Summary:
This diff has two fixes.
1. Fix the bug where compaction does not fail when RocksDB can't create a new file.
2. When NewWritableFiles() fails in OpenCompactionOutputFiles(), previously such fail-to-created file will be still be included as a compaction output. This patch also fixes this bug.
3. Allow VersionEdit::EncodeTo() to return Status and add basic check.
Test Plan:
./version_edit_test
export ROCKSDB_TESTS=FileCreationRandomFailure
./db_test
Reviewers: ljin, sdong, nkg-, igor
Reviewed By: igor
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D25581
Summary:
Abstract out FlushProcess and take it out of DBImpl.
This also includes taking DeletionState outside of DBImpl.
Currently this diff is only doing the refactoring. Future work includes:
1. Decoupling flush_process.cc, make it depend on less state
2. Write flush_process_test, which will mock out everything that FlushProcess depends on and test it in isolation
Test Plan: make check
Reviewers: rven, yhchiang, sdong, ljin
Reviewed By: ljin
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D27561
Summary: No need to expose them in .h
Test Plan: make release
Reviewers: igor, yhchiang, rven, sdong
Reviewed By: sdong
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D27645
Summary: as title
Test Plan:
make release
will run full test on all stacked diffs
Reviewers: sdong, yhchiang, igor
Reviewed By: igor
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D27597
Summary: as title
Test Plan:
make release
will run full test on all stacked diffs before committing
Reviewers: sdong, yhchiang, rven, igor
Reviewed By: igor
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D27591
Summary:
We have several different types of data structures for file information.
FileLevel is kinda of confusing since it only contains file range and
fd. Rename it to LevelFilesBrief to make it clear.
Unfriend CompactedDBImpl as a by product
Test Plan:
make release / make all
will run full test with all stacked diffs
Reviewers: sdong, yhchiang, rven, igor
Reviewed By: igor
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D27585
Summary: as title
Test Plan:
make release
I will run make all check for all stacked diffs before commit
Reviewers: sdong, yhchiang, rven, igor
Reviewed By: igor
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D27573
Summary: as title
Test Plan: running make all check
Reviewers: sdong, yhchiang, rven, igor
Reviewed By: igor
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D27549
Summary: RocksDB already depends on C++11, so we might as well all the goodness that C++11 provides. This means that we don't need AtomicPointer anymore. The less things in port/, the easier it will be to port to other platforms.
Test Plan: make check + careful visual review verifying that NoBarried got memory_order_relaxed, while Acquire/Release methods got memory_order_acquire and memory_order_release
Reviewers: rven, yhchiang, ljin, sdong
Reviewed By: ljin
Subscribers: leveldb
Differential Revision: https://reviews.facebook.net/D27543
Summary:
Make inplace_update_support and inplace_update_num_locks dynamic.
inplace_callback becomes immutable
We are almost free of references to cfd->options() in db_impl
Test Plan: unit test
Reviewers: igor, yhchiang, rven, sdong
Reviewed By: sdong
Subscribers: leveldb
Differential Revision: https://reviews.facebook.net/D25293
Summary: D24513 introduced a bug that a variable is not initialized. It also causes valgrind issue.
Test Plan: Run tests used to fail valgrind and make sure it passes
Reviewers: yhchiang, ljin, igor
Reviewed By: igor
Subscribers: leveldb
Differential Revision: https://reviews.facebook.net/D25569
Summary:
Previously, the log for Universal Compaction does not include the current
number of files in case the compaction is triggered by the number of files.
This diff includes the number of files in the log.
Test Plan:
make
Summary: as title
Test Plan: unit test
Reviewers: sdong, yhchiang, rven, igor
Reviewed By: igor
Subscribers: leveldb
Differential Revision: https://reviews.facebook.net/D25347
Summary: as title
Test Plan: n/a
Reviewers: yhchiang, sdong, igor
Reviewed By: igor
Subscribers: leveldb
Differential Revision: https://reviews.facebook.net/D24963
Summary:
This is not a critical options. Making it dynamic so that we can remove
more reference to cfd->options()
Test Plan: unit test
Reviewers: yhchiang, sdong, igor
Reviewed By: igor
Subscribers: leveldb
Differential Revision: https://reviews.facebook.net/D24957
Summary: Now --bench_size is only used in multireadrandom tests, although the codes allow it to run in all write tests. I don't see a reason why we can't enable it.
Test Plan:
Run
./db_bench -benchmarks multirandomwrite --threads=5 -batch_size=16
and see the stats printed out in LOG to make sure batching really happened.
Reviewers: ljin, yhchiang, rven, igor
Reviewed By: igor
Subscribers: leveldb
Differential Revision: https://reviews.facebook.net/D25509
Summary: It is useful to print out number of keys in DB Stats
Test Plan:
./db_bench --benchmarks fillrandom --num 1000000 -threads 16 -batch_size=16
and watch the outputs in LOG files
Reviewers: MarkCallaghan, ljin, yhchiang, igor
Reviewed By: igor
Subscribers: leveldb
Differential Revision: https://reviews.facebook.net/D24513
Summary:
DeleteFile() call was broken for non-default column family. This fixes it. We might need this feature for mongo.
I also introduced a possibility of deleting oldest file in level 0.
Test Plan: added unit test to deletefile_test
Reviewers: ljin, yhchiang, rven, sdong
Reviewed By: sdong
Subscribers: leveldb
Differential Revision: https://reviews.facebook.net/D24909
Summary:
Add a function as tittle.
Also use the same parameter to fillseekseq too.
Test Plan: Run seekrandom using the new parameter
Reviewers: ljin, MarkCallaghan
Reviewed By: MarkCallaghan
Subscribers: rven, igor, yhchiang, leveldb
Differential Revision: https://reviews.facebook.net/D25035
Summary:
This diff speeds up DB::Open() and Version creation by limiting the number of FileMetaData initialization. The behavior of Version::UpdateAccumulatedStats() is changed as follows:
* It only initializes the first 20 uninitialized FileMetaData from file. This guarantees the size of the latest 20 files will always be compensated when they have any deletion entries. Previously it may initialize all FileMetaData by loading all files at DB::Open().
* In case none the first 20 files has any data entry, UpdateAccumulatedStats() will initialize the FileMetaData of the oldest file.
Test Plan: db_test
Reviewers: igor, sdong, ljin
Reviewed By: ljin
Subscribers: leveldb
Differential Revision: https://reviews.facebook.net/D24255
Summary:
Fix the following Mac compile error.
db/db_test.cc:8686:52: error: C++11 forbids default arguments for lambda expressions [-Werror,-Wlambda-extensions]
auto gen_l0_kb = [this](int start, int size, int stride = 1) {
^ ~
Test Plan:
db_test
Summary: as title
Test Plan: make release
Reviewers: igor
Reviewed By: igor
Subscribers: leveldb
Differential Revision: https://reviews.facebook.net/D25029
Summary: Added one new counter for GetProperty
Test Plan: Not sure if needs a test case. compiles
Reviewers: sdong
Reviewed By: sdong
Subscribers: leveldb
Differential Revision: https://reviews.facebook.net/D25023
Summary: as title
Test Plan: make release
Reviewers: sdong, yhchiang, igor
Reviewed By: igor
Subscribers: leveldb
Differential Revision: https://reviews.facebook.net/D24903
Summary: as title
Test Plan:
unit test
I am only able to build the test case for hard_rate_limit.
soft_rate_limit is essentially the same thing as hard_rate_limit
Reviewers: igor, sdong, yhchiang
Reviewed By: yhchiang
Subscribers: leveldb
Differential Revision: https://reviews.facebook.net/D24759
Summary: Add more tests as well
Test Plan: unit test
Reviewers: igor, sdong, yhchiang
Reviewed By: sdong
Subscribers: leveldb
Differential Revision: https://reviews.facebook.net/D24747
Summary: as title
Test Plan: unit test
Reviewers: sdong, yhchiang, igor
Reviewed By: igor
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D24729
Summary:
Fixed the following compile error on Mac.
db/db_test.cc:8618:52: error: C++11 forbids default arguments for lambda expressions [-Werror,-Wlambda-extensions]
auto gen_l0_kb = [this](int start, int size, int stride = 1) {
^ ~
1 error generated.
Test Plan:
db_test
Summary: option max_background_flushes doesn't make sense if thread pool size is not set accordingly. Set the thread pool size as what we do for max_background_compactions.
Test Plan: Run db_bench with max_background_flushes > 1
Reviewers: yhchiang, igor, rven, ljin
Reviewed By: ljin
Subscribers: MarkCallaghan, leveldb
Differential Revision: https://reviews.facebook.net/D24717
Summary:
This diff introduces the `lookahead` argument to `SkipListFactory()`. This is an
optimization for the tailing use case which includes many seeks. E.g. consider
the following operations on a skip list iterator:
Seek(x), Next(), Next(), Seek(x+2), Next(), Seek(x+3), Next(), Next(), ...
If `lookahead` is positive, `SkipListRep` will return an iterator which also
keeps track of the previously visited node. Seek() then first does a linear
search starting from that node (up to `lookahead` steps). As in the tailing
example above, this may require fewer than ~log(n) comparisons as with regular
skip list search.
Test Plan:
Added a new benchmark (`fillseekseq`) which simulates the usage pattern. It
first writes N records (with consecutive keys), then measures how much time it
takes to read them by calling `Seek()` and `Next()`.
$ time ./db_bench -num 10000000 -benchmarks fillseekseq -prefix_size 1 \
-key_size 8 -write_buffer_size $[1024*1024*1024] -value_size 50 \
-seekseq_next 2 -skip_list_lookahead=0
[...]
DB path: [/dev/shm/rocksdbtest/dbbench]
fillseekseq : 0.389 micros/op 2569047 ops/sec;
real 0m21.806s
user 0m12.106s
sys 0m9.672s
$ time ./db_bench [...] -skip_list_lookahead=2
[...]
DB path: [/dev/shm/rocksdbtest/dbbench]
fillseekseq : 0.153 micros/op 6540684 ops/sec;
real 0m19.469s
user 0m10.192s
sys 0m9.252s
Reviewers: ljin, sdong, igor
Reviewed By: igor
Subscribers: dhruba, leveldb, march, lovro
Differential Revision: https://reviews.facebook.net/D23997
Summary: This will be much easier than reviewing git sha's we currently have in our LOGs
Test Plan: none
Reviewers: sdong, yhchiang, ljin
Reviewed By: ljin
Subscribers: leveldb
Differential Revision: https://reviews.facebook.net/D24591
Summary:
The test only covers changing write_buffer_size. Other changable
parameters such bloom bits/probes are not obvious how to test.
Suggestions are welcome
Test Plan: db_test
Reviewers: sdong, yhchiang, igor
Reviewed By: igor
Subscribers: leveldb
Differential Revision: https://reviews.facebook.net/D24429
Summary:
Fix a check in database shutdown or Column family drop during flush.
Special thanks to Maurice Barnum who spots the problem :)
Test Plan: db_test
Reviewers: ljin, igor, sdong
Reviewed By: sdong
Subscribers: leveldb
Differential Revision: https://reviews.facebook.net/D24273
Summary:
Add two stats to compaction summary:
1. Total input records from previous level
2. Total number of records dropped after compaction
Test Plan: See outputs of printing when runnning locally
Reviewers: ljin, igor, MarkCallaghan
Reviewed By: MarkCallaghan
Subscribers: leveldb
Differential Revision: https://reviews.facebook.net/D24411
Summary: perf_context.get_from_output_files_time is now only set writable DB's DB::Get(). Extend it to MultiGet() and read only DB.
Test Plan:
make all check
Fix perf_context_test and extend it to cover MultiGet(), as long as read-only DB. Run it and watch the results
Reviewers: ljin, yhchiang, igor
Reviewed By: igor
Subscribers: rven, leveldb
Differential Revision: https://reviews.facebook.net/D24207
Summary:
Fixed signed-unsigned comparison warning in db_test.cc
db/db_test.cc:8606:3: note: in instantiation of function template specialization 'rocksdb::test::Tester::IsEq<int, unsigned long>' requested here
ASSERT_EQ(2, metadata.size());
^
Test Plan:
make db_test