Summary:
While profiling compaction in our service I noticed a lot of CPU (~15% of compaction) being spent in MergingIterator and key comparison. Looking at the code I found MergingIterator was (understandably) using std::priority_queue for the multiway merge.
Keys in our dataset include sequence numbers that increase with time. Adjacent keys in an L0 file are very likely to be adjacent in the full database. Consequently, compaction will often pick a chunk of rows from the same L0 file before switching to another one. It would be great to avoid the O(log K) operation per row while compacting.
This diff replaces std::priority_queue with a custom binary heap implementation. It has a "replace top" operation that is cheap when the new top is the same as the old one (i.e. the priority of the top entry is decreased but it still stays on top).
Test Plan:
make check
To test the effect on performance, I generated databases with data patterns that mimic what I describe in the summary (rows have a mostly increasing sequence number). I see a 10-15% CPU decrease for compaction (and a matching throughput improvement on tmpfs). The exact improvement depends on the number of L0 files and the amount of locality. Performance on randomly distributed keys seems on par with the old code.
Reviewers: kailiu, sdong, igor
Reviewed By: igor
Subscribers: yoshinorim, dhruba, tnovak
Differential Revision: https://reviews.facebook.net/D29133
Summary:
Introduced a new category in the enum InfoLogLevel in env.h.
Modifed Log() in env.cc to use the Header()
when the InfoLogLevel == HEADER_LEVEL.
Updated tests in auto_roll_logger_test to ensure
the header is handled properly in these cases.
Test Plan: Augment existing tests in auto_roll_logger_test
Reviewers: igor, sdong, yhchiang
Reviewed By: yhchiang
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D41067
Summary:
Add a new field: BackupableDBOptions.max_background_copies.
CreateNewBackup() and RestoreDBFromBackup() will use this number of threads to perform copies.
If there is a backup rate limit, then max_background_copies must be 1.
Update backupable_db_test.cc to test multi-threaded backup and restore.
Update backupable_db_test.cc to test backups when the backup environment is not the same as the database environment.
Test Plan:
Run ./backupable_db_test
Run valgrind ./backupable_db_test
Run with TSAN and ASAN
Reviewers: yhchiang, rven, anthony, sdong, igor
Reviewed By: igor
Subscribers: yhchiang, anthony, sdong, leveldb, dhruba
Differential Revision: https://reviews.facebook.net/D40725
invalid suffix on literal
no return statement in function returning non-void CuckooStep::operator=
extra qualification ‘rocksdb::spatial::Variant::
dereferencing type-punned pointer will break strict-aliasing rules
Summary: Make RocksDb build and run on Windows to be functionally
complete and performant. All existing test cases run with no
regressions. Performance numbers are in the pull-request.
Test plan: make all of the existing unit tests pass, obtain perf numbers.
Co-authored-by: Praveen Rao praveensinghrao@outlook.com
Co-authored-by: Sherlock Huang baihan.huang@gmail.com
Co-authored-by: Alex Zinoviev alexander.zinoviev@me.com
Co-authored-by: Dmitri Smirnov dmitrism@microsoft.com
Summary:
Implementation of a table-level row cache.
It only caches point queries done through the `DB::Get` interface, queries done through the `Iterator` interface will completely skip the cache.
Supports snapshots and merge operations.
Test Plan: Ran `make valgrind_check commit-prereq`
Reviewers: igor, philipp, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D39849
Summary:
The "one size fits all" approach with WAL recovery will only introduce inconvenience for our varied clients as we go forward. The current recovery is a bit heuristic. We introduce the following levels of consistency while replaying the WAL.
1. RecoverAfterRestart (kTolerateCorruptedTailRecords)
This mocks the current recovery mode.
2. RecoverAfterCleanShutdown (kAbsoluteConsistency)
This is ideal for unit test and cases where the store is shutdown cleanly. We tolerate no corruption or incomplete writes.
3. RecoverPointInTime (kPointInTimeRecovery)
This is ideal when using devices with controller cache or file systems which can loose data on restart. We recover upto the point were is no corruption or incomplete write.
4. RecoverAfterDisaster (kSkipAnyCorruptRecord)
This is ideal mode to recover data. We tolerate corruption and incomplete writes, and we hop over those sections that we cannot make sense of salvaging as many records as possible.
Test Plan:
(1) Run added unit test to cover all levels.
(2) Run make check.
Reviewers: leveldb, sdong, igor
Subscribers: yoshinorim, dhruba
Differential Revision: https://reviews.facebook.net/D38487
Summary: MyRocks need a mechanism to track read outliers. We need to expose this
stat.
Test Plan: None
Reviewers: sdong
CC: leveldb
Task ID: #7152512
Blame Rev:
Summary: See title
Test Plan: Run valgrind ./cache_test
Reviewers: igor
Reviewed By: igor
Subscribers: anthony, dhruba
Differential Revision: https://reviews.facebook.net/D40419
Summary: Make autovector_test runnable in ROCKSDB_LITE
Test Plan: autovector_test
Reviewers: sdong, rven, anthony, kradhakrishnan, IslamAbdelRahman, igor
Reviewed By: igor
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D40245
Summary:
Currently RocksDB silently ignores this issue and doesn't compress the data. Based on discussion, we agree that this is pretty bad because it can cause confusion for our users.
This patch fails DB::Open() if we don't support the compression that is specified in the options.
Test Plan: make check with LZ4 not present. If Snappy is not present all tests will just fail because Snappy is our default library. We should make Snappy the requirement, since without it our default DB::Open() fails.
Reviewers: sdong, MarkCallaghan, rven, yhchiang
Reviewed By: yhchiang
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D39687
Summary:
Add the funcion Cache.GetPinnedUsage() to return the memory size of entries
that are in use by the system (that is, all the entries not in the LRU list).
Test Plan:
Run ./cache_test and examine PinnedUsageTest.
Reviewers: tnovak, igor
Reviewed By: igor
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D40305
Summary: Currently we dump DBOptions for each column family options we dump. This leads to duplicate lines in our LOG file. This diff fixes that.
Test Plan: Check out the LOG
Reviewers: sdong, rven, yhchiang
Reviewed By: yhchiang
Subscribers: IslamAbdelRahman, yoshinorim, dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D39729
Summary:
This diff update DB::CompactRange to use RangeCompactionOptions instead of using multiple parameters
Old CompactRange is still available but deprecated
Test Plan:
make all check
make rocksdbjava
USE_CLANG=1 make all
OPT=-DROCKSDB_LITE make release
Reviewers: sdong, yhchiang, igor
Reviewed By: igor
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D40209
Summary:
Before this patch, any function call to ThreadStatusUtil might automatically initialize and register the thread status data. However, if it is the user-thread making this call, the allocated thread-status-data will never be released as such threads are not managed by rocksdb.
In this patch, I remove the automatic-initialization part. Thread-status data is only initialized and uninitialized in Env during the thread creation and destruction.
Test Plan:
db_test
thread_list_test
listener_test
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D40017
Summary:
Add an option in GetApproximateSize() so that the result will include estimated sizes in mem tables.
To implement it, implement an estimated count from the beginning to a key in skip list. The approach is to count to find the entry, how many Next() is issued from each level, and sum them with a weight that is <branching factor> ^ <level>.
Test Plan: Add a test case
Subscribers: leveldb, dhruba
Differential Revision: https://reviews.facebook.net/D40119
Summary: Removed two unused macros in iostats_context
Test Plan: make all check
Reviewers: sdong, rven, IslamAbdelRahman, kradhakrishnan, igor
Reviewed By: igor
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D40005
Summary:
We slow down data into the database to the rate of options.delayed_write_rate (a new option) with this patch.
The thread synchronization approach I take is to still synchronize write controller by DB mutex and GetDelay() is inside DB mutex. Try to minimize the frequency of getting time in GetDelay(). I verified it through db_bench and it seems to work
hard_rate_limit is deprecated.
options.delayed_write_rate is still not dynamically changeable. Need to work on it as a follow-up.
Test Plan: Add new unit tests in db_test
Reviewers: yhchiang, rven, kradhakrishnan, anthony, MarkCallaghan, igor
Reviewed By: igor
Subscribers: ikabiljo, leveldb, dhruba
Differential Revision: https://reviews.facebook.net/D36351
Summary:
Add Env::GetThreadID(), which returns the ID of the current thread.
In addition, make GetThreadList() and InfoLog use same unique ID for the same thread.
Test Plan:
db_test
listener_test
Reviewers: igor, rven, IslamAbdelRahman, kradhakrishnan, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D39735
Summary:
Replacing the default value for compaction_filter_factory and compaction_filter_factory_v2 to be nullptr instead of DefaultCompactionFilterFactory / DefaultCompactionFilterFactoryV2
The reason for this is to be able to determine easily if we have compaction filter factory or not without depending on RTTI
Test Plan: make check
Reviewers: yoshinorim, ott, igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D39693
Summary:
The type of smallest_output_key_prefix and largest_output_key_prefix
have been changed to std::string in https://reviews.facebook.net/D39537.
As a result, we shouldn't do smallest_output_key_prefix[0] = 0 in the
initialization.
Test Plan: compile db_test with tsan enabled and repeat DBTest.CompactionDeletionTrigger test to verify the tsan issue has been gone.
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D39645
Summary:
Allow EventListener::OnCompactionCompleted to return CompactionJobStats,
which contains useful information about a compaction.
Example CompactionJobStats returned by OnCompactionCompleted():
smallest_output_key_prefix 05000000
largest_output_key_prefix 06990000
elapsed_time 42419
num_input_records 300
num_input_files 3
num_input_files_at_output_level 2
num_output_records 200
num_output_files 1
actual_bytes_input 167200
actual_bytes_output 110688
total_input_raw_key_bytes 5400
total_input_raw_value_bytes 300000
num_records_replaced 100
is_manual_compaction 1
Test Plan: Developed a mega test in db_test which covers 20 variables in CompactionJobStats.
Reviewers: rven, igor, anthony, sdong
Reviewed By: sdong
Subscribers: tnovak, dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D38463
Summary:
We occasionally get write stalls (>1s Write() calls) on HDD under read load. The following timers explain almost all of the stalls:
- perf_context.db_mutex_lock_nanos
- perf_context.db_condition_wait_nanos
- iostats_context.open_time
- iostats_context.allocate_time
- iostats_context.write_time
- iostats_context.range_sync_time
- iostats_context.logger_time
In my experiments each of these occasionally takes >1s on write path under some workload. There are rare cases when Write() takes long but none of these takes long.
Test Plan: Added code to our application to write the listed timings to log for slow writes. They usually add up to almost exactly the time Write() call took.
Reviewers: rven, yhchiang, sdong
Reviewed By: sdong
Subscribers: march, dhruba, tnovak
Differential Revision: https://reviews.facebook.net/D39177
Summary: It used to be no good (known to me) non-intrusive way to wrap WritableFile - you can't call protected virtual methods of the wrapped pointer to WritableFile. This diff adds a convenience class WritableFileWrapper that makes wrapping WritableFile both possible and easy.
Test Plan: `make clean; make -j release`, `make clean; OPT=-DROCKSDB_LITE make release`, `make clean; USE_CLANG=1 make -j all`.
Reviewers: sdong, yhchiang, rven
Reviewed By: rven
Subscribers: dhruba, tnovak, march
Differential Revision: https://reviews.facebook.net/D39147
Summary: Optimistic transactions supporting begin/commit/rollback semantics. Currently relies on checking the memtable to determine if there are any collisions at commit time. Not yet implemented would be a way of enuring the memtable has some minimum amount of history so that we won't fail to commit when the memtable is empty. You should probably start with transaction.h to get an overview of what is currently supported.
Test Plan: Added a new test, but still need to look into stress testing.
Reviewers: yhchiang, igor, rven, sdong
Reviewed By: sdong
Subscribers: adamretter, MarkCallaghan, leveldb, dhruba
Differential Revision: https://reviews.facebook.net/D33435
Summary:
For transactions, we are using the memtables to validate that there are no write conflicts. But after flushing, we don't have any memtables, and transactions could fail to commit. So we want to someone keep around some extra history to use for conflict checking. In addition, we want to provide a way to increase the size of this history if too many transactions fail to commit.
After chatting with people, it seems like everyone prefers just using Memtables to store this history (instead of a separate history structure). It seems like the best place for this is abstracted inside the memtable_list. I decide to create a separate list in MemtableListVersion as using the same list complicated the flush/installalflushresults logic too much.
This diff adds a new parameter to control how much memtable history to keep around after flushing. However, it sounds like people aren't too fond of adding new parameters. So I am making the default size of flushed+not-flushed memtables be set to max_write_buffers. This should not change the maximum amount of memory used, but make it more likely we're using closer the the limit. (We are now postponing deleting flushed memtables until the max_write_buffer limit is reached). So while we might use more memory on average, we are still obeying the limit set (and you could argue it's better to go ahead and use up memory now instead of waiting for a write stall to happen to test this limit).
However, if people are opposed to this default behavior, we can easily set it to 0 and require this parameter be set in order to use transactions.
Test Plan: Added a xfunc test to play around with setting different values of this parameter in all tests. Added testing in memtablelist_test and planning on adding more testing here.
Reviewers: sdong, rven, igor
Reviewed By: igor
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D37443
Summary:
Allow EventLogger to directly log from a JSONWriter. This allows
the JSONWriter to be shared by EventLogger and potentially EventListener,
which is an important step to integrate EventLogger and EventListener.
This patch also rewrites EventLoggerHelpers::LogTableFileCreation(),
which uses the new API to generate identical log.
Test Plan:
Run db_bench in debug mode and make sure the log is correct and no
assertions fail.
Reviewers: sdong, anthony, igor
Reviewed By: igor
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D38709
Summary: Having stats in our LOG more often will help a lot with perf debugging.
Test Plan: none
Reviewers: sdong, MarkCallaghan
Reviewed By: MarkCallaghan
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D38781
Summary: Rename JSONWritter to JSONWriter
Test Plan: make
Reviewers: igor
Reviewed By: igor
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D38733
Summary:
sync_file_range is not always asyncronous and thus can block writes if we do this for WAL in the foreground thread. See more here: http://yoshinorimatsunobu.blogspot.com/2014/03/how-syncfilerange-really-works.html
Some users don't want us to call sync_file_range on WALs. Some other do.
Thus, I'm adding a separate option wal_bytes_per_sync to control calling
sync_file_range on WAL files. bytes_per_sync will apply only to table
files now.
Test Plan: no more sync_file_range for WAL as evidenced by strace
Reviewers: yhchiang, rven, sdong
Reviewed By: sdong
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D38253
Summary:
Fixed the following compile errors due to some gcc does not have std::map::emplace
util/thread_status_impl.cc: In static member function ‘static std::map<std::basic_string<char>, long unsigned int> rocksdb::ThreadStatus::InterpretOperationProperties(rocksdb::ThreadStatus::OperationType, const uint64_t*)’:
util/thread_status_impl.cc:88:20: error: ‘class std::map<std::basic_string<char>, long unsigned int>’ has no member named ‘emplace’
util/thread_status_impl.cc:90:20: error: ‘class std::map<std::basic_string<char>, long unsigned int>’ has no member named ‘emplace’
util/thread_status_impl.cc:94:20: error: ‘class std::map<std::basic_string<char>, long unsigned int>’ has no member named ‘emplace’
util/thread_status_impl.cc:96:20: error: ‘class std::map<std::basic_string<char>, long unsigned int>’ has no member named ‘emplace’
util/thread_status_impl.cc:98:20: error: ‘class std::map<std::basic_string<char>, long unsigned int>’ has no member named ‘emplace’
util/thread_status_impl.cc:101:20: error: ‘class std::map<std::basic_string<char>, long unsigned int>’ has no member named ‘emplace’
make: *** [util/thread_status_impl.o] Error 1
Test Plan: make db_bench
Reviewers: igor
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D38643
Summary: Call Flush() function instead
Test Plan: make all check
Reviewers: igor
Reviewed By: igor
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D38583
Summary:
Allow GetThreadList to report Flush properties, which includes:
* job id
* number of bytes that has been written since flush started.
* total size of input mem-tables
Test Plan:
./db_bench --threads=30 --num=1000000 --benchmarks=fillrandom --thread_status_per_interval=100 --value_size=1000
Sample output from db_bench which tracks same flush job
ThreadID ThreadType cfName Operation ElapsedTime Stage State OperationProperties
140213879898240 High Pri default Flush 5789 us FlushJob::WriteLevel0Table BytesMemtables 4112835 | BytesWritten 577104 | JobID 8 |
ThreadID ThreadType cfName Operation ElapsedTime Stage State OperationProperties
140213879898240 High Pri default Flush 30.634 ms FlushJob::WriteLevel0Table BytesMemtables 4112835 | BytesWritten 1734865 | JobID 8 |
Reviewers: rven, igor, sdong
Reviewed By: sdong
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D38505
Summary: Make ThreadStatus::InterpretOperationProperties take const uint64_t*
Test Plan:
make
make OPT=-DROCKSDB_LITE shared_lib
Reviewers: igor
Reviewed By: igor
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D38445
Summary:
Added these events:
* Recovery start, finish and also when recovery creates a file
* Trivial move
* Compaction start, finish and when compaction creates a file
* Flush start, finish
Also includes small fix to EventLogger
Also added option ROCKSDB_PRINT_EVENTS_TO_STDOUT which is useful when we debug things. I've spent far too much time chasing LOG files.
Still didn't get sst table properties in JSON. They are written very deeply into the stack. I'll address in separate diff.
TODO:
* Write specification. Let's first use this for a while and figure out what's good data to put here, too. After that we'll write spec
* Write tools that parse and analyze LOGs. This can be in python or go. Good intern task.
Test Plan: Ran db_bench with ROCKSDB_PRINT_EVENTS_TO_STDOUT. Here's the output: https://phabricator.fb.com/P19811976
Reviewers: sdong, yhchiang, rven, MarkCallaghan, kradhakrishnan, anthony
Reviewed By: anthony
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D37521
Test Plan: Verified that valgrind build passes for cache_test
Reviewers: igor
Reviewed By: igor
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D37665
Summary: Added keyword override for SetCapacity()
Test Plan: Fixes build
Reviewers: igor
Reviewed By: igor
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D37647
Summary:
When new capacity is larger than existing capacity, simply update the capacity to the new valie
When new capacity is less than existing capacity, but more than the usage, simply update the capacity to new value
When new capacity is less than the existing capacity and existing usage both, try to purge entries in LRU if feasible to make usage < capacity
Test Plan: Created unit tests in cache_test.cc
Reviewers: sdong, rven, yhchiang, igor
Reviewed By: igor
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D37527
lovro