2022-02-03 23:00:03 +00:00
|
|
|
// Copyright (c) 2011-present, Facebook, Inc. All rights reserved.
|
|
|
|
// This source code is licensed under both the GPLv2 (found in the
|
|
|
|
// COPYING file in the root directory) and Apache 2.0 License
|
|
|
|
// (found in the LICENSE.Apache file in the root directory).
|
|
|
|
|
|
|
|
#include "utilities/counted_fs.h"
|
|
|
|
|
|
|
|
#include <sstream>
|
|
|
|
|
|
|
|
#include "rocksdb/file_system.h"
|
|
|
|
#include "rocksdb/utilities/options_type.h"
|
|
|
|
|
|
|
|
namespace ROCKSDB_NAMESPACE {
|
|
|
|
namespace {
|
|
|
|
class CountedSequentialFile : public FSSequentialFileOwnerWrapper {
|
|
|
|
private:
|
|
|
|
CountedFileSystem* fs_;
|
|
|
|
|
|
|
|
public:
|
|
|
|
CountedSequentialFile(std::unique_ptr<FSSequentialFile>&& f,
|
|
|
|
CountedFileSystem* fs)
|
|
|
|
: FSSequentialFileOwnerWrapper(std::move(f)), fs_(fs) {}
|
|
|
|
|
|
|
|
~CountedSequentialFile() override { fs_->counters()->closes++; }
|
|
|
|
|
|
|
|
IOStatus Read(size_t n, const IOOptions& options, Slice* result,
|
|
|
|
char* scratch, IODebugContext* dbg) override {
|
|
|
|
IOStatus rv = target()->Read(n, options, result, scratch, dbg);
|
|
|
|
fs_->counters()->reads.RecordOp(rv, result->size());
|
|
|
|
return rv;
|
|
|
|
}
|
|
|
|
|
|
|
|
IOStatus PositionedRead(uint64_t offset, size_t n, const IOOptions& options,
|
|
|
|
Slice* result, char* scratch,
|
|
|
|
IODebugContext* dbg) override {
|
|
|
|
IOStatus rv =
|
|
|
|
target()->PositionedRead(offset, n, options, result, scratch, dbg);
|
|
|
|
fs_->counters()->reads.RecordOp(rv, result->size());
|
|
|
|
return rv;
|
|
|
|
}
|
|
|
|
};
|
|
|
|
|
|
|
|
class CountedRandomAccessFile : public FSRandomAccessFileOwnerWrapper {
|
|
|
|
private:
|
|
|
|
CountedFileSystem* fs_;
|
|
|
|
|
|
|
|
public:
|
|
|
|
CountedRandomAccessFile(std::unique_ptr<FSRandomAccessFile>&& f,
|
|
|
|
CountedFileSystem* fs)
|
|
|
|
: FSRandomAccessFileOwnerWrapper(std::move(f)), fs_(fs) {}
|
|
|
|
|
|
|
|
~CountedRandomAccessFile() override { fs_->counters()->closes++; }
|
|
|
|
|
|
|
|
IOStatus Read(uint64_t offset, size_t n, const IOOptions& options,
|
|
|
|
Slice* result, char* scratch,
|
|
|
|
IODebugContext* dbg) const override {
|
|
|
|
IOStatus rv = target()->Read(offset, n, options, result, scratch, dbg);
|
|
|
|
fs_->counters()->reads.RecordOp(rv, result->size());
|
|
|
|
return rv;
|
|
|
|
}
|
|
|
|
|
|
|
|
IOStatus MultiRead(FSReadRequest* reqs, size_t num_reqs,
|
|
|
|
const IOOptions& options, IODebugContext* dbg) override {
|
|
|
|
IOStatus rv = target()->MultiRead(reqs, num_reqs, options, dbg);
|
|
|
|
for (size_t r = 0; r < num_reqs; r++) {
|
|
|
|
fs_->counters()->reads.RecordOp(reqs[r].status, reqs[r].result.size());
|
|
|
|
}
|
|
|
|
return rv;
|
|
|
|
}
|
|
|
|
};
|
|
|
|
|
|
|
|
class CountedWritableFile : public FSWritableFileOwnerWrapper {
|
|
|
|
private:
|
|
|
|
CountedFileSystem* fs_;
|
|
|
|
|
|
|
|
public:
|
|
|
|
CountedWritableFile(std::unique_ptr<FSWritableFile>&& f,
|
|
|
|
CountedFileSystem* fs)
|
|
|
|
: FSWritableFileOwnerWrapper(std::move(f)), fs_(fs) {}
|
|
|
|
|
|
|
|
IOStatus Append(const Slice& data, const IOOptions& options,
|
|
|
|
IODebugContext* dbg) override {
|
|
|
|
IOStatus rv = target()->Append(data, options, dbg);
|
|
|
|
fs_->counters()->writes.RecordOp(rv, data.size());
|
|
|
|
return rv;
|
|
|
|
}
|
|
|
|
|
|
|
|
IOStatus Append(const Slice& data, const IOOptions& options,
|
|
|
|
const DataVerificationInfo& info,
|
|
|
|
IODebugContext* dbg) override {
|
|
|
|
IOStatus rv = target()->Append(data, options, info, dbg);
|
|
|
|
fs_->counters()->writes.RecordOp(rv, data.size());
|
|
|
|
return rv;
|
|
|
|
}
|
|
|
|
|
|
|
|
IOStatus PositionedAppend(const Slice& data, uint64_t offset,
|
|
|
|
const IOOptions& options,
|
|
|
|
IODebugContext* dbg) override {
|
|
|
|
IOStatus rv = target()->PositionedAppend(data, offset, options, dbg);
|
|
|
|
fs_->counters()->writes.RecordOp(rv, data.size());
|
|
|
|
return rv;
|
|
|
|
}
|
|
|
|
|
|
|
|
IOStatus PositionedAppend(const Slice& data, uint64_t offset,
|
|
|
|
const IOOptions& options,
|
|
|
|
const DataVerificationInfo& info,
|
|
|
|
IODebugContext* dbg) override {
|
|
|
|
IOStatus rv = target()->PositionedAppend(data, offset, options, info, dbg);
|
|
|
|
fs_->counters()->writes.RecordOp(rv, data.size());
|
|
|
|
return rv;
|
|
|
|
}
|
|
|
|
|
|
|
|
IOStatus Close(const IOOptions& options, IODebugContext* dbg) override {
|
|
|
|
IOStatus rv = target()->Close(options, dbg);
|
|
|
|
if (rv.ok()) {
|
|
|
|
fs_->counters()->closes++;
|
|
|
|
}
|
|
|
|
return rv;
|
|
|
|
}
|
|
|
|
|
|
|
|
IOStatus Flush(const IOOptions& options, IODebugContext* dbg) override {
|
|
|
|
IOStatus rv = target()->Flush(options, dbg);
|
|
|
|
if (rv.ok()) {
|
|
|
|
fs_->counters()->flushes++;
|
|
|
|
}
|
|
|
|
return rv;
|
|
|
|
}
|
|
|
|
|
|
|
|
IOStatus Sync(const IOOptions& options, IODebugContext* dbg) override {
|
|
|
|
IOStatus rv = target()->Sync(options, dbg);
|
|
|
|
if (rv.ok()) {
|
|
|
|
fs_->counters()->syncs++;
|
|
|
|
}
|
|
|
|
return rv;
|
|
|
|
}
|
|
|
|
|
|
|
|
IOStatus Fsync(const IOOptions& options, IODebugContext* dbg) override {
|
|
|
|
IOStatus rv = target()->Fsync(options, dbg);
|
|
|
|
if (rv.ok()) {
|
|
|
|
fs_->counters()->fsyncs++;
|
|
|
|
}
|
|
|
|
return rv;
|
|
|
|
}
|
|
|
|
|
|
|
|
IOStatus RangeSync(uint64_t offset, uint64_t nbytes, const IOOptions& options,
|
|
|
|
IODebugContext* dbg) override {
|
|
|
|
IOStatus rv = target()->RangeSync(offset, nbytes, options, dbg);
|
|
|
|
if (rv.ok()) {
|
|
|
|
fs_->counters()->syncs++;
|
|
|
|
}
|
|
|
|
return rv;
|
|
|
|
}
|
|
|
|
};
|
|
|
|
|
|
|
|
class CountedRandomRWFile : public FSRandomRWFileOwnerWrapper {
|
|
|
|
private:
|
|
|
|
mutable CountedFileSystem* fs_;
|
|
|
|
|
|
|
|
public:
|
|
|
|
CountedRandomRWFile(std::unique_ptr<FSRandomRWFile>&& f,
|
|
|
|
CountedFileSystem* fs)
|
|
|
|
: FSRandomRWFileOwnerWrapper(std::move(f)), fs_(fs) {}
|
|
|
|
IOStatus Write(uint64_t offset, const Slice& data, const IOOptions& options,
|
|
|
|
IODebugContext* dbg) override {
|
|
|
|
IOStatus rv = target()->Write(offset, data, options, dbg);
|
|
|
|
fs_->counters()->writes.RecordOp(rv, data.size());
|
|
|
|
return rv;
|
|
|
|
}
|
|
|
|
|
|
|
|
IOStatus Read(uint64_t offset, size_t n, const IOOptions& options,
|
|
|
|
Slice* result, char* scratch,
|
|
|
|
IODebugContext* dbg) const override {
|
|
|
|
IOStatus rv = target()->Read(offset, n, options, result, scratch, dbg);
|
|
|
|
fs_->counters()->reads.RecordOp(rv, result->size());
|
|
|
|
return rv;
|
|
|
|
}
|
|
|
|
|
|
|
|
IOStatus Flush(const IOOptions& options, IODebugContext* dbg) override {
|
|
|
|
IOStatus rv = target()->Flush(options, dbg);
|
|
|
|
if (rv.ok()) {
|
|
|
|
fs_->counters()->flushes++;
|
|
|
|
}
|
|
|
|
return rv;
|
|
|
|
}
|
|
|
|
|
|
|
|
IOStatus Sync(const IOOptions& options, IODebugContext* dbg) override {
|
|
|
|
IOStatus rv = target()->Sync(options, dbg);
|
|
|
|
if (rv.ok()) {
|
|
|
|
fs_->counters()->syncs++;
|
|
|
|
}
|
|
|
|
return rv;
|
|
|
|
}
|
|
|
|
|
|
|
|
IOStatus Fsync(const IOOptions& options, IODebugContext* dbg) override {
|
|
|
|
IOStatus rv = target()->Fsync(options, dbg);
|
|
|
|
if (rv.ok()) {
|
|
|
|
fs_->counters()->fsyncs++;
|
|
|
|
}
|
|
|
|
return rv;
|
|
|
|
}
|
|
|
|
|
|
|
|
IOStatus Close(const IOOptions& options, IODebugContext* dbg) override {
|
|
|
|
IOStatus rv = target()->Close(options, dbg);
|
|
|
|
if (rv.ok()) {
|
|
|
|
fs_->counters()->closes++;
|
|
|
|
}
|
|
|
|
return rv;
|
|
|
|
}
|
|
|
|
};
|
|
|
|
|
|
|
|
class CountedDirectory : public FSDirectoryWrapper {
|
|
|
|
private:
|
|
|
|
mutable CountedFileSystem* fs_;
|
Fix serious FSDirectory use-after-Close bug (missing fsync) (#10460)
Summary:
TL;DR: due to a recent change, if you drop a column family,
often that DB will no longer fsync after writing new SST files
to remaining or new column families, which could lead to data
loss on power loss.
More bug detail:
The intent of https://github.com/facebook/rocksdb/issues/10049 was to Close FSDirectory objects at
DB::Close time rather than waiting for DB object destruction.
Unfortunately, it also closes shared FSDirectory objects on
DropColumnFamily (& destroy remaining handles), which can lead
to use-after-Close on FSDirectory shared with remaining column
families. Those "uses" are only Fsyncs (or redundant Closes). In
the default Posix filesystem, an Fsync on a closed FSDirectory is a
quiet no-op. Consequently (under most configurations), if you drop
a column family, that DB will no longer fsync after writing new SST
files to column families sharing the same directory (true under most
configurations).
More fix detail:
Basically, this removes unnecessary Close ops on destroying
ColumnFamilyData. We let `shared_ptr` take care of calling the
destructor at the right time. If the intent was to require Close be
called before destroying FSDirectory, that was not made clear by the
author of FileSystem and was not at all enforced by https://github.com/facebook/rocksdb/issues/10049, which
could have added `assert(fd_ == -1)` to `~PosixDirectory()` but did
not. To keep this fix simple, we relax the unit test for https://github.com/facebook/rocksdb/issues/10049 to allow
timely destruction of FSDirectory to suffice as Close (in
CountedFileSystem). Added a TODO to revisit that.
Also in this PR:
* Added a TODO to share FSDirectory instances between DB and its column
families. (Already shared among column families.)
* Made DB::Close attempt to close all its open FSDirectory objects even
if there is a failure in closing one. Also code clean-up around this
logic.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10460
Test Plan:
add an assert to check for use-after-Close. With that
existing tests can detect the misuse. With fix, tests pass (except noted
relaxing of unit test for https://github.com/facebook/rocksdb/issues/10049)
Reviewed By: ajkr
Differential Revision: D38357922
Pulled By: pdillinger
fbshipit-source-id: d42079cadbedf0a969f03389bf586b3b4e1f9137
2022-08-02 17:54:32 +00:00
|
|
|
bool closed_ = false;
|
2022-02-03 23:00:03 +00:00
|
|
|
|
|
|
|
public:
|
|
|
|
CountedDirectory(std::unique_ptr<FSDirectory>&& f, CountedFileSystem* fs)
|
|
|
|
: FSDirectoryWrapper(std::move(f)), fs_(fs) {}
|
|
|
|
|
|
|
|
IOStatus Fsync(const IOOptions& options, IODebugContext* dbg) override {
|
|
|
|
IOStatus rv = FSDirectoryWrapper::Fsync(options, dbg);
|
|
|
|
if (rv.ok()) {
|
|
|
|
fs_->counters()->dsyncs++;
|
|
|
|
}
|
|
|
|
return rv;
|
|
|
|
}
|
|
|
|
|
Explicitly closing all directory file descriptors (#10049)
Summary:
Currently, the DB directory file descriptor is left open until the deconstruction process (`DB::Close()` does not close the file descriptor). To verify this, comment out the lines between `db_ = nullptr` and `db_->Close()` (line 512, 513, 514, 515 in ldb_cmd.cc) to leak the ``db_'' object, build `ldb` tool and run
```
strace --trace=open,openat,close ./ldb --db=$TEST_TMPDIR --ignore_unknown_options put K1 V1 --create_if_missing
```
There is one directory file descriptor that is not closed in the strace log.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10049
Test Plan: Add a new unit test DBBasicTest.DBCloseAllDirectoryFDs: Open a database with different WAL directory and three different data directories, and all directory file descriptors should be closed after calling Close(). Explicitly call Close() after a directory file descriptor is not used so that the counter of directory open and close should be equivalent.
Reviewed By: ajkr, hx235
Differential Revision: D36722135
Pulled By: littlepig2013
fbshipit-source-id: 07bdc2abc417c6b30997b9bbef1f79aa757b21ff
2022-06-02 01:03:34 +00:00
|
|
|
IOStatus Close(const IOOptions& options, IODebugContext* dbg) override {
|
|
|
|
IOStatus rv = FSDirectoryWrapper::Close(options, dbg);
|
|
|
|
if (rv.ok()) {
|
|
|
|
fs_->counters()->closes++;
|
|
|
|
fs_->counters()->dir_closes++;
|
Fix serious FSDirectory use-after-Close bug (missing fsync) (#10460)
Summary:
TL;DR: due to a recent change, if you drop a column family,
often that DB will no longer fsync after writing new SST files
to remaining or new column families, which could lead to data
loss on power loss.
More bug detail:
The intent of https://github.com/facebook/rocksdb/issues/10049 was to Close FSDirectory objects at
DB::Close time rather than waiting for DB object destruction.
Unfortunately, it also closes shared FSDirectory objects on
DropColumnFamily (& destroy remaining handles), which can lead
to use-after-Close on FSDirectory shared with remaining column
families. Those "uses" are only Fsyncs (or redundant Closes). In
the default Posix filesystem, an Fsync on a closed FSDirectory is a
quiet no-op. Consequently (under most configurations), if you drop
a column family, that DB will no longer fsync after writing new SST
files to column families sharing the same directory (true under most
configurations).
More fix detail:
Basically, this removes unnecessary Close ops on destroying
ColumnFamilyData. We let `shared_ptr` take care of calling the
destructor at the right time. If the intent was to require Close be
called before destroying FSDirectory, that was not made clear by the
author of FileSystem and was not at all enforced by https://github.com/facebook/rocksdb/issues/10049, which
could have added `assert(fd_ == -1)` to `~PosixDirectory()` but did
not. To keep this fix simple, we relax the unit test for https://github.com/facebook/rocksdb/issues/10049 to allow
timely destruction of FSDirectory to suffice as Close (in
CountedFileSystem). Added a TODO to revisit that.
Also in this PR:
* Added a TODO to share FSDirectory instances between DB and its column
families. (Already shared among column families.)
* Made DB::Close attempt to close all its open FSDirectory objects even
if there is a failure in closing one. Also code clean-up around this
logic.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10460
Test Plan:
add an assert to check for use-after-Close. With that
existing tests can detect the misuse. With fix, tests pass (except noted
relaxing of unit test for https://github.com/facebook/rocksdb/issues/10049)
Reviewed By: ajkr
Differential Revision: D38357922
Pulled By: pdillinger
fbshipit-source-id: d42079cadbedf0a969f03389bf586b3b4e1f9137
2022-08-02 17:54:32 +00:00
|
|
|
closed_ = true;
|
Explicitly closing all directory file descriptors (#10049)
Summary:
Currently, the DB directory file descriptor is left open until the deconstruction process (`DB::Close()` does not close the file descriptor). To verify this, comment out the lines between `db_ = nullptr` and `db_->Close()` (line 512, 513, 514, 515 in ldb_cmd.cc) to leak the ``db_'' object, build `ldb` tool and run
```
strace --trace=open,openat,close ./ldb --db=$TEST_TMPDIR --ignore_unknown_options put K1 V1 --create_if_missing
```
There is one directory file descriptor that is not closed in the strace log.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10049
Test Plan: Add a new unit test DBBasicTest.DBCloseAllDirectoryFDs: Open a database with different WAL directory and three different data directories, and all directory file descriptors should be closed after calling Close(). Explicitly call Close() after a directory file descriptor is not used so that the counter of directory open and close should be equivalent.
Reviewed By: ajkr, hx235
Differential Revision: D36722135
Pulled By: littlepig2013
fbshipit-source-id: 07bdc2abc417c6b30997b9bbef1f79aa757b21ff
2022-06-02 01:03:34 +00:00
|
|
|
}
|
|
|
|
return rv;
|
|
|
|
}
|
|
|
|
|
2022-02-03 23:00:03 +00:00
|
|
|
IOStatus FsyncWithDirOptions(const IOOptions& options, IODebugContext* dbg,
|
|
|
|
const DirFsyncOptions& dir_options) override {
|
|
|
|
IOStatus rv =
|
|
|
|
FSDirectoryWrapper::FsyncWithDirOptions(options, dbg, dir_options);
|
|
|
|
if (rv.ok()) {
|
|
|
|
fs_->counters()->dsyncs++;
|
|
|
|
}
|
|
|
|
return rv;
|
|
|
|
}
|
Fix serious FSDirectory use-after-Close bug (missing fsync) (#10460)
Summary:
TL;DR: due to a recent change, if you drop a column family,
often that DB will no longer fsync after writing new SST files
to remaining or new column families, which could lead to data
loss on power loss.
More bug detail:
The intent of https://github.com/facebook/rocksdb/issues/10049 was to Close FSDirectory objects at
DB::Close time rather than waiting for DB object destruction.
Unfortunately, it also closes shared FSDirectory objects on
DropColumnFamily (& destroy remaining handles), which can lead
to use-after-Close on FSDirectory shared with remaining column
families. Those "uses" are only Fsyncs (or redundant Closes). In
the default Posix filesystem, an Fsync on a closed FSDirectory is a
quiet no-op. Consequently (under most configurations), if you drop
a column family, that DB will no longer fsync after writing new SST
files to column families sharing the same directory (true under most
configurations).
More fix detail:
Basically, this removes unnecessary Close ops on destroying
ColumnFamilyData. We let `shared_ptr` take care of calling the
destructor at the right time. If the intent was to require Close be
called before destroying FSDirectory, that was not made clear by the
author of FileSystem and was not at all enforced by https://github.com/facebook/rocksdb/issues/10049, which
could have added `assert(fd_ == -1)` to `~PosixDirectory()` but did
not. To keep this fix simple, we relax the unit test for https://github.com/facebook/rocksdb/issues/10049 to allow
timely destruction of FSDirectory to suffice as Close (in
CountedFileSystem). Added a TODO to revisit that.
Also in this PR:
* Added a TODO to share FSDirectory instances between DB and its column
families. (Already shared among column families.)
* Made DB::Close attempt to close all its open FSDirectory objects even
if there is a failure in closing one. Also code clean-up around this
logic.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10460
Test Plan:
add an assert to check for use-after-Close. With that
existing tests can detect the misuse. With fix, tests pass (except noted
relaxing of unit test for https://github.com/facebook/rocksdb/issues/10049)
Reviewed By: ajkr
Differential Revision: D38357922
Pulled By: pdillinger
fbshipit-source-id: d42079cadbedf0a969f03389bf586b3b4e1f9137
2022-08-02 17:54:32 +00:00
|
|
|
|
|
|
|
~CountedDirectory() {
|
|
|
|
if (!closed_) {
|
|
|
|
// TODO: fix DB+CF code to use explicit Close, not rely on destructor
|
|
|
|
fs_->counters()->closes++;
|
|
|
|
fs_->counters()->dir_closes++;
|
|
|
|
}
|
|
|
|
}
|
2022-02-03 23:00:03 +00:00
|
|
|
};
|
|
|
|
} // anonymous namespace
|
|
|
|
|
|
|
|
std::string FileOpCounters::PrintCounters() const {
|
|
|
|
std::stringstream ss;
|
|
|
|
ss << "Num files opened: " << opens.load(std::memory_order_relaxed)
|
|
|
|
<< std::endl;
|
|
|
|
ss << "Num files deleted: " << deletes.load(std::memory_order_relaxed)
|
|
|
|
<< std::endl;
|
|
|
|
ss << "Num files renamed: " << renames.load(std::memory_order_relaxed)
|
|
|
|
<< std::endl;
|
|
|
|
ss << "Num Flush(): " << flushes.load(std::memory_order_relaxed) << std::endl;
|
|
|
|
ss << "Num Sync(): " << syncs.load(std::memory_order_relaxed) << std::endl;
|
|
|
|
ss << "Num Fsync(): " << fsyncs.load(std::memory_order_relaxed) << std::endl;
|
|
|
|
ss << "Num Dir Fsync(): " << dsyncs.load(std::memory_order_relaxed)
|
|
|
|
<< std::endl;
|
|
|
|
ss << "Num Close(): " << closes.load(std::memory_order_relaxed) << std::endl;
|
Explicitly closing all directory file descriptors (#10049)
Summary:
Currently, the DB directory file descriptor is left open until the deconstruction process (`DB::Close()` does not close the file descriptor). To verify this, comment out the lines between `db_ = nullptr` and `db_->Close()` (line 512, 513, 514, 515 in ldb_cmd.cc) to leak the ``db_'' object, build `ldb` tool and run
```
strace --trace=open,openat,close ./ldb --db=$TEST_TMPDIR --ignore_unknown_options put K1 V1 --create_if_missing
```
There is one directory file descriptor that is not closed in the strace log.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10049
Test Plan: Add a new unit test DBBasicTest.DBCloseAllDirectoryFDs: Open a database with different WAL directory and three different data directories, and all directory file descriptors should be closed after calling Close(). Explicitly call Close() after a directory file descriptor is not used so that the counter of directory open and close should be equivalent.
Reviewed By: ajkr, hx235
Differential Revision: D36722135
Pulled By: littlepig2013
fbshipit-source-id: 07bdc2abc417c6b30997b9bbef1f79aa757b21ff
2022-06-02 01:03:34 +00:00
|
|
|
ss << "Num Dir Open(): " << dir_opens.load(std::memory_order_relaxed)
|
|
|
|
<< std::endl;
|
|
|
|
ss << "Num Dir Close(): " << dir_closes.load(std::memory_order_relaxed)
|
|
|
|
<< std::endl;
|
2022-02-03 23:00:03 +00:00
|
|
|
ss << "Num Read(): " << reads.ops.load(std::memory_order_relaxed)
|
|
|
|
<< std::endl;
|
|
|
|
ss << "Num Append(): " << writes.ops.load(std::memory_order_relaxed)
|
|
|
|
<< std::endl;
|
|
|
|
ss << "Num bytes read: " << reads.bytes.load(std::memory_order_relaxed)
|
|
|
|
<< std::endl;
|
|
|
|
ss << "Num bytes written: " << writes.bytes.load(std::memory_order_relaxed)
|
|
|
|
<< std::endl;
|
|
|
|
return ss.str();
|
|
|
|
}
|
|
|
|
|
|
|
|
CountedFileSystem::CountedFileSystem(const std::shared_ptr<FileSystem>& base)
|
|
|
|
: FileSystemWrapper(base) {}
|
|
|
|
|
|
|
|
IOStatus CountedFileSystem::NewSequentialFile(
|
|
|
|
const std::string& f, const FileOptions& options,
|
|
|
|
std::unique_ptr<FSSequentialFile>* r, IODebugContext* dbg) {
|
|
|
|
std::unique_ptr<FSSequentialFile> base;
|
|
|
|
IOStatus s = target()->NewSequentialFile(f, options, &base, dbg);
|
|
|
|
if (s.ok()) {
|
|
|
|
counters_.opens++;
|
|
|
|
r->reset(new CountedSequentialFile(std::move(base), this));
|
|
|
|
}
|
|
|
|
return s;
|
|
|
|
}
|
|
|
|
|
|
|
|
IOStatus CountedFileSystem::NewRandomAccessFile(
|
|
|
|
const std::string& f, const FileOptions& options,
|
|
|
|
std::unique_ptr<FSRandomAccessFile>* r, IODebugContext* dbg) {
|
|
|
|
std::unique_ptr<FSRandomAccessFile> base;
|
|
|
|
IOStatus s = target()->NewRandomAccessFile(f, options, &base, dbg);
|
|
|
|
if (s.ok()) {
|
|
|
|
counters_.opens++;
|
|
|
|
r->reset(new CountedRandomAccessFile(std::move(base), this));
|
|
|
|
}
|
|
|
|
return s;
|
|
|
|
}
|
|
|
|
|
|
|
|
IOStatus CountedFileSystem::NewWritableFile(const std::string& f,
|
|
|
|
const FileOptions& options,
|
|
|
|
std::unique_ptr<FSWritableFile>* r,
|
|
|
|
IODebugContext* dbg) {
|
|
|
|
std::unique_ptr<FSWritableFile> base;
|
|
|
|
IOStatus s = target()->NewWritableFile(f, options, &base, dbg);
|
|
|
|
if (s.ok()) {
|
|
|
|
counters_.opens++;
|
|
|
|
r->reset(new CountedWritableFile(std::move(base), this));
|
|
|
|
}
|
|
|
|
return s;
|
|
|
|
}
|
|
|
|
|
|
|
|
IOStatus CountedFileSystem::ReopenWritableFile(
|
|
|
|
const std::string& fname, const FileOptions& options,
|
|
|
|
std::unique_ptr<FSWritableFile>* result, IODebugContext* dbg) {
|
|
|
|
std::unique_ptr<FSWritableFile> base;
|
|
|
|
IOStatus s = target()->ReopenWritableFile(fname, options, &base, dbg);
|
|
|
|
if (s.ok()) {
|
|
|
|
counters_.opens++;
|
|
|
|
result->reset(new CountedWritableFile(std::move(base), this));
|
|
|
|
}
|
|
|
|
return s;
|
|
|
|
}
|
|
|
|
|
|
|
|
IOStatus CountedFileSystem::ReuseWritableFile(
|
|
|
|
const std::string& fname, const std::string& old_fname,
|
|
|
|
const FileOptions& options, std::unique_ptr<FSWritableFile>* result,
|
|
|
|
IODebugContext* dbg) {
|
|
|
|
std::unique_ptr<FSWritableFile> base;
|
|
|
|
IOStatus s =
|
|
|
|
target()->ReuseWritableFile(fname, old_fname, options, &base, dbg);
|
|
|
|
if (s.ok()) {
|
|
|
|
counters_.opens++;
|
|
|
|
result->reset(new CountedWritableFile(std::move(base), this));
|
|
|
|
}
|
|
|
|
return s;
|
|
|
|
}
|
|
|
|
|
|
|
|
IOStatus CountedFileSystem::NewRandomRWFile(
|
|
|
|
const std::string& name, const FileOptions& options,
|
|
|
|
std::unique_ptr<FSRandomRWFile>* result, IODebugContext* dbg) {
|
|
|
|
std::unique_ptr<FSRandomRWFile> base;
|
|
|
|
IOStatus s = target()->NewRandomRWFile(name, options, &base, dbg);
|
|
|
|
if (s.ok()) {
|
|
|
|
counters_.opens++;
|
|
|
|
result->reset(new CountedRandomRWFile(std::move(base), this));
|
|
|
|
}
|
|
|
|
return s;
|
|
|
|
}
|
|
|
|
|
|
|
|
IOStatus CountedFileSystem::NewDirectory(const std::string& name,
|
|
|
|
const IOOptions& options,
|
|
|
|
std::unique_ptr<FSDirectory>* result,
|
|
|
|
IODebugContext* dbg) {
|
|
|
|
std::unique_ptr<FSDirectory> base;
|
|
|
|
IOStatus s = target()->NewDirectory(name, options, &base, dbg);
|
|
|
|
if (s.ok()) {
|
|
|
|
counters_.opens++;
|
Explicitly closing all directory file descriptors (#10049)
Summary:
Currently, the DB directory file descriptor is left open until the deconstruction process (`DB::Close()` does not close the file descriptor). To verify this, comment out the lines between `db_ = nullptr` and `db_->Close()` (line 512, 513, 514, 515 in ldb_cmd.cc) to leak the ``db_'' object, build `ldb` tool and run
```
strace --trace=open,openat,close ./ldb --db=$TEST_TMPDIR --ignore_unknown_options put K1 V1 --create_if_missing
```
There is one directory file descriptor that is not closed in the strace log.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/10049
Test Plan: Add a new unit test DBBasicTest.DBCloseAllDirectoryFDs: Open a database with different WAL directory and three different data directories, and all directory file descriptors should be closed after calling Close(). Explicitly call Close() after a directory file descriptor is not used so that the counter of directory open and close should be equivalent.
Reviewed By: ajkr, hx235
Differential Revision: D36722135
Pulled By: littlepig2013
fbshipit-source-id: 07bdc2abc417c6b30997b9bbef1f79aa757b21ff
2022-06-02 01:03:34 +00:00
|
|
|
counters_.dir_opens++;
|
2022-02-03 23:00:03 +00:00
|
|
|
result->reset(new CountedDirectory(std::move(base), this));
|
|
|
|
}
|
|
|
|
return s;
|
|
|
|
}
|
|
|
|
|
|
|
|
} // namespace ROCKSDB_NAMESPACE
|