mirror of
https://github.com/facebook/rocksdb.git
synced 2024-11-30 22:41:48 +00:00
ac24f152a1
Summary: This is setting up for a fix to a data race in SetOptions on BlockBasedTableOptions (BBTO), https://github.com/facebook/rocksdb/issues/10079 The race will be fixed by replacing `table_factory` with a modified copy whenever we want to modify a BBTO field. An argument could be made that this change creates more entaglement between features (e.g. BlobSource <-> MutableCFOptions), rather than (conceptually) minimizing the dependencies of each feature, but * Most of these things already depended on ImmutableOptions * Historically there has been a lot of plumbing (and possible small CPU overhead) involved in adding features that need to reach a lot of places, like `block_protection_bytes_per_key`. Keeping those wrapped up in options simplifies that. * SuperVersion management generally takes care of lifetime management of MutableCFOptions, so is not that difficult. (Crash test agrees so far.) There are some FIXME places where it is known to be unsafe to replace `block_cache` unless/until we handle shared_ptr tracking properly. HOWEVER, replacing `block_cache` is generally dubious, at least while existing users of the old block cache (e.g. table readers) can continue indefinitely. The change to cf_options.cc is essentially just moving code (not changing). I'm not concerned about the performance of copying another shared_ptr with MutableCFOptions, but I left a note about considering an improvement if more shared_ptr are added to it. Pull Request resolved: https://github.com/facebook/rocksdb/pull/13077 Test Plan: existing tests, crash test. Unit test DBOptionsTest.GetLatestCFOptions updated with some temporary logic. MemoryTest required some refactoring (simplification) for the change. Reviewed By: cbi42 Differential Revision: D64546903 Pulled By: pdillinger fbshipit-source-id: 69ae97ce5cf4c01b58edc4c5d4687eb1e5bf5855
166 lines
6.6 KiB
C++
166 lines
6.6 KiB
C++
// Copyright (c) Meta Platforms, Inc. and affiliates.
|
|
// 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).
|
|
|
|
#pragma once
|
|
|
|
#include <cinttypes>
|
|
#include <memory>
|
|
|
|
#include "cache/cache_key.h"
|
|
#include "cache/typed_cache.h"
|
|
#include "db/blob/blob_contents.h"
|
|
#include "db/blob/blob_file_cache.h"
|
|
#include "db/blob/blob_read_request.h"
|
|
#include "rocksdb/cache.h"
|
|
#include "rocksdb/rocksdb_namespace.h"
|
|
#include "table/block_based/cachable_entry.h"
|
|
#include "util/autovector.h"
|
|
|
|
namespace ROCKSDB_NAMESPACE {
|
|
|
|
struct ImmutableOptions;
|
|
struct MutableCFOptions;
|
|
class Status;
|
|
class FilePrefetchBuffer;
|
|
class Slice;
|
|
|
|
// BlobSource is a class that provides universal access to blobs, regardless of
|
|
// whether they are in the blob cache, secondary cache, or (remote) storage.
|
|
// Depending on user settings, it always fetch blobs from multi-tier cache and
|
|
// storage with minimal cost.
|
|
class BlobSource {
|
|
public:
|
|
// NOTE: db_id, db_session_id, and blob_file_cache are saved by reference or
|
|
// pointer.
|
|
BlobSource(const ImmutableOptions& immutable_options,
|
|
const MutableCFOptions& mutable_cf_options,
|
|
const std::string& db_id, const std::string& db_session_id,
|
|
BlobFileCache* blob_file_cache);
|
|
|
|
BlobSource(const BlobSource&) = delete;
|
|
BlobSource& operator=(const BlobSource&) = delete;
|
|
|
|
~BlobSource();
|
|
|
|
// Read a blob from the underlying cache or one blob file.
|
|
//
|
|
// If successful, returns ok and sets "*value" to the newly retrieved
|
|
// uncompressed blob. If there was an error while fetching the blob, sets
|
|
// "*value" to empty and returns a non-ok status.
|
|
//
|
|
// Note: For consistency, whether the blob is found in the cache or on disk,
|
|
// sets "*bytes_read" to the size of on-disk (possibly compressed) blob
|
|
// record.
|
|
Status GetBlob(const ReadOptions& read_options, const Slice& user_key,
|
|
uint64_t file_number, uint64_t offset, uint64_t file_size,
|
|
uint64_t value_size, CompressionType compression_type,
|
|
FilePrefetchBuffer* prefetch_buffer, PinnableSlice* value,
|
|
uint64_t* bytes_read);
|
|
|
|
// Read multiple blobs from the underlying cache or blob file(s).
|
|
//
|
|
// If successful, returns ok and sets "result" in the elements of "blob_reqs"
|
|
// to the newly retrieved uncompressed blobs. If there was an error while
|
|
// fetching one of blobs, sets its "result" to empty and sets its
|
|
// corresponding "status" to a non-ok status.
|
|
//
|
|
// Note:
|
|
// - The main difference between this function and MultiGetBlobFromOneFile is
|
|
// that this function can read multiple blobs from multiple blob files.
|
|
//
|
|
// - For consistency, whether the blob is found in the cache or on disk, sets
|
|
// "*bytes_read" to the total size of on-disk (possibly compressed) blob
|
|
// records.
|
|
void MultiGetBlob(const ReadOptions& read_options,
|
|
autovector<BlobFileReadRequests>& blob_reqs,
|
|
uint64_t* bytes_read);
|
|
|
|
// Read multiple blobs from the underlying cache or one blob file.
|
|
//
|
|
// If successful, returns ok and sets "result" in the elements of "blob_reqs"
|
|
// to the newly retrieved uncompressed blobs. If there was an error while
|
|
// fetching one of blobs, sets its "result" to empty and sets its
|
|
// corresponding "status" to a non-ok status.
|
|
//
|
|
// Note:
|
|
// - The main difference between this function and MultiGetBlob is that this
|
|
// function is only used for the case where the demanded blobs are stored in
|
|
// one blob file. MultiGetBlob will call this function multiple times if the
|
|
// demanded blobs are stored in multiple blob files.
|
|
//
|
|
// - For consistency, whether the blob is found in the cache or on disk, sets
|
|
// "*bytes_read" to the total size of on-disk (possibly compressed) blob
|
|
// records.
|
|
void MultiGetBlobFromOneFile(const ReadOptions& read_options,
|
|
uint64_t file_number, uint64_t file_size,
|
|
autovector<BlobReadRequest>& blob_reqs,
|
|
uint64_t* bytes_read);
|
|
|
|
inline Status GetBlobFileReader(
|
|
const ReadOptions& read_options, uint64_t blob_file_number,
|
|
CacheHandleGuard<BlobFileReader>* blob_file_reader) {
|
|
return blob_file_cache_->GetBlobFileReader(read_options, blob_file_number,
|
|
blob_file_reader);
|
|
}
|
|
|
|
inline Cache* GetBlobCache() const { return blob_cache_.get(); }
|
|
|
|
bool TEST_BlobInCache(uint64_t file_number, uint64_t file_size,
|
|
uint64_t offset, size_t* charge = nullptr) const;
|
|
|
|
// For TypedSharedCacheInterface
|
|
void Create(BlobContents** out, const char* buf, size_t size,
|
|
MemoryAllocator* alloc);
|
|
|
|
using SharedCacheInterface =
|
|
FullTypedSharedCacheInterface<BlobContents, BlobContentsCreator>;
|
|
using TypedHandle = SharedCacheInterface::TypedHandle;
|
|
|
|
private:
|
|
Status GetBlobFromCache(const Slice& cache_key,
|
|
CacheHandleGuard<BlobContents>* cached_blob) const;
|
|
|
|
Status PutBlobIntoCache(const Slice& cache_key,
|
|
std::unique_ptr<BlobContents>* blob,
|
|
CacheHandleGuard<BlobContents>* cached_blob) const;
|
|
|
|
static void PinCachedBlob(CacheHandleGuard<BlobContents>* cached_blob,
|
|
PinnableSlice* value);
|
|
|
|
static void PinOwnedBlob(std::unique_ptr<BlobContents>* owned_blob,
|
|
PinnableSlice* value);
|
|
|
|
TypedHandle* GetEntryFromCache(const Slice& key) const;
|
|
|
|
Status InsertEntryIntoCache(const Slice& key, BlobContents* value,
|
|
TypedHandle** cache_handle,
|
|
Cache::Priority priority) const;
|
|
|
|
inline CacheKey GetCacheKey(uint64_t file_number, uint64_t /*file_size*/,
|
|
uint64_t offset) const {
|
|
OffsetableCacheKey base_cache_key(db_id_, db_session_id_, file_number);
|
|
return base_cache_key.WithOffset(offset);
|
|
}
|
|
|
|
const std::string& db_id_;
|
|
const std::string& db_session_id_;
|
|
|
|
Statistics* statistics_;
|
|
|
|
// A cache to store blob file reader.
|
|
BlobFileCache* blob_file_cache_;
|
|
|
|
// A cache to store uncompressed blobs.
|
|
mutable SharedCacheInterface blob_cache_;
|
|
|
|
// The control option of how the cache tiers will be used. Currently rocksdb
|
|
// support block/blob cache (volatile tier) and secondary cache (this tier
|
|
// isn't strictly speaking a non-volatile tier since the compressed cache in
|
|
// this tier is in volatile memory).
|
|
const CacheTier lowest_used_cache_tier_;
|
|
};
|
|
|
|
} // namespace ROCKSDB_NAMESPACE
|