_t huge_page_tlb_size = 0; huge_page_tlb_size <= 2 * 1024 * 1024;
huge_page_tlb_size += 2 * 1024 * 1024) {
for (EncodingType encoding_type : {kPlain, kPrefix}) {
for (int bloom_bits = 0; bloom_bits <= 117; bloom_bits += 117) {
for (int total_order = 0; total_order <= 1; total_order++) {
if (encoding_type == kPrefix && total_order == 1) {
continue;
}
bool expect_bloom_not_match = false;
Options options = CurrentOptions();
options.create_if_missing = true;
// Set only one bucket to force bucket conflict.
// Test index interval for the same prefix to be 1, 2 and 4
if (total_order) {
options.prefix_extractor = nullptr;
PlainTableOptions plain_table_options;
plain_table_options.user_key_len = 16;
plain_table_options.bloom_bits_per_key = bloom_bits;
plain_table_options.hash_table_ratio = 0;
plain_table_options.index_sparseness = 2;
plain_table_options.huge_page_tlb_size = huge_page_tlb_size;
plain_table_options.encoding_type = encoding_type;
options.table_factory.reset(new TestPlainTableFactory(
&expect_bloom_not_match, plain_table_options,
0 /* column_family_id */, kDefaultColumnFamilyName));
} else {
PlainTableOptions plain_table_options;
plain_table_options.user_key_len = 16;
plain_table_options.bloom_bits_per_key = bloom_bits;
plain_table_options.hash_table_ratio = 0.75;
plain_table_options.index_sparseness = 16;
plain_table_options.huge_page_tlb_size = huge_page_tlb_size;
plain_table_options.encoding_type = encoding_type;
options.table_factory.reset(new TestPlainTableFactory(
&expect_bloom_not_match, plain_table_options,
0 /* column_family_id */, kDefaultColumnFamilyName));
}
DestroyAndReopen(&options);
ASSERT_OK(Put("1000000000foo002", "v_2"));
ASSERT_OK(Put("0000000000000bar", "random"));
ASSERT_OK(Put("1000000000foo001", "v1"));
ASSERT_OK(Put("3000000000000bar", "bar_v"));
ASSERT_OK(Put("1000000000foo003", "v__3"));
ASSERT_OK(Put("1000000000foo004", "v__4"));
ASSERT_OK(Put("1000000000foo005", "v__5"));
ASSERT_OK(Put("1000000000foo007", "v__7"));
ASSERT_OK(Put("1000000000foo008", "v__8"));
ASSERT_OK(dbfull()->TEST_FlushMemTable());
ASSERT_EQ("v1", Get("1000000000foo001"));
ASSERT_EQ("v__3", Get("1000000000foo003"));
Iterator* iter = dbfull()->NewIterator(ReadOptions());
iter->Seek("1000000000foo000");
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("1000000000foo001", iter->key().ToString());
ASSERT_EQ("v1", iter->value().ToString());
iter->Next();
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("1000000000foo002", iter->key().ToString());
ASSERT_EQ("v_2", iter->value().ToString());
iter->Next();
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("1000000000foo003", iter->key().ToString());
ASSERT_EQ("v__3", iter->value().ToString());
iter->Next();
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("1000000000foo004", iter->key().ToString());
ASSERT_EQ("v__4", iter->value().ToString());
iter->Seek("3000000000000bar");
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("3000000000000bar", iter->key().ToString());
ASSERT_EQ("bar_v", iter->value().ToString());
iter->Seek("1000000000foo000");
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("1000000000foo001", iter->key().ToString());
ASSERT_EQ("v1", iter->value().ToString());
iter->Seek("1000000000foo005");
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("1000000000foo005", iter->key().ToString());
ASSERT_EQ("v__5", iter->value().ToString());
iter->Seek("1000000000foo006");
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("1000000000foo007", iter->key().ToString());
ASSERT_EQ("v__7", iter->value().ToString());
iter->Seek("1000000000foo008");
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("1000000000foo008", iter->key().ToString());
ASSERT_EQ("v__8", iter->value().ToString());
if (total_order == 0) {
iter->Seek("1000000000foo009");
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("3000000000000bar", iter->key().ToString());
}
// Test Bloom Filter
if (bloom_bits > 0) {
if (!total_order) {
// Neither key nor value should exist.
expect_bloom_not_match = true;
iter->Seek("2not000000000bar");
ASSERT_TRUE(!iter->Valid());
ASSERT_EQ("NOT_FOUND", Get("2not000000000bar"));
expect_bloom_not_match = false;
} else {
expect_bloom_not_match = true;
ASSERT_EQ("NOT_FOUND", Get("2not000000000bar"));
expect_bloom_not_match = false;
}
}
ASSERT_OK(iter->status());
delete iter;
}
}
}
}
}
namespace {
std::string NthKey(size_t n, char filler) {
std::string rv(16, filler);
rv[0] = n % 10;
rv[1] = (n / 10) % 10;
rv[2] = (n / 100) % 10;
rv[3] = (n / 1000) % 10;
return rv;
}
} // anonymous namespace
TEST_P(PlainTableDBTest, BloomSchema) {
Options options = CurrentOptions();
options.create_if_missing = true;
for (int bloom_locality = 0; bloom_locality <= 1; bloom_locality++) {
options.bloom_locality = bloom_locality;
PlainTableOptions plain_table_options;
plain_table_options.user_key_len = 16;
plain_table_options.bloom_bits_per_key = 3; // high FP rate for test
plain_table_options.hash_table_ratio = 0.75;
plain_table_options.index_sparseness = 16;
plain_table_options.huge_page_tlb_size = 0;
plain_table_options.encoding_type = kPlain;
bool expect_bloom_not_match = false;
options.table_factory.reset(new TestPlainTableFactory(
&expect_bloom_not_match, plain_table_options, 0 /* column_family_id */,
kDefaultColumnFamilyName));
DestroyAndReopen(&options);
for (unsigned i = 0; i < 2345; ++i) {
ASSERT_OK(Put(NthKey(i, 'y'), "added"));
}
ASSERT_OK(dbfull()->TEST_FlushMemTable());
ASSERT_EQ("added", Get(NthKey(42, 'y')));
for (unsigned i = 0; i < 32; ++i) {
// Known pattern of Bloom filter false positives can detect schema change
// with high probability. Known FPs stuffed into bits:
uint32_t pattern;
if (!bloom_locality) {
pattern = 1785868347UL;
} else if (CACHE_LINE_SIZE == 64U) {
pattern = 2421694657UL;
} else if (CACHE_LINE_SIZE == 128U) {
pattern = 788710956UL;
} else {
ASSERT_EQ(CACHE_LINE_SIZE, 256U);
pattern = 163905UL;
}
bool expect_fp = pattern & (1UL << i);
// fprintf(stderr, "expect_fp@%u: %d\n", i, (int)expect_fp);
expect_bloom_not_match = !expect_fp;
ASSERT_EQ("NOT_FOUND", Get(NthKey(i, 'n')));
}
}
}
namespace {
std::string MakeLongKey(size_t length, char c) {
return std::string(length, c);
}
} // anonymous namespace
TEST_P(PlainTableDBTest, IteratorLargeKeys) {
Options options = CurrentOptions();
PlainTableOptions plain_table_options;
plain_table_options.user_key_len = 0;
plain_table_options.bloom_bits_per_key = 0;
plain_table_options.hash_table_ratio = 0;
options.table_factory.reset(NewPlainTableFactory(plain_table_options));
options.create_if_missing = true;
options.prefix_extractor.reset();
DestroyAndReopen(&options);
std::string key_list[] = {MakeLongKey(30, '0'), MakeLongKey(16, '1'),
MakeLongKey(32, '2'), MakeLongKey(60, '3'),
MakeLongKey(90, '4'), MakeLongKey(50, '5'),
MakeLongKey(26, '6')};
for (size_t i = 0; i < 7; i++) {
ASSERT_OK(Put(key_list[i], std::to_string(i)));
}
ASSERT_OK(dbfull()->TEST_FlushMemTable());
Iterator* iter = dbfull()->NewIterator(ReadOptions());
iter->Seek(key_list[0]);
for (size_t i = 0; i < 7; i++) {
ASSERT_TRUE(iter->Valid());
ASSERT_EQ(key_list[i], iter->key().ToString());
ASSERT_EQ(std::to_string(i), iter->value().ToString());
iter->Next();
}
ASSERT_TRUE(!iter->Valid());
ASSERT_OK(iter->status());
delete iter;
}
namespace {
std::string MakeLongKeyWithPrefix(size_t length, char c) {
return "00000000" + std::string(length - 8, c);
}
} // anonymous namespace
TEST_P(PlainTableDBTest, IteratorLargeKeysWithPrefix) {
Options options = CurrentOptions();
PlainTableOptions plain_table_options;
plain_table_options.user_key_len = 16;
plain_table_options.bloom_bits_per_key = 0;
plain_table_options.hash_table_ratio = 0.8;
plain_table_options.index_sparseness = 3;
plain_table_options.huge_page_tlb_size = 0;
plain_table_options.encoding_type = kPrefix;
options.table_factory.reset(NewPlainTableFactory(plain_table_options));
options.create_if_missing = true;
DestroyAndReopen(&options);
std::string key_list[] = {
MakeLongKeyWithPrefix(30, '0'), MakeLongKeyWithPrefix(16, '1'),
MakeLongKeyWithPrefix(32, '2'), MakeLongKeyWithPrefix(60, '3'),
MakeLongKeyWithPrefix(90, '4'), MakeLongKeyWithPrefix(50, '5'),
MakeLongKeyWithPrefix(26, '6')};
for (size_t i = 0; i < 7; i++) {
ASSERT_OK(Put(key_list[i], std::to_string(i)));
}
ASSERT_OK(dbfull()->TEST_FlushMemTable());
Iterator* iter = dbfull()->NewIterator(ReadOptions());
iter->Seek(key_list[0]);
for (size_t i = 0; i < 7; i++) {
ASSERT_TRUE(iter->Valid());
ASSERT_EQ(key_list[i], iter->key().ToString());
ASSERT_EQ(std::to_string(i), iter->value().ToString());
iter->Next();
}
ASSERT_TRUE(!iter->Valid());
ASSERT_OK(iter->status());
delete iter;
}
TEST_P(PlainTableDBTest, IteratorReverseSuffixComparator) {
Options options = CurrentOptions();
options.create_if_missing = true;
// Set only one bucket to force bucket conflict.
// Test index interval for the same prefix to be 1, 2 and 4
test::SimpleSuffixReverseComparator comp;
options.comparator = &comp;
DestroyAndReopen(&options);
ASSERT_OK(Put("1000000000foo002", "v_2"));
ASSERT_OK(Put("0000000000000bar", "random"));
ASSERT_OK(Put("1000000000foo001", "v1"));
ASSERT_OK(Put("3000000000000bar", "bar_v"));
ASSERT_OK(Put("1000000000foo003", "v__3"));
ASSERT_OK(Put("1000000000foo004", "v__4"));
ASSERT_OK(Put("1000000000foo005", "v__5"));
ASSERT_OK(Put("1000000000foo007", "v__7"));
ASSERT_OK(Put("1000000000foo008", "v__8"));
ASSERT_OK(dbfull()->TEST_FlushMemTable());
ASSERT_EQ("v1", Get("1000000000foo001"));
ASSERT_EQ("v__3", Get("1000000000foo003"));
Iterator* iter = dbfull()->NewIterator(ReadOptions());
iter->Seek("1000000000foo009");
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("1000000000foo008", iter->key().ToString());
ASSERT_EQ("v__8", iter->value().ToString());
iter->Next();
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("1000000000foo007", iter->key().ToString());
ASSERT_EQ("v__7", iter->value().ToString());
iter->Next();
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("1000000000foo005", iter->key().ToString());
ASSERT_EQ("v__5", iter->value().ToString());
iter->Next();
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("1000000000foo004", iter->key().ToString());
ASSERT_EQ("v__4", iter->value().ToString());
iter->Seek("3000000000000bar");
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("3000000000000bar", iter->key().ToString());
ASSERT_EQ("bar_v", iter->value().ToString());
iter->Seek("1000000000foo005");
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("1000000000foo005", iter->key().ToString());
ASSERT_EQ("v__5", iter->value().ToString());
iter->Seek("1000000000foo006");
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("1000000000foo005", iter->key().ToString());
ASSERT_EQ("v__5", iter->value().ToString());
iter->Seek("1000000000foo008");
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("1000000000foo008", iter->key().ToString());
ASSERT_EQ("v__8", iter->value().ToString());
iter->Seek("1000000000foo000");
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("3000000000000bar", iter->key().ToString());
delete iter;
}
TEST_P(PlainTableDBTest, HashBucketConflict) {
for (size_t huge_page_tlb_size = 0; huge_page_tlb_size <= 2 * 1024 * 1024;
huge_page_tlb_size += 2 * 1024 * 1024) {
for (unsigned char i = 1; i <= 3; i++) {
Options options = CurrentOptions();
options.create_if_missing = true;
// Set only one bucket to force bucket conflict.
// Test index interval for the same prefix to be 1, 2 and 4
PlainTableOptions plain_table_options;
plain_table_options.user_key_len = 16;
plain_table_options.bloom_bits_per_key = 0;
plain_table_options.hash_table_ratio = 0;
plain_table_options.index_sparseness = 2 ^ i;
plain_table_options.huge_page_tlb_size = huge_page_tlb_size;
options.table_factory.reset(NewPlainTableFactory(plain_table_options));
DestroyAndReopen(&options);
ASSERT_OK(Put("5000000000000fo0", "v1"));
ASSERT_OK(Put("5000000000000fo1", "v2"));
ASSERT_OK(Put("5000000000000fo2", "v"));
ASSERT_OK(Put("2000000000000fo0", "v3"));
ASSERT_OK(Put("2000000000000fo1", "v4"));
ASSERT_OK(Put("2000000000000fo2", "v"));
ASSERT_OK(Put("2000000000000fo3", "v"));
ASSERT_OK(dbfull()->TEST_FlushMemTable());
ASSERT_EQ("v1", Get("5000000000000fo0"));
ASSERT_EQ("v2", Get("5000000000000fo1"));
ASSERT_EQ("v3", Get("2000000000000fo0"));
ASSERT_EQ("v4", Get("2000000000000fo1"));
ASSERT_EQ("NOT_FOUND", Get("5000000000000bar"));
ASSERT_EQ("NOT_FOUND", Get("2000000000000bar"));
ASSERT_EQ("NOT_FOUND", Get("5000000000000fo8"));
ASSERT_EQ("NOT_FOUND", Get("2000000000000fo8"));
ReadOptions ro;
Iterator* iter = dbfull()->NewIterator(ro);
iter->Seek("5000000000000fo0");
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("5000000000000fo0", iter->key().ToString());
iter->Next();
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("5000000000000fo1", iter->key().ToString());
iter->Seek("5000000000000fo1");
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("5000000000000fo1", iter->key().ToString());
iter->Seek("2000000000000fo0");
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("2000000000000fo0", iter->key().ToString());
iter->Next();
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("2000000000000fo1", iter->key().ToString());
iter->Seek("2000000000000fo1");
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("2000000000000fo1", iter->key().ToString());
iter->Seek("2000000000000bar");
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("2000000000000fo0", iter->key().ToString());
iter->Seek("5000000000000bar");
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("5000000000000fo0", iter->key().ToString());
iter->Seek("2000000000000fo8");
ASSERT_TRUE(!iter->Valid() ||
options.comparator->Compare(iter->key(), "20000001") > 0);
iter->Seek("5000000000000fo8");
ASSERT_TRUE(!iter->Valid());
iter->Seek("1000000000000fo2");
ASSERT_TRUE(!iter->Valid());
iter->Seek("3000000000000fo2");
ASSERT_TRUE(!iter->Valid());
iter->Seek("8000000000000fo2");
ASSERT_TRUE(!iter->Valid());
ASSERT_OK(iter->status());
delete iter;
}
}
}
TEST_P(PlainTableDBTest, HashBucketConflictReverseSuffixComparator) {
for (size_t huge_page_tlb_size = 0; huge_page_tlb_size <= 2 * 1024 * 1024;
huge_page_tlb_size += 2 * 1024 * 1024) {
for (unsigned char i = 1; i <= 3; i++) {
Options options = CurrentOptions();
options.create_if_missing = true;
test::SimpleSuffixReverseComparator comp;
options.comparator = &comp;
// Set only one bucket to force bucket conflict.
// Test index interval for the same prefix to be 1, 2 and 4
PlainTableOptions plain_table_options;
plain_table_options.user_key_len = 16;
plain_table_options.bloom_bits_per_key = 0;
plain_table_options.hash_table_ratio = 0;
plain_table_options.index_sparseness = 2 ^ i;
plain_table_options.huge_page_tlb_size = huge_page_tlb_size;
options.table_factory.reset(NewPlainTableFactory(plain_table_options));
DestroyAndReopen(&options);
ASSERT_OK(Put("5000000000000fo0", "v1"));
ASSERT_OK(Put("5000000000000fo1", "v2"));
ASSERT_OK(Put("5000000000000fo2", "v"));
ASSERT_OK(Put("2000000000000fo0", "v3"));
ASSERT_OK(Put("2000000000000fo1", "v4"));
ASSERT_OK(Put("2000000000000fo2", "v"));
ASSERT_OK(Put("2000000000000fo3", "v"));
ASSERT_OK(dbfull()->TEST_FlushMemTable());
ASSERT_EQ("v1", Get("5000000000000fo0"));
ASSERT_EQ("v2", Get("5000000000000fo1"));
ASSERT_EQ("v3", Get("2000000000000fo0"));
ASSERT_EQ("v4", Get("2000000000000fo1"));
ASSERT_EQ("NOT_FOUND", Get("5000000000000bar"));
ASSERT_EQ("NOT_FOUND", Get("2000000000000bar"));
ASSERT_EQ("NOT_FOUND", Get("5000000000000fo8"));
ASSERT_EQ("NOT_FOUND", Get("2000000000000fo8"));
ReadOptions ro;
Iterator* iter = dbfull()->NewIterator(ro);
iter->Seek("5000000000000fo1");
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("5000000000000fo1", iter->key().ToString());
iter->Next();
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("5000000000000fo0", iter->key().ToString());
iter->Seek("5000000000000fo1");
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("5000000000000fo1", iter->key().ToString());
iter->Seek("2000000000000fo1");
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("2000000000000fo1", iter->key().ToString());
iter->Next();
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("2000000000000fo0", iter->key().ToString());
iter->Seek("2000000000000fo1");
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("2000000000000fo1", iter->key().ToString());
iter->Seek("2000000000000var");
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("2000000000000fo3", iter->key().ToString());
iter->Seek("5000000000000var");
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("5000000000000fo2", iter->key().ToString());
std::string seek_key = "2000000000000bar";
iter->Seek(seek_key);
ASSERT_TRUE(!iter->Valid() ||
options.prefix_extractor->Transform(iter->key()) !=
options.prefix_extractor->Transform(seek_key));
iter->Seek("1000000000000fo2");
ASSERT_TRUE(!iter->Valid());
iter->Seek("3000000000000fo2");
ASSERT_TRUE(!iter->Valid());
iter->Seek("8000000000000fo2");
ASSERT_TRUE(!iter->Valid());
ASSERT_OK(iter->status());
delete iter;
}
}
}
TEST_P(PlainTableDBTest, NonExistingKeyToNonEmptyBucket) {
Options options = CurrentOptions();
options.create_if_missing = true;
// Set only one bucket to force bucket conflict.
// Test index interval for the same prefix to be 1, 2 and 4
PlainTableOptions plain_table_options;
plain_table_options.user_key_len = 16;
plain_table_options.bloom_bits_per_key = 0;
plain_table_options.hash_table_ratio = 0;
plain_table_options.index_sparseness = 5;
options.table_factory.reset(NewPlainTableFactory(plain_table_options));
DestroyAndReopen(&options);
ASSERT_OK(Put("5000000000000fo0", "v1"));
ASSERT_OK(Put("5000000000000fo1", "v2"));
ASSERT_OK(Put("5000000000000fo2", "v3"));
ASSERT_OK(dbfull()->TEST_FlushMemTable());
ASSERT_EQ("v1", Get("5000000000000fo0"));
ASSERT_EQ("v2", Get("5000000000000fo1"));
ASSERT_EQ("v3", Get("5000000000000fo2"));
ASSERT_EQ("NOT_FOUND", Get("8000000000000bar"));
ASSERT_EQ("NOT_FOUND", Get("1000000000000bar"));
Iterator* iter = dbfull()->NewIterator(ReadOptions());
iter->Seek("5000000000000bar");
ASSERT_TRUE(iter->Valid());
ASSERT_EQ("5000000000000fo0", iter->key().ToString());
iter->Seek("5000000000000fo8");
ASSERT_TRUE(!iter->Valid());
iter->Seek("1000000000000fo2");
ASSERT_TRUE(!iter->Valid());
iter->Seek("8000000000000fo2");
ASSERT_TRUE(!iter->Valid());
ASSERT_OK(iter->status());
delete iter;
}
static std::string Key(int i) {
char buf[100];
snprintf(buf, sizeof(buf), "key_______%06d", i);
return std::string(buf);
}
TEST_P(PlainTableDBTest, CompactionTrigger) {
Options options = CurrentOptions();
options.write_buffer_size = 120 << 10; // 120KB
options.num_levels = 3;
options.level0_file_num_compaction_trigger = 3;
Reopen(&options);
Random rnd(301);
for (int num = 0; num < options.level0_file_num_compaction_trigger - 1;
num++) {
std::vector<std::string> values;
// Write 120KB (10 values, each 12K)
for (int i = 0; i < 10; i++) {
values.push_back(rnd.RandomString(12 << 10));
ASSERT_OK(Put(Key(i), values[i]));
}
ASSERT_OK(Put(Key(999), ""));
ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable());
ASSERT_EQ(NumTableFilesAtLevel(0), num + 1);
}
// generate one more file in level-0, and should trigger level-0 compaction
std::vector<std::string> values;
for (int i = 0; i < 12; i++) {
values.push_back(rnd.RandomString(10000));
ASSERT_OK(Put(Key(i), values[i]));
}
ASSERT_OK(Put(Key(999), ""));
ASSERT_OK(dbfull()->TEST_WaitForCompact());
ASSERT_EQ(NumTableFilesAtLevel(0), 0);
ASSERT_EQ(NumTableFilesAtLevel(1), 1);
}
TEST_P(PlainTableDBTest, AdaptiveTable) {
Options options = CurrentOptions();
options.create_if_missing = true;
options.table_factory.reset(NewPlainTableFactory());
DestroyAndReopen(&options);
ASSERT_OK(Put("1000000000000foo", "v1"));
ASSERT_OK(Put("0000000000000bar", "v2"));
ASSERT_OK(Put("1000000000000foo", "v3"));
ASSERT_OK(dbfull()->TEST_FlushMemTable());
options.create_if_missing = false;
std::shared_ptr<TableFactory> block_based_factory(
NewBlockBasedTableFactory());
std::shared_ptr<TableFactory> plain_table_factory(NewPlainTableFactory());
std::shared_ptr<TableFactory> dummy_factory;
options.table_factory.reset(NewAdaptiveTableFactory(
block_based_factory, block_based_factory, plain_table_factory));
Reopen(&options);
ASSERT_EQ("v3", Get("1000000000000foo"));
ASSERT_EQ("v2", Get("0000000000000bar"));
ASSERT_OK(Put("2000000000000foo", "v4"));
ASSERT_OK(Put("3000000000000bar", "v5"));
ASSERT_OK(dbfull()->TEST_FlushMemTable());
ASSERT_EQ("v4", Get("2000000000000foo"));
ASSERT_EQ("v5", Get("3000000000000bar"));
Reopen(&options);
ASSERT_EQ("v3", Get("1000000000000foo"));
ASSERT_EQ("v2", Get("0000000000000bar"));
ASSERT_EQ("v4", Get("2000000000000foo"));
ASSERT_EQ("v5", Get("3000000000000bar"));
options.paranoid_checks = false;
options.table_factory.reset(NewBlockBasedTableFactory());
Reopen(&options);
ASSERT_NE("v3", Get("1000000000000foo"));
options.paranoid_checks = false;
options.table_factory.reset(NewPlainTableFactory());
Reopen(&options);
ASSERT_NE("v5", Get("3000000000000bar"));
}
INSTANTIATE_TEST_CASE_P(PlainTableDBTest, PlainTableDBTest, ::testing::Bool());
} // namespace ROCKSDB_NAMESPACE
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();
}