rocksdb/cache/lru_cache.h
Peter Dillinger df5dc73bec Don't hold DB mutex for block cache entry stat scans (#8538)
Summary:
I previously didn't notice the DB mutex was being held during
block cache entry stat scans, probably because I primarily checked for
read performance regressions, because they require the block cache and
are traditionally latency-sensitive.

This change does some refactoring to avoid holding DB mutex and to
avoid triggering and waiting for a scan in GetProperty("rocksdb.cfstats").
Some tests have to be updated because now the stats collector is
populated in the Cache aggressively on DB startup rather than lazily.
(I hope to clean up some of this added complexity in the future.)

This change also ensures proper treatment of need_out_of_mutex for
non-int DB properties.

Pull Request resolved: https://github.com/facebook/rocksdb/pull/8538

Test Plan:
Added unit test logic that uses sync points to fail if the DB mutex
is held during a scan, covering the various ways that a scan might be
triggered.

Performance test - the known impact to holding the DB mutex is on
TransactionDB, and the easiest way to see the impact is to hack the
scan code to almost always miss and take an artificially long time
scanning. Here I've injected an unconditional 5s sleep at the call to
ApplyToAllEntries.

Before (hacked):

    $ TEST_TMPDIR=/dev/shm ./db_bench.base_xxx -benchmarks=randomtransaction,stats -cache_index_and_filter_blocks=1 -bloom_bits=10 -partition_index_and_filters=1 -duration=30 -stats_dump_period_sec=12 -cache_size=100000000 -statistics -transaction_db 2>&1 | egrep 'db.db.write.micros|micros/op'
    randomtransaction :     433.219 micros/op 2308 ops/sec;    0.1 MB/s ( transactions:78999 aborts:0)
    rocksdb.db.write.micros P50 : 16.135883 P95 : 36.622503 P99 : 66.036115 P100 : 5000614.000000 COUNT : 149677 SUM : 8364856
    $ TEST_TMPDIR=/dev/shm ./db_bench.base_xxx -benchmarks=randomtransaction,stats -cache_index_and_filter_blocks=1 -bloom_bits=10 -partition_index_and_filters=1 -duration=30 -stats_dump_period_sec=12 -cache_size=100000000 -statistics -transaction_db 2>&1 | egrep 'db.db.write.micros|micros/op'
    randomtransaction :     448.802 micros/op 2228 ops/sec;    0.1 MB/s ( transactions:75999 aborts:0)
    rocksdb.db.write.micros P50 : 16.629221 P95 : 37.320607 P99 : 72.144341 P100 : 5000871.000000 COUNT : 143995 SUM : 13472323

Notice the 5s P100 write time.

After (hacked):

    $ TEST_TMPDIR=/dev/shm ./db_bench.new_xxx -benchmarks=randomtransaction,stats -cache_index_and_filter_blocks=1 -bloom_bits=10 -partition_index_and_filters=1 -duration=30 -stats_dump_period_sec=12 -cache_size=100000000 -statistics -transaction_db 2>&1 | egrep 'db.db.write.micros|micros/op'
    randomtransaction :     303.645 micros/op 3293 ops/sec;    0.1 MB/s ( transactions:98999 aborts:0)
    rocksdb.db.write.micros P50 : 16.061871 P95 : 33.978834 P99 : 60.018017 P100 : 616315.000000 COUNT : 187619 SUM : 4097407
    $ TEST_TMPDIR=/dev/shm ./db_bench.new_xxx -benchmarks=randomtransaction,stats -cache_index_and_filter_blocks=1 -bloom_bits=10 -partition_index_and_filters=1 -duration=30 -stats_dump_period_sec=12 -cache_size=100000000 -statistics -transaction_db 2>&1 | egrep 'db.db.write.micros|micros/op'
    randomtransaction :     310.383 micros/op 3221 ops/sec;    0.1 MB/s ( transactions:96999 aborts:0)
    rocksdb.db.write.micros P50 : 16.270026 P95 : 35.786844 P99 : 64.302878 P100 : 603088.000000 COUNT : 183819 SUM : 4095918

P100 write is now ~0.6s. Not good, but it's the same even if I completely bypass all the scanning code:

    $ TEST_TMPDIR=/dev/shm ./db_bench.new_skip -benchmarks=randomtransaction,stats -cache_index_and_filter_blocks=1 -bloom_bits=10 -partition_index_and_filters=1 -duration=30 -stats_dump_period_sec=12 -cache_size=100000000 -statistics -transaction_db 2>&1 | egrep 'db.db.write.micros|micros/op'
    randomtransaction :     311.365 micros/op 3211 ops/sec;    0.1 MB/s ( transactions:96999 aborts:0)
    rocksdb.db.write.micros P50 : 16.274362 P95 : 36.221184 P99 : 68.809783 P100 : 649808.000000 COUNT : 183819 SUM : 4156767
    $ TEST_TMPDIR=/dev/shm ./db_bench.new_skip -benchmarks=randomtransaction,stats -cache_index_and_filter_blocks=1 -bloom_bits=10 -partition_index_and_filters=1 -duration=30 -stats_dump_period_sec=12 -cache_size=100000000 -statistics -transaction_db 2>&1 | egrep 'db.db.write.micros|micros/op'
    randomtransaction :     308.395 micros/op 3242 ops/sec;    0.1 MB/s ( transactions:97999 aborts:0)
    rocksdb.db.write.micros P50 : 16.106222 P95 : 37.202403 P99 : 67.081875 P100 : 598091.000000 COUNT : 185714 SUM : 4098832

No substantial difference.

Reviewed By: siying

Differential Revision: D29738847

Pulled By: pdillinger

fbshipit-source-id: 1c5c155f5a1b62e4fea0fd4eeb515a8b7474027b
2021-07-16 14:13:08 -07:00

481 lines
17 KiB
C++

// 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.
#pragma once
#include <memory>
#include <string>
#include "cache/sharded_cache.h"
#include "port/lang.h"
#include "port/malloc.h"
#include "port/port.h"
#include "rocksdb/secondary_cache.h"
#include "util/autovector.h"
namespace ROCKSDB_NAMESPACE {
// LRU cache implementation. This class is not thread-safe.
// An entry is a variable length heap-allocated structure.
// Entries are referenced by cache and/or by any external entity.
// The cache keeps all its entries in a hash table. Some elements
// are also stored on LRU list.
//
// LRUHandle can be in these states:
// 1. Referenced externally AND in hash table.
// In that case the entry is *not* in the LRU list
// (refs >= 1 && in_cache == true)
// 2. Not referenced externally AND in hash table.
// In that case the entry is in the LRU list and can be freed.
// (refs == 0 && in_cache == true)
// 3. Referenced externally AND not in hash table.
// In that case the entry is not in the LRU list and not in hash table.
// The entry can be freed when refs becomes 0.
// (refs >= 1 && in_cache == false)
//
// All newly created LRUHandles are in state 1. If you call
// LRUCacheShard::Release on entry in state 1, it will go into state 2.
// To move from state 1 to state 3, either call LRUCacheShard::Erase or
// LRUCacheShard::Insert with the same key (but possibly different value).
// To move from state 2 to state 1, use LRUCacheShard::Lookup.
// Before destruction, make sure that no handles are in state 1. This means
// that any successful LRUCacheShard::Lookup/LRUCacheShard::Insert have a
// matching LRUCache::Release (to move into state 2) or LRUCacheShard::Erase
// (to move into state 3).
struct LRUHandle {
void* value;
union Info {
Info() {}
~Info() {}
Cache::DeleterFn deleter;
const ShardedCache::CacheItemHelper* helper;
} info_;
// An entry is not added to the LRUHandleTable until the secondary cache
// lookup is complete, so its safe to have this union.
union {
LRUHandle* next_hash;
SecondaryCacheResultHandle* sec_handle;
};
LRUHandle* next;
LRUHandle* prev;
size_t charge; // TODO(opt): Only allow uint32_t?
size_t key_length;
// The hash of key(). Used for fast sharding and comparisons.
uint32_t hash;
// The number of external refs to this entry. The cache itself is not counted.
uint32_t refs;
enum Flags : uint8_t {
// Whether this entry is referenced by the hash table.
IN_CACHE = (1 << 0),
// Whether this entry is high priority entry.
IS_HIGH_PRI = (1 << 1),
// Whether this entry is in high-pri pool.
IN_HIGH_PRI_POOL = (1 << 2),
// Whether this entry has had any lookups (hits).
HAS_HIT = (1 << 3),
// Can this be inserted into the secondary cache
IS_SECONDARY_CACHE_COMPATIBLE = (1 << 4),
// Is the handle still being read from a lower tier
IS_PENDING = (1 << 5),
// Has the item been promoted from a lower tier
IS_PROMOTED = (1 << 6),
};
uint8_t flags;
#ifdef __SANITIZE_THREAD__
// TSAN can report a false data race on flags, where one thread is writing
// to one of the mutable bits and another thread is reading this immutable
// bit. So precisely suppress that TSAN warning, we separate out this bit
// during TSAN runs.
bool is_secondary_cache_compatible_for_tsan;
#endif // __SANITIZE_THREAD__
// Beginning of the key (MUST BE THE LAST FIELD IN THIS STRUCT!)
char key_data[1];
Slice key() const { return Slice(key_data, key_length); }
// Increase the reference count by 1.
void Ref() { refs++; }
// Just reduce the reference count by 1. Return true if it was last reference.
bool Unref() {
assert(refs > 0);
refs--;
return refs == 0;
}
// Return true if there are external refs, false otherwise.
bool HasRefs() const { return refs > 0; }
bool InCache() const { return flags & IN_CACHE; }
bool IsHighPri() const { return flags & IS_HIGH_PRI; }
bool InHighPriPool() const { return flags & IN_HIGH_PRI_POOL; }
bool HasHit() const { return flags & HAS_HIT; }
bool IsSecondaryCacheCompatible() const {
#ifdef __SANITIZE_THREAD__
return is_secondary_cache_compatible_for_tsan;
#else
return flags & IS_SECONDARY_CACHE_COMPATIBLE;
#endif // __SANITIZE_THREAD__
}
bool IsPending() const { return flags & IS_PENDING; }
bool IsPromoted() const { return flags & IS_PROMOTED; }
void SetInCache(bool in_cache) {
if (in_cache) {
flags |= IN_CACHE;
} else {
flags &= ~IN_CACHE;
}
}
void SetPriority(Cache::Priority priority) {
if (priority == Cache::Priority::HIGH) {
flags |= IS_HIGH_PRI;
} else {
flags &= ~IS_HIGH_PRI;
}
}
void SetInHighPriPool(bool in_high_pri_pool) {
if (in_high_pri_pool) {
flags |= IN_HIGH_PRI_POOL;
} else {
flags &= ~IN_HIGH_PRI_POOL;
}
}
void SetHit() { flags |= HAS_HIT; }
void SetSecondaryCacheCompatible(bool compat) {
if (compat) {
flags |= IS_SECONDARY_CACHE_COMPATIBLE;
} else {
flags &= ~IS_SECONDARY_CACHE_COMPATIBLE;
}
#ifdef __SANITIZE_THREAD__
is_secondary_cache_compatible_for_tsan = compat;
#endif // __SANITIZE_THREAD__
}
void SetIncomplete(bool incomp) {
if (incomp) {
flags |= IS_PENDING;
} else {
flags &= ~IS_PENDING;
}
}
void SetPromoted(bool promoted) {
if (promoted) {
flags |= IS_PROMOTED;
} else {
flags &= ~IS_PROMOTED;
}
}
void Free() {
assert(refs == 0);
#ifdef __SANITIZE_THREAD__
// Here we can safely assert they are the same without a data race reported
assert(((flags & IS_SECONDARY_CACHE_COMPATIBLE) != 0) ==
is_secondary_cache_compatible_for_tsan);
#endif // __SANITIZE_THREAD__
if (!IsSecondaryCacheCompatible() && info_.deleter) {
(*info_.deleter)(key(), value);
} else if (IsSecondaryCacheCompatible()) {
if (IsPending()) {
assert(sec_handle != nullptr);
SecondaryCacheResultHandle* tmp_sec_handle = sec_handle;
tmp_sec_handle->Wait();
value = tmp_sec_handle->Value();
delete tmp_sec_handle;
}
if (value) {
(*info_.helper->del_cb)(key(), value);
}
}
delete[] reinterpret_cast<char*>(this);
}
// Calculate the memory usage by metadata
inline size_t CalcTotalCharge(
CacheMetadataChargePolicy metadata_charge_policy) {
size_t meta_charge = 0;
if (metadata_charge_policy == kFullChargeCacheMetadata) {
#ifdef ROCKSDB_MALLOC_USABLE_SIZE
meta_charge += malloc_usable_size(static_cast<void*>(this));
#else
// This is the size that is used when a new handle is created
meta_charge += sizeof(LRUHandle) - 1 + key_length;
#endif
}
return charge + meta_charge;
}
};
// We provide our own simple hash table since it removes a whole bunch
// of porting hacks and is also faster than some of the built-in hash
// table implementations in some of the compiler/runtime combinations
// we have tested. E.g., readrandom speeds up by ~5% over the g++
// 4.4.3's builtin hashtable.
class LRUHandleTable {
public:
// If the table uses more hash bits than `max_upper_hash_bits`,
// it will eat into the bits used for sharding, which are constant
// for a given LRUHandleTable.
explicit LRUHandleTable(int max_upper_hash_bits);
~LRUHandleTable();
LRUHandle* Lookup(const Slice& key, uint32_t hash);
LRUHandle* Insert(LRUHandle* h);
LRUHandle* Remove(const Slice& key, uint32_t hash);
template <typename T>
void ApplyToEntriesRange(T func, uint32_t index_begin, uint32_t index_end) {
for (uint32_t i = index_begin; i < index_end; i++) {
LRUHandle* h = list_[i];
while (h != nullptr) {
auto n = h->next_hash;
assert(h->InCache());
func(h);
h = n;
}
}
}
int GetLengthBits() const { return length_bits_; }
private:
// Return a pointer to slot that points to a cache entry that
// matches key/hash. If there is no such cache entry, return a
// pointer to the trailing slot in the corresponding linked list.
LRUHandle** FindPointer(const Slice& key, uint32_t hash);
void Resize();
// Number of hash bits (upper because lower bits used for sharding)
// used for table index. Length == 1 << length_bits_
int length_bits_;
// The table consists of an array of buckets where each bucket is
// a linked list of cache entries that hash into the bucket.
std::unique_ptr<LRUHandle*[]> list_;
// Number of elements currently in the table
uint32_t elems_;
// Set from max_upper_hash_bits (see constructor)
const int max_length_bits_;
};
// A single shard of sharded cache.
class ALIGN_AS(CACHE_LINE_SIZE) LRUCacheShard final : public CacheShard {
public:
LRUCacheShard(size_t capacity, bool strict_capacity_limit,
double high_pri_pool_ratio, bool use_adaptive_mutex,
CacheMetadataChargePolicy metadata_charge_policy,
int max_upper_hash_bits,
const std::shared_ptr<SecondaryCache>& secondary_cache);
virtual ~LRUCacheShard() override = default;
// Separate from constructor so caller can easily make an array of LRUCache
// if current usage is more than new capacity, the function will attempt to
// free the needed space
virtual void SetCapacity(size_t capacity) override;
// Set the flag to reject insertion if cache if full.
virtual void SetStrictCapacityLimit(bool strict_capacity_limit) override;
// Set percentage of capacity reserved for high-pri cache entries.
void SetHighPriorityPoolRatio(double high_pri_pool_ratio);
// Like Cache methods, but with an extra "hash" parameter.
virtual Status Insert(const Slice& key, uint32_t hash, void* value,
size_t charge, Cache::DeleterFn deleter,
Cache::Handle** handle,
Cache::Priority priority) override {
return Insert(key, hash, value, charge, deleter, nullptr, handle, priority);
}
virtual Status Insert(const Slice& key, uint32_t hash, void* value,
const Cache::CacheItemHelper* helper, size_t charge,
Cache::Handle** handle,
Cache::Priority priority) override {
assert(helper);
return Insert(key, hash, value, charge, nullptr, helper, handle, priority);
}
// If helper_cb is null, the values of the following arguments don't
// matter
virtual Cache::Handle* Lookup(const Slice& key, uint32_t hash,
const ShardedCache::CacheItemHelper* helper,
const ShardedCache::CreateCallback& create_cb,
ShardedCache::Priority priority,
bool wait) override;
virtual Cache::Handle* Lookup(const Slice& key, uint32_t hash) override {
return Lookup(key, hash, nullptr, nullptr, Cache::Priority::LOW, true);
}
virtual bool Release(Cache::Handle* handle, bool /*useful*/,
bool force_erase) override {
return Release(handle, force_erase);
}
virtual bool IsReady(Cache::Handle* /*handle*/) override;
virtual void Wait(Cache::Handle* /*handle*/) override {}
virtual bool Ref(Cache::Handle* handle) override;
virtual bool Release(Cache::Handle* handle,
bool force_erase = false) override;
virtual void Erase(const Slice& key, uint32_t hash) override;
// Although in some platforms the update of size_t is atomic, to make sure
// GetUsage() and GetPinnedUsage() work correctly under any platform, we'll
// protect them with mutex_.
virtual size_t GetUsage() const override;
virtual size_t GetPinnedUsage() const override;
virtual void ApplyToSomeEntries(
const std::function<void(const Slice& key, void* value, size_t charge,
DeleterFn deleter)>& callback,
uint32_t average_entries_per_lock, uint32_t* state) override;
virtual void EraseUnRefEntries() override;
virtual std::string GetPrintableOptions() const override;
void TEST_GetLRUList(LRUHandle** lru, LRUHandle** lru_low_pri);
// Retrieves number of elements in LRU, for unit test purpose only
// not threadsafe
size_t TEST_GetLRUSize();
// Retrieves high pri pool ratio
double GetHighPriPoolRatio();
private:
friend class LRUCache;
// Insert an item into the hash table and, if handle is null, insert into
// the LRU list. Older items are evicted as necessary. If the cache is full
// and free_handle_on_fail is true, the item is deleted and handle is set to.
Status InsertItem(LRUHandle* item, Cache::Handle** handle,
bool free_handle_on_fail);
Status Insert(const Slice& key, uint32_t hash, void* value, size_t charge,
DeleterFn deleter, const Cache::CacheItemHelper* helper,
Cache::Handle** handle, Cache::Priority priority);
// Promote an item looked up from the secondary cache to the LRU cache. The
// item is only inserted into the hash table and not the LRU list, and only
// if the cache is not at full capacity, as is the case during Insert. The
// caller should hold a reference on the LRUHandle. When the caller releases
// the last reference, the item is added to the LRU list.
// The item is promoted to the high pri or low pri pool as specified by the
// caller in Lookup.
void Promote(LRUHandle* e);
void LRU_Remove(LRUHandle* e);
void LRU_Insert(LRUHandle* e);
// Overflow the last entry in high-pri pool to low-pri pool until size of
// high-pri pool is no larger than the size specify by high_pri_pool_pct.
void MaintainPoolSize();
// Free some space following strict LRU policy until enough space
// to hold (usage_ + charge) is freed or the lru list is empty
// This function is not thread safe - it needs to be executed while
// holding the mutex_
void EvictFromLRU(size_t charge, autovector<LRUHandle*>* deleted);
// Initialized before use.
size_t capacity_;
// Memory size for entries in high-pri pool.
size_t high_pri_pool_usage_;
// Whether to reject insertion if cache reaches its full capacity.
bool strict_capacity_limit_;
// Ratio of capacity reserved for high priority cache entries.
double high_pri_pool_ratio_;
// High-pri pool size, equals to capacity * high_pri_pool_ratio.
// Remember the value to avoid recomputing each time.
double high_pri_pool_capacity_;
// Dummy head of LRU list.
// lru.prev is newest entry, lru.next is oldest entry.
// LRU contains items which can be evicted, ie reference only by cache
LRUHandle lru_;
// Pointer to head of low-pri pool in LRU list.
LRUHandle* lru_low_pri_;
// ------------^^^^^^^^^^^^^-----------
// Not frequently modified data members
// ------------------------------------
//
// We separate data members that are updated frequently from the ones that
// are not frequently updated so that they don't share the same cache line
// which will lead into false cache sharing
//
// ------------------------------------
// Frequently modified data members
// ------------vvvvvvvvvvvvv-----------
LRUHandleTable table_;
// Memory size for entries residing in the cache
size_t usage_;
// Memory size for entries residing only in the LRU list
size_t lru_usage_;
// mutex_ protects the following state.
// We don't count mutex_ as the cache's internal state so semantically we
// don't mind mutex_ invoking the non-const actions.
mutable port::Mutex mutex_;
std::shared_ptr<SecondaryCache> secondary_cache_;
};
class LRUCache
#ifdef NDEBUG
final
#endif
: public ShardedCache {
public:
LRUCache(size_t capacity, int num_shard_bits, bool strict_capacity_limit,
double high_pri_pool_ratio,
std::shared_ptr<MemoryAllocator> memory_allocator = nullptr,
bool use_adaptive_mutex = kDefaultToAdaptiveMutex,
CacheMetadataChargePolicy metadata_charge_policy =
kDontChargeCacheMetadata,
const std::shared_ptr<SecondaryCache>& secondary_cache = nullptr);
virtual ~LRUCache();
virtual const char* Name() const override { return "LRUCache"; }
virtual CacheShard* GetShard(uint32_t shard) override;
virtual const CacheShard* GetShard(uint32_t shard) const override;
virtual void* Value(Handle* handle) override;
virtual size_t GetCharge(Handle* handle) const override;
virtual uint32_t GetHash(Handle* handle) const override;
virtual DeleterFn GetDeleter(Handle* handle) const override;
virtual void DisownData() override;
virtual void WaitAll(std::vector<Handle*>& handles) override;
// Retrieves number of elements in LRU, for unit test purpose only
size_t TEST_GetLRUSize();
// Retrieves high pri pool ratio
double GetHighPriPoolRatio();
private:
LRUCacheShard* shards_ = nullptr;
int num_shards_ = 0;
std::shared_ptr<SecondaryCache> secondary_cache_;
};
} // namespace ROCKSDB_NAMESPACE