Summary:
Consider the following WAL with 4 batch entries prefixed with their sequence at time of memtable insert.
[1: BEGIN_PREPARE, PUT, PUT, PUT, PUT, END_PREPARE(a)]
[1: BEGIN_PREPARE, PUT, PUT, PUT, PUT, END_PREPARE(b)]
[4: COMMIT(a)]
[7: COMMIT(b)]
The first two batches do not consume any sequence numbers so are both prefixed with seq=1.
For 2pc commit, memtable insertion takes place before COMMIT batch is written to WAL.
We can see that sequence number consumption takes place between WAL entries giving us the seemingly sparse sequence prefix for WAL entries.
This is a valid WAL.
Because with 2PC markers one WriteBatch points to another batch containing its inserts a writebatch can consume more or less sequence numbers than the number of sequence consuming entries that it contains.
We can see that, given the entries in the WAL, 6 sequence ids were consumed. Yet on recovery the maximum sequence consumed would be 7 + 3 (the number of sequence numbers consumed by COMMIT(b))
So, now upon recovery we must track the actual consumption of sequence numbers.
In the provided scenario there will be no sequence gaps, but it is possible to produce a sequence gap. This should not be a problem though. correct?
Test Plan: provided test.
Reviewers: sdong
Subscribers: andrewkr, leveldb, dhruba, hermanlee4
Differential Revision: https://reviews.facebook.net/D57645
Summary:
This diff is built on top of WriteBatch modification: https://reviews.facebook.net/D54093 and adds the required functionality to rocksdb core necessary for rocksdb to support 2PC.
modfication of DBImpl::WriteImpl()
- added two arguments *uint64_t log_used = nullptr, uint64_t log_ref = 0;
- *log_used is an output argument which will return the log number which the incoming batch was inserted into, 0 if no WAL insert took place.
- log_ref is a supplied log_number which all memtables inserted into will reference after the batch insert takes place. This number will reside in 'FindMinPrepLogReferencedByMemTable()' until all Memtables insertinto have flushed.
- Recovery/writepath is now aware of prepared batches and commit and rollback markers.
Test Plan: There is currently no test on this diff. All testing of this functionality takes place in the Transaction layer/diff but I will add some testing.
Reviewers: IslamAbdelRahman, sdong
Subscribers: leveldb, santoshb, andrewkr, vasilep, dhruba, hermanlee4
Differential Revision: https://reviews.facebook.net/D56919
Summary:
On Mac OS X, the chroot directory we typically use ("/tmp") is actually
a symlink for "/private/tmp". Since we dereference symlinks in user-defined
paths, we must also dereference symlinks in chroot_dir_ such that we can perform
string comparisons on those paths.
Test Plan: ran env_test on Mac OS X and devserver
Reviewers: sdong, IslamAbdelRahman
Reviewed By: IslamAbdelRahman
Subscribers: andrewkr, dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D57957
Summary:
Add a new option that can be used to set a specific compression algorithm for bottommost level.
This option will only affect levels larger than base level.
I have also updated CompactionJobInfo to include the compression algorithm used in compaction
Test Plan:
added new unittest
existing unittests
Reviewers: andrewkr, yhchiang, sdong
Reviewed By: sdong
Subscribers: lightmark, andrewkr, dhruba, yoshinorim
Differential Revision: https://reviews.facebook.net/D57669
Summary:
For testing backups, we needed an Env that is fully isolated from other
Envs on the same machine. Our in-memory Envs (MockEnv and InMemoryEnv) were
insufficient because they don't implement most directory operations.
This diff introduces a new Env, "ChrootEnv", that translates paths such that the
chroot directory appears to be the root directory. This way, multiple Envs can
be isolated in the filesystem by using different chroot directories. Since we
use the filesystem, all directory operations are trivially supported.
Test Plan:
I parameterized the existing EnvPosixTest so it runs tests on ChrootEnv
except the ioctl-related cases.
Reviewers: sdong, lightmark, IslamAbdelRahman
Reviewed By: IslamAbdelRahman
Subscribers: andrewkr, dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D57543
Summary:
This is needed so that rocksdb users can add more
commands to the included ldb tool by adding more custom
commands.
Test Plan: make -j ldb
Reviewers: sdong, IslamAbdelRahman
Reviewed By: IslamAbdelRahman
Subscribers: andrewkr, dhruba
Differential Revision: https://reviews.facebook.net/D57243
Summary:
For backwards compatibility with older option strings, the parser needs
to treat this argument as optional.
Test Plan:
Updated unit test to cover case where compression_opts is present but
max_dict_bytes is omitted.
Reviewers: MarkCallaghan, sdong, IslamAbdelRahman
Reviewed By: IslamAbdelRahman
Subscribers: andrewkr, dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D57759
Summary: We changed default options of max_open_files and max_file_opening_threads but didn't revert it in OptimizeForSmallDb().
Test Plan: Add a unit test
Reviewers: igor, yhchiang, IslamAbdelRahman
Reviewed By: IslamAbdelRahman
Subscribers: leveldb, andrewkr, dhruba
Differential Revision: https://reviews.facebook.net/D57675
Summary:
Add an option `iterator_readahead_size` to `ReadOptions` to enable
configurable readahead for iterators similar to the corresponding
option for compaction.
Test Plan:
```
make commit_prereq
```
Reviewers: kumar.rangarajan, ott, igor, sdong
Reviewed By: sdong
Subscribers: yiwu, andrewkr, dhruba
Differential Revision: https://reviews.facebook.net/D55419
Summary:
Changing several option defaults:
options.max_open_files changes from 5000 to -1
options.base_background_compactions changes from max_background_compactions to 1
options.wal_recovery_mode changes from kTolerateCorruptedTailRecords to kTolerateCorruptedTailRecords
options.compaction_pri changes from kByCompensatedSize to kByCompensatedSize
Test Plan: Write unit tests to see OldDefaults() works as expected.
Reviewers: IslamAbdelRahman, yhchiang, igor
Reviewed By: igor
Subscribers: MarkCallaghan, yiwu, kradhakrishnan, leveldb, andrewkr, dhruba
Differential Revision: https://reviews.facebook.net/D56427
Summary:
This adds a new metablock containing a shared dictionary that is used
to compress all data blocks in the SST file. The size of the shared dictionary
is configurable in CompressionOptions and defaults to 0. It's currently only
used for zlib/lz4/lz4hc, but the block will be stored in the SST regardless of
the compression type if the user chooses a nonzero dictionary size.
During compaction, computes the dictionary by randomly sampling the first
output file in each subcompaction. It pre-computes the intervals to sample
by assuming the output file will have the maximum allowable length. In case
the file is smaller, some of the pre-computed sampling intervals can be beyond
end-of-file, in which case we skip over those samples and the dictionary will
be a bit smaller. After the dictionary is generated using the first file in a
subcompaction, it is loaded into the compression library before writing each
block in each subsequent file of that subcompaction.
On the read path, gets the dictionary from the metablock, if it exists. Then,
loads that dictionary into the compression library before reading each block.
Test Plan: new unit test
Reviewers: yhchiang, IslamAbdelRahman, cyan, sdong
Reviewed By: sdong
Subscribers: andrewkr, yoshinorim, kradhakrishnan, dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D52287
Summary:
Introduced option to dump malloc statistics using new option flag.
Added new command line option to db_bench tool to enable this
funtionality.
Also extended build to support environments with/without jemalloc.
Test Plan:
1) Build rocksdb using `make` command. Launch the following command
`./db_bench --benchmarks=fillrandom --dump_malloc_stats=true
--num=10000000` end verified that jemalloc dump is present in LOG file.
2) Build rocksdb using `DISABLE_JEMALLOC=1 make db_bench -j32` and ran
the same db_bench tool and found the following message in LOG file:
"Please compile with jemalloc to enable malloc dump".
3) Also built rocksdb using `make` command on MacOS to verify behavior
in non-FB environment.
Also to debug build configuration change temporary changed
AM_DEFAULT_VERBOSITY = 1 in Makefile to see compiler and build
tools output. For case 1) -DROCKSDB_JEMALLOC was present in compiler
command line. For both 2) and 3) this flag was not present.
Reviewers: sdong
Reviewed By: sdong
Subscribers: andrewkr, dhruba
Differential Revision: https://reviews.facebook.net/D57321
Summary:
The current implementation find the first different byte and try to increment it, if it cannot it return the original key
we can improve this by keep going after the first different byte to find the first non 0xFF byte and increment it
After trying this patch on some logdevice sst files I see decrease in there index block size by 8.5%
Test Plan: existing tests and updated test
Reviewers: yhchiang, andrewkr, sdong
Reviewed By: sdong
Subscribers: andrewkr, dhruba
Differential Revision: https://reviews.facebook.net/D56241
* Musl libc does not provide adaptive mutex. Added feature test for PTHREAD_MUTEX_ADAPTIVE_NP.
* Musl libc does not provide backtrace(3). Added a feature check for backtrace(3).
* Fixed compiler error.
* Musl libc does not implement backtrace(3). Added platform check for libexecinfo.
* Alpine does not appear to support gcc -pg option. By default (gcc has PIE option enabled) it fails with:
gcc: error: -pie and -pg|p|profile are incompatible when linking
When -fno-PIE and -nopie are used it fails with:
/usr/lib/gcc/x86_64-alpine-linux-musl/5.3.0/../../../../x86_64-alpine-linux-musl/bin/ld: cannot find gcrt1.o: No such file or directory
Added gcc -pg platform test and output PROFILING_FLAGS accordingly. Replaced pg var in Makefile with PROFILING_FLAGS.
* fix segfault when TEST_IOCTL_FRIENDLY_TMPDIR is undefined and default candidates are not suitable
* use ASSERT_DOUBLE_EQ instead of ASSERT_EQ
* When compiled with ROCKSDB_MALLOC_USABLE_SIZE UniversalCompactionFourPaths and UniversalCompactionSecondPathRatio tests fail due to premature memtable flushes on systems with 16-byte alignment. Arena runs out of block space before GenerateNewFile() completes.
Increased options.write_buffer_size.
Comparable with Snappy on comp ratio.
Implemented using Windows API, does not require external package.
Avaiable since Windows 8 and server 2012.
Use -DXPRESS=1 with CMake to enable.
Summary: It is useful to print out IO stats in flush jobs too. Extend options.compaction_measure_io_stats to flush jobs and raname it.
Test Plan: Try db_bench and see the stats are printed out.
Reviewers: yhchiang
Reviewed By: yhchiang
Subscribers: kradhakrishnan, yiwu, IslamAbdelRahman, leveldb, andrewkr, dhruba
Differential Revision: https://reviews.facebook.net/D56769
Summary: If the user specified a small enough value for the rate limiter's bytes per second, the calculation for the number of refill bytes per period could become zero which would effectively cause the server to hang forever.
Test Plan: Existing tests
Reviewers: sdong, yhchiang, igor
Reviewed By: igor
Subscribers: leveldb, andrewkr, dhruba
Differential Revision: https://reviews.facebook.net/D56631
Summary: In option settable tests, bytes for pointers are not all skipped, so that they may be the same as the special character and cause false positive.
Test Plan: Run the test. Manually verify the issue is not there any more.
Reviewers: IslamAbdelRahman, andrewkr
Reviewed By: IslamAbdelRahman
Subscribers: kradhakrishnan, yiwu, yhchiang, leveldb, andrewkr, dhruba
Differential Revision: https://reviews.facebook.net/D56553
Summary: Test DBOptionsAllFieldsSettable sometimes fails under valgrind. Move option settable tests to a separate test file and disable it in valgrind..
Test Plan: Run valgrind test and make sure the test doesn't run.
Reviewers: andrewkr, IslamAbdelRahman
Reviewed By: IslamAbdelRahman
Subscribers: kradhakrishnan, yiwu, yhchiang, leveldb, andrewkr, dhruba
Differential Revision: https://reviews.facebook.net/D56529
Summary:
- Need to use unsigned long long for 64-bit literals on windows
- Need size_t for backup meta-file length since clang doesn't let us assign size_t to int
Test Plan: backupable_db_test and options_test
Reviewers: IslamAbdelRahman, yhchiang, sdong
Reviewed By: sdong
Subscribers: andrewkr, dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D56391
Summary: Cache shard bit 4 is sometimes too small and 6 is a more common value picked by users. Make that default. It shouldn't hurt much to change options.max_file_opening_threads default to be 16, which will reduce the worst case DB open time.
Test Plan: Run all existing tests.
Reviewers: IslamAbdelRahman, yhchiang, kradhakrishnan, igor, andrewkr
Reviewed By: andrewkr
Subscribers: andrewkr, MarkCallaghan, leveldb, dhruba
Differential Revision: https://reviews.facebook.net/D55047
Summary:
Adapted a stderr logger from the option tests. Moved it to a separate
header so we can reuse it, e.g., from ldb subcommands for faster debugging. This
is especially useful to make errors/warnings more visible when running
"ldb repair", which involves potential data loss.
Test Plan:
ran options_test and "ldb repair"
$ ./ldb repair --db=./tmp/
[WARN] **** Repaired rocksdb ./tmp/; recovered 1 files; 588bytes. Some data may have been lost. ****
OK
Reviewers: IslamAbdelRahman, yhchiang, sdong
Reviewed By: sdong
Subscribers: andrewkr, dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D56151
Summary:
When a block based table file is opened, if prefetch_index_and_filter is true, it will prefetch the index and filter blocks, putting them into the block cache.
What this feature adds: when a L0 block based table file is opened, if pin_l0_filter_and_index_blocks_in_cache is true in the options (and prefetch_index_and_filter is true), then the filter and index blocks aren't released back to the block cache at the end of BlockBasedTableReader::Open(). Instead the table reader takes ownership of them, hence pinning them, ie. the LRU cache will never push them out. Meanwhile in the table reader, further accesses will not hit the block cache, thus avoiding lock contention.
Test Plan:
'export TEST_TMPDIR=/dev/shm/ && DISABLE_JEMALLOC=1 OPT=-g make all valgrind_check -j32' is OK.
I didn't run the Java tests, I don't have Java set up on my devserver.
Reviewers: sdong
Reviewed By: sdong
Subscribers: andrewkr, dhruba
Differential Revision: https://reviews.facebook.net/D56133
Summary: Change some RocksDB default options to make it more friendly to server workloads.
Test Plan: Run all existing tests
Reviewers: yhchiang, IslamAbdelRahman
Reviewed By: IslamAbdelRahman
Subscribers: sumeet, muthu, benj, MarkCallaghan, igor, leveldb, andrewkr, dhruba
Differential Revision: https://reviews.facebook.net/D55941
Summary:
Expose the inverse of ToString(hex=true) on Slice: Slice::DecodeHex
Refactor the other implementation of to/from hex in ldb_cmd.h to use the Slice
version
(Difference between the 2 is whether 0x is expected/produced in front of the hex
string or not)
Eliminated support for invalid odd length hex string - this is now invalid
instead of having 1/2 byte set
Added (inverse of HexToString) test for LDBCommand::StringToHex which also
indirectly tests Slice::ToString(true)
After moving the original implementation from ldb_cmd.h, updated it to much simpler/efficient version
(originally/inspired from https://github.com/facebook/wdt/blob/master/util/EncryptionUtils.cpp#L140-L169 )
Test Plan: run tests
Reviewers: uddipta, sdong
Reviewed By: sdong
Subscribers: andrewkr, dhruba
Differential Revision: https://reviews.facebook.net/D56121
Summary: Something changed and the special charactor seems to be conflict with an exisitng value. Change it to unblock the build.
Test Plan: Run the test and make sure it passes
Reviewers: kradhakrishnan
Reviewed By: kradhakrishnan
Subscribers: leveldb, andrewkr, dhruba
Differential Revision: https://reviews.facebook.net/D55845
Summary:
When a block based table file is opened, if prefetch_index_and_filter is true, it will prefetch the index and filter blocks, putting them into the block cache.
What this feature adds: when a L0 block based table file is opened, if pin_l0_filter_and_index_blocks_in_cache is true in the options (and prefetch_index_and_filter is true), then the filter and index blocks aren't released back to the block cache at the end of BlockBasedTableReader::Open(). Instead the table reader takes ownership of them, hence pinning them, ie. the LRU cache will never push them out. Meanwhile in the table reader, further accesses will not hit the block cache, thus avoiding lock contention.
When the table reader is destroyed, it releases the pinned blocks (if there were any). This has to happen before the cache is destroyed, so I had to introduce a TableReader::Close(), to guarantee the order of destruction.
Test Plan:
Added two unit tests for this. Existing unit tests run fine (default is pin_l0_filter_and_index_blocks_in_cache=false).
DISABLE_JEMALLOC=1 OPT=-g make all valgrind_check -j32
Mac: OK.
Linux: with D55287 patched in it's OK.
Reviewers: sdong
Reviewed By: sdong
Subscribers: andrewkr, leveldb, dhruba
Differential Revision: https://reviews.facebook.net/D54801
Summary:
This fixes a similar issue as D54711: "CURRENT" file can mutate between
GetLiveFiles() and copy to the tmp directory, in which case it would reference
the wrong manifest filename. To fix this, I forge the "CURRENT" file such that
it simply contains the filename for the manifest returned by GetLiveFiles().
- Changed CreateCheckpoint() to forge current file
- Added CreateFile() utility function
- Added test case that rolls manifest during checkpoint creation
Test Plan:
$ ./checkpoint_test
Reviewers: sdong, IslamAbdelRahman
Reviewed By: IslamAbdelRahman
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D55065
Summary: Refactored db_bench transaction stress tests so that they can be called from unit tests as well.
Test Plan: run new unit test as well as db_bench
Reviewers: yhchiang, IslamAbdelRahman, sdong
Reviewed By: IslamAbdelRahman
Subscribers: andrewkr, dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D55203
Summary:
Cache to have an option to fail Cache::Insert() when full. Update call sites to check status and handle error.
I totally have no idea what's correct behavior of all the call sites when they encounter error. Please let me know if you see something wrong or more unit test is needed.
Test Plan: make check -j32, see tests pass.
Reviewers: anthony, yhchiang, andrewkr, IslamAbdelRahman, kradhakrishnan, sdong
Reviewed By: sdong
Subscribers: andrewkr, dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D54705
Summary:
Use pure if-then check instead of assert in EraseColumnFamilyInfo
when the specified column family does not found in the cf_info_map_.
So the second deletion will be no op instead of crash.
Test Plan: existing test.
Reviewers: sdong, anthony, kradhakrishnan, IslamAbdelRahman
Reviewed By: IslamAbdelRahman
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D55023
Summary: There are a few options in struct DBOptions that aren't handled by options_helper.cc. Add those missing options so they can be used by GetDBOptionsFromString() and friends.
Test Plan: Updated options_test.cc, reran all tests.
Reviewers: sdong, yhchiang
Reviewed By: yhchiang
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D54603
Summary: similar to D52809 add option to exclude zero counters.
Test Plan:
[yiwu@dev4504.prn1 ~/rocksdb] ./iostats_context_test
[==========] Running 1 test from 1 test case.
[----------] Global test environment set-up.
[----------] 1 test from IOStatsContextTest
[ RUN ] IOStatsContextTest.ToString
[ OK ] IOStatsContextTest.ToString (0 ms)
[----------] 1 test from IOStatsContextTest (0 ms total)
[----------] Global test environment tear-down
[==========] 1 test from 1 test case ran. (0 ms total)
[ PASSED ] 1 test.
Reviewers: anthony, yhchiang, andrewkr, IslamAbdelRahman, kradhakrishnan, sdong
Reviewed By: sdong
Subscribers: leveldb, dhruba
Differential Revision: https://reviews.facebook.net/D54591
Summary:
There was a race condition in the test where the rolling thread
acquired the mutex before the flush thread pinned the logger. Rather than add
more complicated synchronization to fix it, I followed Siying's suggestion to
use SyncPoint in the test code.
Comments in the LoadDependency() invocation explain the reason for each of the
sync points.
Test Plan:
Ran test 1000 times for tsan/asan. Will wait for all sandcastle tests
to finish before committing since this is a tricky test.
Reviewers: IslamAbdelRahman, sdong
Reviewed By: sdong
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D54615
Summary:
Relax the check condition of prefix_extractor in CheckOptionsCompatibility
by allowing changing value from non-nullptr to nullptr or nullptr to
non-nullptr.
Test Plan:
options_test
options_util_test
Reviewers: sdong, anthony, IslamAbdelRahman, kradhakrishnan, gunnarku
Reviewed By: gunnarku
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D54477
Summary: Recent change break thread_local_test by introducing exception, which is disabled in LITE build. Fix it by disabling exception handling in LITE build.
Test Plan: Build with both of LITE and non-LITE
Reviewers: anthony, IslamAbdelRahman, yhchiang, kradhakrishnan, andrewkr
Reviewed By: andrewkr
Subscribers: leveldb, dhruba
Differential Revision: https://reviews.facebook.net/D54513
Summary: I have introduced max_allowed_space_ but did not initialize it
Test Plan: make check
Reviewers: sdong, yhchiang, anthony
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D54357
Summary:
Remove the SyncPoint usage in the destructor of PosixEnv as none
of any active tests is using it.
SyncPoint is a test-only utility class, and it's a static varible.
As a result, using SyncPoint in the destructor of PosixEnv will
make default Env depends on SyncPoint. Removing such dependency
could solve the problem crash issue only reproducable in Mac
environment.
Test Plan: OPT=-DTRAVIS V=1 make -j4 check on Mac environment
Reviewers: sdong, anthony
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D54333
Summary:
Introude SstFileManager::SetMaxAllowedSpaceUsage() that can be used to limit the maximum space usage allowed for RocksDB.
When this limit is exceeded WriteImpl() will fail and return Status::Aborted()
Test Plan: unit testing
Reviewers: yhchiang, anthony, andrewkr, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D53763
Summary:
For GetLogFileSize() and Flush(), they previously did not follow the
synchronization pattern for accessing logger_. This meant ResetLogger() could
cause logger_ destruction while the unsynchronized functions were accessing it,
causing a segfault.
Also made the mutex instance variable mutable so we can preserve
GetLogFileSize()'s const-ness.
Test Plan:
new test case, it's quite ugly because both threads need to access
one of the functions with SyncPoints (PosixLogger::Flush()), and also special
handling is needed to prevent the mutex and sync points from conflicting.
Reviewers: kradhakrishnan, IslamAbdelRahman, sdong
Reviewed By: sdong
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D54237
Summary:
Previously compilation failed when ROCKSDB_NO_FBCODE=1 because fcntl.h
wasn't included for open().
Related issue: https://github.com/facebook/rocksdb/issues/977
Test Plan:
verified below command works now:
$ make clean && ROCKSDB_NO_FBCODE=1 ROCKSDB_DISABLE_FALLOCATE=1 make -j32 env_test
Reviewers: sdong, yhchiang, igor
Reviewed By: igor
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D54135
Summary:
When a child thread that uses ThreadLocalPtr, ThreadLocalPtr::OnThreadExit
will be called when that child thread is destroyed. However,
OnThreadExit will try to access a static singleton of ThreadLocalPtr,
which will be destroyed when the main thread exit. As a result,
when a child thread that uses ThreadLocalPtr exits AFTER the main thread
exits, illegal memory access will occur.
This diff includes a test that reproduce this legacy bug.
==2095206==ERROR: AddressSanitizer: heap-use-after-free on address
0x608000007fa0 at pc 0x959b79 bp 0x7f5fa7426b60 sp 0x7f5fa7426b58
READ of size 8 at 0x608000007fa0 thread T1
This patch fix this issue by having the thread local mutex never be deleted
(but will leak small piece of memory at the end.) The patch also describe
a better solution (thread_local) in the comment that requires gcc 4.8.1 and
in latest clang as a future work once we agree to move toward gcc 4.8.
Test Plan:
COMPILE_WITH_ASAN=1 make thread_local_test -j32
./thread_local_test --gtest_filter="*MainThreadDiesFirst"
Reviewers: anthony, hermanlee4, sdong
Reviewed By: sdong
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D53013
Summary:
Added this new function, which returns filename, size, and modified
timestamp for each file in the provided directory. The default implementation
retrieves the metadata sequentially using existing functions. In the next diff
I'll make HdfsEnv override this function to use libhdfs's bulk get function.
This won't work on windows due to the path separator.
Test Plan:
new unit test
$ ./env_test --gtest_filter=EnvPosixTest.ConsistentChildrenMetadata
Reviewers: yhchiang, sdong
Reviewed By: sdong
Subscribers: IslamAbdelRahman, dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D53781
Summary: Add a new option to BlockBasedTableOptions that will allow us to change the restart interval for the index block
Test Plan: unit tests
Reviewers: yhchiang, anthony, andrewkr, sdong
Reviewed By: sdong
Subscribers: march, dhruba
Differential Revision: https://reviews.facebook.net/D53721
Summary:
If options.base_background_compactions is given, we try to schedule number of compactions not existing this number, only when L0 files increase to certain number, or pending compaction bytes more than certain threshold, we schedule compactions based on options.max_background_compactions.
The watermarks are calculated based on slowdown thresholds.
Test Plan:
Add new test cases in column_family_test.
Adding more unit tests.
Reviewers: IslamAbdelRahman, yhchiang, kradhakrishnan, rven, anthony
Reviewed By: anthony
Subscribers: leveldb, dhruba, yoshinorim
Differential Revision: https://reviews.facebook.net/D53409
Summary:
Add a new class SstFileTracker that will be notified whenever a DB add/delete/move and sst file, it will also replace DeleteScheduler
SstFileTracker can be used later to abort writes when we exceed a specific size
Test Plan: unit tests
Reviewers: rven, anthony, yhchiang, sdong
Reviewed By: sdong
Subscribers: igor, lovro, march, dhruba
Differential Revision: https://reviews.facebook.net/D50469
Summary: Measuring mutex duration will measure time inside DB mutex, which breaks our best practice. Add a stat level in Statistics class. By default, disable to measure the mutex operations.
Test Plan: Add a unit test to make sure it is off by default.
Reviewers: rven, anthony, IslamAbdelRahman, kradhakrishnan, andrewkr, yhchiang
Reviewed By: yhchiang
Subscribers: MarkCallaghan, leveldb, dhruba
Differential Revision: https://reviews.facebook.net/D53367
Summary:
We can avoid the dependency by forward-declaring ColumnFamilyData and
then treating it as a black box. That means callers of ThreadStatusUtil need to
explicitly provide more options, even if they can be derived from the
ColumnFamilyData, since ThreadStatusUtil doesn't include the definition.
This is part of a series of diffs to eliminate circular dependencies between
directories (e.g., db/* files depending on util/* files and vice-versa).
Test Plan:
$ ./db_test --gtest_filter=DBTest.GetThreadStatus
$ make -j32 commit-prereq
Reviewers: sdong, yhchiang, IslamAbdelRahman
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D53361
Summary:
I split the db-specific test points out into a separate file under db/
directory. There were also a few bugs to fix in xfunc.{h,cc} that prevented it
from compiling previously; see https://reviews.facebook.net/D36825.
Test Plan:
compilation works now, below command works, will also run "make xfunc".
$ make check ROCKSDB_XFUNC_TEST='managed_new' tests-regexp='DBTest' -j32
Reviewers: sdong, yhchiang
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D53343
Summary: Timing mutex operations can impact scalability of the system. Add a new perf context level that can measure time counters except for mutex.
Test Plan: Add a new unit test case to make sure it is not set.
Reviewers: IslamAbdelRahman, rven, kradhakrishnan, yhchiang, anthony
Reviewed By: anthony
Subscribers: MarkCallaghan, leveldb, dhruba
Differential Revision: https://reviews.facebook.net/D53199
Summary: Add a test to fail if someone adds a DB options.
Test Plan: Run the test, run the test with valgrind. Add an option to DB option in the middle or in the end and make sure it fails.
Reviewers: yhchiang, anthony, IslamAbdelRahman, kradhakrishnan, rven, andrewkr
Reviewed By: andrewkr
Subscribers: leveldb, dhruba
Differential Revision: https://reviews.facebook.net/D53097
Summary: OptionsParserTest.BlockBasedTableOptionsAllFieldsSettable is failiong under CLANG. Disable the test to unblock the build.
Test Plan: Run it both of CLANG and GCC
Reviewers: kradhakrishnan, rven, andrewkr, anthony, yhchiang, IslamAbdelRahman
Reviewed By: IslamAbdelRahman
Subscribers: leveldb, dhruba
Differential Revision: https://reviews.facebook.net/D53157
Summary: Add a test OptionsParserTest.BlockBasedTableOptionsAdded, which will fail if a new option is added to BlockBasedTableOptions but is not settable through GetBlockBasedTableOptionsFromString().
Test Plan: Run the test. Also manually remove and add options and make sure it fails.
Reviewers: anthony, IslamAbdelRahman, kradhakrishnan, rven, yhchiang, andrewkr
Reviewed By: andrewkr
Subscribers: leveldb, dhruba
Differential Revision: https://reviews.facebook.net/D52953
Summary:
In crash test, when coming to each kill point, we start a random class using seed as current second. With this approach, for every second, the random number used is the same. However, in each second, there are multiple kill points with different frequency. It makes it hard to reason about chance of kill point to trigger. With this commit, we use thread local random seed to generate the random number, so that it will take different values per second, hoping it makes chances of killing much easier to reason about.
Also significantly reduce the kill odd to make sure time before kiling is similar as before.
Test Plan: Run white box crash test and see the killing happens as expected and the run time time before killing reasonable.
Reviewers: kradhakrishnan, IslamAbdelRahman, rven, yhchiang, andrewkr, anthony
Reviewed By: anthony
Subscribers: leveldb, dhruba
Differential Revision: https://reviews.facebook.net/D52971
Makefile adjust paths for solaris build
Makefile enable _GLIBCXX_USE_C99 so that std::to_string is available
db_compaction_test.cc Initialise a variable to avoid a compilation error
db_impl.cc Include <alloca.h>
db_test.cc Include <alloca.h>
Environment.java recognise solaris envrionment
options_bulder.cc Make log unambiguous
geodb_impl.cc Make log and floor unambiguous
Summary: Depending on the order of include paths and versions of various headers we may end up in a situation where we'll encounter a build break caused by redefinition of constants. gcc-4.9-glibc-2.20 header update to include/bits/fcntl-linux.h introduced the definitions of FALLOC_FL_* constants. However, linux/falloc.h from kernel-headers also has FALLOC_FL_* constants defined. Therefore during the compilation we'll get "previously defined" errors.
Test Plan:
Both in the environment where the build break manifests (to make sure that the change fixed the problem) and in the environment where everything builds fine (to make sure that there are no regressions):
make clean
make -j 32
Reviewers: sdong, igor
Reviewed By: igor
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D52821
Summary: In plain table reader's non-mmap mode, we only keep the most recent read buffer. However, for binary search, it is likely we come back to a location to read. To avoid one pread in such a case, we keep two read buffers. It should cover most of the cases.
Test Plan:
1. run tests
2. check the optimization works through strace when running
./table_reader_bench -mmap_read=false --num_keys2=1 -num_keys1=5000 -table_factory=plain_table --iterator --through_db
Reviewers: anthony, rven, kradhakrishnan, igor, yhchiang, IslamAbdelRahman
Reviewed By: IslamAbdelRahman
Subscribers: leveldb, dhruba
Differential Revision: https://reviews.facebook.net/D51171
Summary: TSAN fails on DynamicBloomTest.concurrent_with_perf. This change fixes it. Not sure why though.
Test Plan: Run the test with TSAN and make sure no warning shown.
Reviewers: yhchiang, IslamAbdelRahman, anthony, ngbronson, rven
Reviewed By: rven
Subscribers: rven, leveldb, dhruba
Differential Revision: https://reviews.facebook.net/D52383
Summary: Adding "__attribute__((__unused__))" after padding fields will pass CLANG build but will fail gcc 4.8.1. Fix it by not generating it under GCC 4.8.1.
Test Plan: Build under four combinations of USE_CLANG=0,1 and ROCKSDB_FBCODE_BUILD_WITH_481=0.1.
Reviewers: yhchiang, rven, ngbronson, anthony, IslamAbdelRahman
Reviewed By: anthony
Subscribers: leveldb, dhruba
Differential Revision: https://reviews.facebook.net/D52371
Summary: Fix some CLANG errors introduced in 7d87f02799
Test Plan: Build with both of CLANG and gcc
Reviewers: rven, yhchiang, kradhakrishnan, anthony, IslamAbdelRahman, ngbronson
Subscribers: leveldb, dhruba
Differential Revision: https://reviews.facebook.net/D52329
Summary:
This diff adds support for concurrent adds to the skiplist memtable
implementations. Memory allocation is made thread-safe by the addition of
a spinlock, with small per-core buffers to avoid contention. Concurrent
memtable writes are made via an additional method and don't impose a
performance overhead on the non-concurrent case, so parallelism can be
selected on a per-batch basis.
Write thread synchronization is an increasing bottleneck for higher levels
of concurrency, so this diff adds --enable_write_thread_adaptive_yield
(default off). This feature causes threads joining a write batch
group to spin for a short time (default 100 usec) using sched_yield,
rather than going to sleep on a mutex. If the timing of the yield calls
indicates that another thread has actually run during the yield then
spinning is avoided. This option improves performance for concurrent
situations even without parallel adds, although it has the potential to
increase CPU usage (and the heuristic adaptation is not yet mature).
Parallel writes are not currently compatible with
inplace updates, update callbacks, or delete filtering.
Enable it with --allow_concurrent_memtable_write (and
--enable_write_thread_adaptive_yield). Parallel memtable writes
are performance neutral when there is no actual parallelism, and in
my experiments (SSD server-class Linux and varying contention and key
sizes for fillrandom) they are always a performance win when there is
more than one thread.
Statistics are updated earlier in the write path, dropping the number
of DB mutex acquisitions from 2 to 1 for almost all cases.
This diff was motivated and inspired by Yahoo's cLSM work. It is more
conservative than cLSM: RocksDB's write batch group leader role is
preserved (along with all of the existing flush and write throttling
logic) and concurrent writers are blocked until all memtable insertions
have completed and the sequence number has been advanced, to preserve
linearizability.
My test config is "db_bench -benchmarks=fillrandom -threads=$T
-batch_size=1 -memtablerep=skip_list -value_size=100 --num=1000000/$T
-level0_slowdown_writes_trigger=9999 -level0_stop_writes_trigger=9999
-disable_auto_compactions --max_write_buffer_number=8
-max_background_flushes=8 --disable_wal --write_buffer_size=160000000
--block_size=16384 --allow_concurrent_memtable_write" on a two-socket
Xeon E5-2660 @ 2.2Ghz with lots of memory and an SSD hard drive. With 1
thread I get ~440Kops/sec. Peak performance for 1 socket (numactl
-N1) is slightly more than 1Mops/sec, at 16 threads. Peak performance
across both sockets happens at 30 threads, and is ~900Kops/sec, although
with fewer threads there is less performance loss when the system has
background work.
Test Plan:
1. concurrent stress tests for InlineSkipList and DynamicBloom
2. make clean; make check
3. make clean; DISABLE_JEMALLOC=1 make valgrind_check; valgrind db_bench
4. make clean; COMPILE_WITH_TSAN=1 make all check; db_bench
5. make clean; COMPILE_WITH_ASAN=1 make all check; db_bench
6. make clean; OPT=-DROCKSDB_LITE make check
7. verify no perf regressions when disabled
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: MarkCallaghan, IslamAbdelRahman, anthony, yhchiang, rven, sdong, guyg8, kradhakrishnan, dhruba
Differential Revision: https://reviews.facebook.net/D50589
Summary: We now have a mechanism to further slowdown writes. Double default options.delayed_write_rate to try to keep the default behavior closer to it used to be.
Test Plan: Run all tests.
Reviewers: IslamAbdelRahman
Reviewed By: IslamAbdelRahman
Subscribers: yhchiang, kradhakrishnan, rven, leveldb, dhruba
Differential Revision: https://reviews.facebook.net/D52281
Summary: It's usually hard for users to set a value of options.delayed_write_rate. With this diff, after slowdown condition triggers, we greedily reduce write rate if estimated pending compaction bytes increase. If estimated compaction pending bytes drop, we increase the write rate.
Test Plan:
Add a unit test
Test with db_bench setting:
TEST_TMPDIR=/dev/shm/ ./db_bench --benchmarks=fillrandom -num=10000000 --soft_pending_compaction_bytes_limit=1000000000 --hard_pending_compaction_bytes_limit=3000000000 --delayed_write_rate=100000000
and make sure without the commit, write stop will happen, but with the commit, it will not happen.
Reviewers: igor, anthony, rven, yhchiang, kradhakrishnan, IslamAbdelRahman
Reviewed By: IslamAbdelRahman
Subscribers: leveldb, dhruba
Differential Revision: https://reviews.facebook.net/D52131
