rocksdb/db/event_helpers.cc

358 lines
12 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).
#include "db/event_helpers.h"
#include "rocksdb/convenience.h"
#include "rocksdb/listener.h"
#include "rocksdb/utilities/customizable_util.h"
namespace ROCKSDB_NAMESPACE {
#ifndef ROCKSDB_LITE
Status EventListener::CreateFromString(const ConfigOptions& config_options,
const std::string& id,
std::shared_ptr<EventListener>* result) {
return LoadSharedObject<EventListener>(config_options, id, nullptr, result);
}
#endif // ROCKSDB_LITE
namespace {
template <class T>
inline T SafeDivide(T a, T b) {
return b == 0 ? 0 : a / b;
}
} // namespace
void EventHelpers::AppendCurrentTime(JSONWriter* jwriter) {
*jwriter << "time_micros"
<< std::chrono::duration_cast<std::chrono::microseconds>(
std::chrono::system_clock::now().time_since_epoch())
.count();
}
#ifndef ROCKSDB_LITE
void EventHelpers::NotifyTableFileCreationStarted(
const std::vector<std::shared_ptr<EventListener>>& listeners,
const std::string& db_name, const std::string& cf_name,
const std::string& file_path, int job_id, TableFileCreationReason reason) {
if (listeners.empty()) {
return;
}
TableFileCreationBriefInfo info;
info.db_name = db_name;
info.cf_name = cf_name;
info.file_path = file_path;
info.job_id = job_id;
info.reason = reason;
for (auto& listener : listeners) {
listener->OnTableFileCreationStarted(info);
}
}
#endif // !ROCKSDB_LITE
void EventHelpers::NotifyOnBackgroundError(
const std::vector<std::shared_ptr<EventListener>>& listeners,
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 22:36:19 +02:00
BackgroundErrorReason reason, Status* bg_error, InstrumentedMutex* db_mutex,
bool* auto_recovery) {
#ifndef ROCKSDB_LITE
if (listeners.empty()) {
return;
}
db_mutex->AssertHeld();
// release lock while notifying events
db_mutex->Unlock();
for (auto& listener : listeners) {
listener->OnBackgroundError(reason, bg_error);
bg_error->PermitUncheckedError();
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 22:36:19 +02:00
if (*auto_recovery) {
listener->OnErrorRecoveryBegin(reason, *bg_error, auto_recovery);
}
}
db_mutex->Lock();
#else
(void)listeners;
(void)reason;
(void)bg_error;
(void)db_mutex;
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 22:36:19 +02:00
(void)auto_recovery;
#endif // ROCKSDB_LITE
}
void EventHelpers::LogAndNotifyTableFileCreationFinished(
EventLogger* event_logger,
const std::vector<std::shared_ptr<EventListener>>& listeners,
const std::string& db_name, const std::string& cf_name,
const std::string& file_path, int job_id, const FileDescriptor& fd,
uint64_t oldest_blob_file_number, const TableProperties& table_properties,
TableFileCreationReason reason, const Status& s,
const std::string& file_checksum,
const std::string& file_checksum_func_name) {
if (s.ok() && event_logger) {
JSONWriter jwriter;
AppendCurrentTime(&jwriter);
jwriter << "cf_name" << cf_name << "job" << job_id << "event"
<< "table_file_creation"
<< "file_number" << fd.GetNumber() << "file_size"
<< fd.GetFileSize() << "file_checksum" << file_checksum
<< "file_checksum_func_name" << file_checksum_func_name;
// table_properties
{
jwriter << "table_properties";
jwriter.StartObject();
// basic properties:
jwriter << "data_size" << table_properties.data_size << "index_size"
<< table_properties.index_size << "index_partitions"
<< table_properties.index_partitions << "top_level_index_size"
<< table_properties.top_level_index_size
<< "index_key_is_user_key"
<< table_properties.index_key_is_user_key
<< "index_value_is_delta_encoded"
<< table_properties.index_value_is_delta_encoded << "filter_size"
<< table_properties.filter_size << "raw_key_size"
<< table_properties.raw_key_size << "raw_average_key_size"
<< SafeDivide(table_properties.raw_key_size,
table_properties.num_entries)
<< "raw_value_size" << table_properties.raw_value_size
<< "raw_average_value_size"
<< SafeDivide(table_properties.raw_value_size,
table_properties.num_entries)
<< "num_data_blocks" << table_properties.num_data_blocks
<< "num_entries" << table_properties.num_entries
<< "num_filter_entries" << table_properties.num_filter_entries
<< "num_deletions" << table_properties.num_deletions
<< "num_merge_operands" << table_properties.num_merge_operands
Fix wrong info log printing for num_range_deletions (#5617) Summary: num_range_deletions printing is wrong in this log line: 2019/07/18-12:59:15.309271 7f869f9ff700 EVENT_LOG_v1 {"time_micros": 1563479955309228, "cf_name": "5", "job": 955, "event": "table_file_creation", "file_number": 34579, "file_size": 2239842, "table_properties": {"data_size": 1988792, "index_size": 3067, "index_partitions": 0, "top_level_index_size": 0, "index_key_is_user_key": 0, "index_value_is_delta_encoded": 1, "filter_size": 170821, "raw_key_size": 1951792, "raw_average_key_size": 16, "raw_value_size": 1731720, "raw_average_value_size": 14, "num_data_blocks": 199, "num_entries": 121987, "num_deletions": 15184, "num_merge_operands": 86512, "num_range_deletions": 86512, "format_version": 0, "fixed_key_len": 0, "filter_policy": "rocksdb.BuiltinBloomFilter", "column_family_name": "5", "column_family_id": 5, "comparator": "leveldb.BytewiseComparator", "merge_operator": "PutOperator", "prefix_extractor_name": "rocksdb.FixedPrefix.7", "property_collectors": "[]", "compression": "ZSTD", "compression_options": "window_bits=-14; level=32767; strategy=0; max_dict_bytes=0; zstd_max_train_bytes=0; enabled=0; ", "creation_time": 1563479951, "oldest_key_time": 0, "file_creation_time": 1563479954}} It actually prints "num_merge_operands" number. Fix it. Pull Request resolved: https://github.com/facebook/rocksdb/pull/5617 Test Plan: Just build. Differential Revision: D16453110 fbshipit-source-id: fc1024b3cd5650312ed47a1379f0d2cf8b2d8a8f
2019-07-24 04:34:56 +02:00
<< "num_range_deletions" << table_properties.num_range_deletions
<< "format_version" << table_properties.format_version
<< "fixed_key_len" << table_properties.fixed_key_len
<< "filter_policy" << table_properties.filter_policy_name
<< "column_family_name" << table_properties.column_family_name
<< "column_family_id" << table_properties.column_family_id
<< "comparator" << table_properties.comparator_name
<< "merge_operator" << table_properties.merge_operator_name
<< "prefix_extractor_name"
<< table_properties.prefix_extractor_name << "property_collectors"
<< table_properties.property_collectors_names << "compression"
<< table_properties.compression_name << "compression_options"
<< table_properties.compression_options << "creation_time"
<< table_properties.creation_time << "oldest_key_time"
<< table_properties.oldest_key_time << "file_creation_time"
<< table_properties.file_creation_time
<< "slow_compression_estimated_data_size"
<< table_properties.slow_compression_estimated_data_size
<< "fast_compression_estimated_data_size"
<< table_properties.fast_compression_estimated_data_size
<< "db_id" << table_properties.db_id << "db_session_id"
Embed original file number in SST table properties (#8686) Summary: I very recently realized that with https://github.com/facebook/rocksdb/issues/8669 we cannot later add file numbers to external SST files (so that more can share db session ids for better uniqueness properties), because of forward compatibility. We would have a version of RocksDB that assumes session IDs are unique on external SST files and therefore can't really break that invariant in future files. This change adds a table property for "orig_file_number" which is populated by normal SST files and also external SST files generated by SstFileWriter. SstFileWriter now keeps a db_session_id for life of the object and increments its own file numbers for embedding in table properties. (They are arguably "fake" file numbers because these numbers and not embedded in the file name.) While updating block_based_table_builder, I removed several unnecessary fields from Rep, because following the pattern would have created another unnecessary field. This change also updates block_based_table_reader to use this new property when available, which means that for newer SST files, we can determine the stable/original <db_session_id,file_number> unique identifier using just the file contents, not the file name. (It's a bit complicated; detailed comments in block_based_table_reader.) Also added DB host id to properties listing by sst_dump, which could be useful in debugging. Pull Request resolved: https://github.com/facebook/rocksdb/pull/8686 Test Plan: majorly overhauled StableCacheKeys test for this change Reviewed By: zhichao-cao Differential Revision: D30457742 Pulled By: pdillinger fbshipit-source-id: 2e5ae7dddeb94fb9d8eac8a928486aed8b8cd445
2021-08-21 05:39:52 +02:00
<< table_properties.db_session_id << "orig_file_number"
<< table_properties.orig_file_number;
// user collected properties
for (const auto& prop : table_properties.readable_properties) {
jwriter << prop.first << prop.second;
}
jwriter.EndObject();
}
if (oldest_blob_file_number != kInvalidBlobFileNumber) {
jwriter << "oldest_blob_file_number" << oldest_blob_file_number;
}
jwriter.EndObject();
event_logger->Log(jwriter);
}
#ifndef ROCKSDB_LITE
if (listeners.empty()) {
return;
}
TableFileCreationInfo info;
info.db_name = db_name;
info.cf_name = cf_name;
info.file_path = file_path;
info.file_size = fd.file_size;
info.job_id = job_id;
info.table_properties = table_properties;
info.reason = reason;
info.status = s;
info.file_checksum = file_checksum;
info.file_checksum_func_name = file_checksum_func_name;
for (auto& listener : listeners) {
listener->OnTableFileCreated(info);
}
info.status.PermitUncheckedError();
#else
(void)listeners;
(void)db_name;
(void)cf_name;
(void)file_path;
(void)reason;
#endif // !ROCKSDB_LITE
}
void EventHelpers::LogAndNotifyTableFileDeletion(
EventLogger* event_logger, int job_id, uint64_t file_number,
const std::string& file_path, const Status& status,
const std::string& dbname,
const std::vector<std::shared_ptr<EventListener>>& listeners) {
JSONWriter jwriter;
AppendCurrentTime(&jwriter);
jwriter << "job" << job_id << "event"
<< "table_file_deletion"
<< "file_number" << file_number;
if (!status.ok()) {
jwriter << "status" << status.ToString();
}
jwriter.EndObject();
event_logger->Log(jwriter);
#ifndef ROCKSDB_LITE
if (listeners.empty()) {
return;
}
TableFileDeletionInfo info;
info.db_name = dbname;
info.job_id = job_id;
info.file_path = file_path;
info.status = status;
for (auto& listener : listeners) {
listener->OnTableFileDeleted(info);
}
info.status.PermitUncheckedError();
#else
(void)file_path;
(void)dbname;
(void)listeners;
#endif // !ROCKSDB_LITE
}
void EventHelpers::NotifyOnErrorRecoveryEnd(
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 22:36:19 +02:00
const std::vector<std::shared_ptr<EventListener>>& listeners,
const Status& old_bg_error, const Status& new_bg_error,
InstrumentedMutex* db_mutex) {
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 22:36:19 +02:00
#ifndef ROCKSDB_LITE
if (!listeners.empty()) {
db_mutex->AssertHeld();
// release lock while notifying events
db_mutex->Unlock();
for (auto& listener : listeners) {
BackgroundErrorRecoveryInfo info;
info.old_bg_error = old_bg_error;
info.new_bg_error = new_bg_error;
listener->OnErrorRecoveryCompleted(old_bg_error);
listener->OnErrorRecoveryEnd(info);
info.old_bg_error.PermitUncheckedError();
info.new_bg_error.PermitUncheckedError();
}
db_mutex->Lock();
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 22:36:19 +02:00
}
#else
(void)listeners;
(void)old_bg_error;
(void)new_bg_error;
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 22:36:19 +02:00
(void)db_mutex;
#endif // ROCKSDB_LITE
}
#ifndef ROCKSDB_LITE
void EventHelpers::NotifyBlobFileCreationStarted(
const std::vector<std::shared_ptr<EventListener>>& listeners,
const std::string& db_name, const std::string& cf_name,
const std::string& file_path, int job_id,
BlobFileCreationReason creation_reason) {
if (listeners.empty()) {
return;
}
BlobFileCreationBriefInfo info(db_name, cf_name, file_path, job_id,
creation_reason);
for (const auto& listener : listeners) {
listener->OnBlobFileCreationStarted(info);
}
}
#endif // !ROCKSDB_LITE
void EventHelpers::LogAndNotifyBlobFileCreationFinished(
EventLogger* event_logger,
const std::vector<std::shared_ptr<EventListener>>& listeners,
const std::string& db_name, const std::string& cf_name,
const std::string& file_path, int job_id, uint64_t file_number,
BlobFileCreationReason creation_reason, const Status& s,
const std::string& file_checksum,
const std::string& file_checksum_func_name, uint64_t total_blob_count,
uint64_t total_blob_bytes) {
if (s.ok() && event_logger) {
JSONWriter jwriter;
AppendCurrentTime(&jwriter);
jwriter << "cf_name" << cf_name << "job" << job_id << "event"
<< "blob_file_creation"
<< "file_number" << file_number << "total_blob_count"
<< total_blob_count << "total_blob_bytes" << total_blob_bytes
<< "file_checksum" << file_checksum << "file_checksum_func_name"
<< file_checksum_func_name << "status" << s.ToString();
jwriter.EndObject();
event_logger->Log(jwriter);
}
#ifndef ROCKSDB_LITE
if (listeners.empty()) {
return;
}
BlobFileCreationInfo info(db_name, cf_name, file_path, job_id,
creation_reason, total_blob_count, total_blob_bytes,
s, file_checksum, file_checksum_func_name);
for (const auto& listener : listeners) {
listener->OnBlobFileCreated(info);
}
info.status.PermitUncheckedError();
#else
(void)listeners;
(void)db_name;
(void)file_path;
(void)creation_reason;
#endif
}
void EventHelpers::LogAndNotifyBlobFileDeletion(
EventLogger* event_logger,
const std::vector<std::shared_ptr<EventListener>>& listeners, int job_id,
uint64_t file_number, const std::string& file_path, const Status& status,
const std::string& dbname) {
if (event_logger) {
JSONWriter jwriter;
AppendCurrentTime(&jwriter);
jwriter << "job" << job_id << "event"
<< "blob_file_deletion"
<< "file_number" << file_number;
if (!status.ok()) {
jwriter << "status" << status.ToString();
}
jwriter.EndObject();
event_logger->Log(jwriter);
}
#ifndef ROCKSDB_LITE
if (listeners.empty()) {
return;
}
BlobFileDeletionInfo info(dbname, file_path, job_id, status);
for (const auto& listener : listeners) {
listener->OnBlobFileDeleted(info);
}
info.status.PermitUncheckedError();
#else
(void)listeners;
(void)dbname;
(void)file_path;
#endif // !ROCKSDB_LITE
}
} // namespace ROCKSDB_NAMESPACE