rocksdb/env/posix_logger.h
Anand Ananthabhotla a27fce408e Auto recovery from out of space errors (#4164)
Summary:
This commit implements automatic recovery from a Status::NoSpace() error
during background operations such as write callback, flush and
compaction. The broad design is as follows -
1. Compaction errors are treated as soft errors and don't put the
database in read-only mode. A compaction is delayed until enough free
disk space is available to accomodate the compaction outputs, which is
estimated based on the input size. This means that users can continue to
write, and we rely on the WriteController to delay or stop writes if the
compaction debt becomes too high due to persistent low disk space
condition
2. Errors during write callback and flush are treated as hard errors,
i.e the database is put in read-only mode and goes back to read-write
only fater certain recovery actions are taken.
3. Both types of recovery rely on the SstFileManagerImpl to poll for
sufficient disk space. We assume that there is a 1-1 mapping between an
SFM and the underlying OS storage container. For cases where multiple
DBs are hosted on a single storage container, the user is expected to
allocate a single SFM instance and use the same one for all the DBs. If
no SFM is specified by the user, DBImpl::Open() will allocate one, but
this will be one per DB and each DB will recover independently. The
recovery implemented by SFM is as follows -
  a) On the first occurance of an out of space error during compaction,
subsequent
  compactions will be delayed until the disk free space check indicates
  enough available space. The required space is computed as the sum of
  input sizes.
  b) The free space check requirement will be removed once the amount of
  free space is greater than the size reserved by in progress
  compactions when the first error occured
  c) If the out of space error is a hard error, a background thread in
  SFM will poll for sufficient headroom before triggering the recovery
  of the database and putting it in write-only mode. The headroom is
  calculated as the sum of the write_buffer_size of all the DB instances
  associated with the SFM
4. EventListener callbacks will be called at the start and completion of
automatic recovery. Users can disable the auto recov ery in the start
callback, and later initiate it manually by calling DB::Resume()

Todo:
1. More extensive testing
2. Add disk full condition to db_stress (follow-on PR)
Pull Request resolved: https://github.com/facebook/rocksdb/pull/4164

Differential Revision: D9846378

Pulled By: anand1976

fbshipit-source-id: 80ea875dbd7f00205e19c82215ff6e37da10da4a
2018-09-15 13:43:04 -07:00

186 lines
5.3 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.
//
// Logger implementation that can be shared by all environments
// where enough posix functionality is available.
#pragma once
#include <algorithm>
#include <stdio.h>
#include "port/sys_time.h"
#include <time.h>
#include <fcntl.h>
#ifdef OS_LINUX
#ifndef FALLOC_FL_KEEP_SIZE
#include <linux/falloc.h>
#endif
#endif
#include <atomic>
#include "env/io_posix.h"
#include "monitoring/iostats_context_imp.h"
#include "rocksdb/env.h"
#include "util/sync_point.h"
namespace rocksdb {
class PosixLogger : public Logger {
private:
Status PosixCloseHelper() {
int ret;
ret = fclose(file_);
if (ret) {
return IOError("Unable to close log file", "", ret);
}
return Status::OK();
}
FILE* file_;
uint64_t (*gettid_)(); // Return the thread id for the current thread
std::atomic_size_t log_size_;
int fd_;
const static uint64_t flush_every_seconds_ = 5;
std::atomic_uint_fast64_t last_flush_micros_;
Env* env_;
std::atomic<bool> flush_pending_;
protected:
virtual Status CloseImpl() override { return PosixCloseHelper(); }
public:
PosixLogger(FILE* f, uint64_t (*gettid)(), Env* env,
const InfoLogLevel log_level = InfoLogLevel::ERROR_LEVEL)
: Logger(log_level),
file_(f),
gettid_(gettid),
log_size_(0),
fd_(fileno(f)),
last_flush_micros_(0),
env_(env),
flush_pending_(false) {}
virtual ~PosixLogger() {
if (!closed_) {
closed_ = true;
PosixCloseHelper();
}
}
virtual void Flush() override {
TEST_SYNC_POINT("PosixLogger::Flush:Begin1");
TEST_SYNC_POINT("PosixLogger::Flush:Begin2");
if (flush_pending_) {
flush_pending_ = false;
fflush(file_);
}
last_flush_micros_ = env_->NowMicros();
}
using Logger::Logv;
virtual void Logv(const char* format, va_list ap) override {
IOSTATS_TIMER_GUARD(logger_nanos);
const uint64_t thread_id = (*gettid_)();
// We try twice: the first time with a fixed-size stack allocated buffer,
// and the second time with a much larger dynamically allocated buffer.
char buffer[500];
for (int iter = 0; iter < 2; iter++) {
char* base;
int bufsize;
if (iter == 0) {
bufsize = sizeof(buffer);
base = buffer;
} else {
bufsize = 65536;
base = new char[bufsize];
}
char* p = base;
char* limit = base + bufsize;
struct timeval now_tv;
gettimeofday(&now_tv, nullptr);
const time_t seconds = now_tv.tv_sec;
struct tm t;
localtime_r(&seconds, &t);
p += snprintf(p, limit - p,
"%04d/%02d/%02d-%02d:%02d:%02d.%06d %llx ",
t.tm_year + 1900,
t.tm_mon + 1,
t.tm_mday,
t.tm_hour,
t.tm_min,
t.tm_sec,
static_cast<int>(now_tv.tv_usec),
static_cast<long long unsigned int>(thread_id));
// Print the message
if (p < limit) {
va_list backup_ap;
va_copy(backup_ap, ap);
p += vsnprintf(p, limit - p, format, backup_ap);
va_end(backup_ap);
}
// Truncate to available space if necessary
if (p >= limit) {
if (iter == 0) {
continue; // Try again with larger buffer
} else {
p = limit - 1;
}
}
// Add newline if necessary
if (p == base || p[-1] != '\n') {
*p++ = '\n';
}
assert(p <= limit);
const size_t write_size = p - base;
#ifdef ROCKSDB_FALLOCATE_PRESENT
const int kDebugLogChunkSize = 128 * 1024;
// If this write would cross a boundary of kDebugLogChunkSize
// space, pre-allocate more space to avoid overly large
// allocations from filesystem allocsize options.
const size_t log_size = log_size_;
const size_t last_allocation_chunk =
((kDebugLogChunkSize - 1 + log_size) / kDebugLogChunkSize);
const size_t desired_allocation_chunk =
((kDebugLogChunkSize - 1 + log_size + write_size) /
kDebugLogChunkSize);
if (last_allocation_chunk != desired_allocation_chunk) {
fallocate(
fd_, FALLOC_FL_KEEP_SIZE, 0,
static_cast<off_t>(desired_allocation_chunk * kDebugLogChunkSize));
}
#endif
size_t sz = fwrite(base, 1, write_size, file_);
flush_pending_ = true;
if (sz > 0) {
log_size_ += write_size;
}
uint64_t now_micros = static_cast<uint64_t>(now_tv.tv_sec) * 1000000 +
now_tv.tv_usec;
if (now_micros - last_flush_micros_ >= flush_every_seconds_ * 1000000) {
Flush();
}
if (base != buffer) {
delete[] base;
}
break;
}
}
size_t GetLogFileSize() const override { return log_size_; }
};
} // namespace rocksdb