Summary:
- Remove default arch-specified flags.
- Move non-default arch-specific flags to arch-specific param.

Reviewed By: yiwu-arbug

Differential Revision: D5597499

fbshipit-source-id: c53108ac39c73ac36893d3fd9aaf3b5e3080f1ae

Summary: Closes https://github.com/facebook/rocksdb/pull/2730

Differential Revision: D5619256

Pulled By: ajkr

fbshipit-source-id: c80d697eeceab91964259132e58f5cd2219efb93

Summary:
`KeyNotExistsBeyondOutputLevel` didn't consider L0 files' key-ranges. So if a key only was covered by older L0 files' key-ranges, we would incorrectly drop deletions of that key. This PR just skips the deletion-dropping optimization when output level is L0.
Closes https://github.com/facebook/rocksdb/pull/2726

Differential Revision: D5617286

Pulled By: ajkr

fbshipit-source-id: 4bff1396b06d49a828ba4542f249191052915bce

Summary:
- like other subcommands, reporting compression sizes should be specified with the `--command` CLI arg.
- also added `--compression_types` arg as it's useful to restrict the types of compression used, at least in my dictionary compression experiments.
Closes https://github.com/facebook/rocksdb/pull/2706

Differential Revision: D5589520

Pulled By: ajkr

fbshipit-source-id: 305bb4ebcc95eecc8a85523cd3b1050619c9ddc5

Summary:
Previously we could only select the CF on which to operate uniformly at random. This is a limitation, e.g., when testing universal compaction as all CFs would need to run full compaction at roughly the same time, which isn't realistic.

This PR allows the user to specify the probability distribution for selecting CFs via the `--column_family_distribution` argument.
Closes https://github.com/facebook/rocksdb/pull/2677

Differential Revision: D5544436

Pulled By: ajkr

fbshipit-source-id: 478d56260995236ae90895ce5bd51f38882e185a

Summary:
sounds like we're willing to tradeoff minor inaccuracy in stats for speed. start with histogram stats. ticker stats will be harder (and, IMO, we shouldn't change them in this manner) as many test cases rely on them being exactly correct.
Closes https://github.com/facebook/rocksdb/pull/2720

Differential Revision: D5607884

Pulled By: ajkr

fbshipit-source-id: 1b754cda35ea6b252d1fdd5aa3cfb58866506372

Summary:
Fix c_test missing deletion of write batch pointer.
Closes https://github.com/facebook/rocksdb/pull/2725

Differential Revision: D5613866

Pulled By: yiwu-arbug

fbshipit-source-id: bf3f59a6812178577c9c25bae558ef36414a1f51

Summary:
While GC, blob DB use optimistic transaction to delete or replace the index entry in LSM, to guarantee correctness if there's a normal write writing to the same key. However, the previous implementation doesn't call SetSnapshot() nor use GetForUpdate() of transaction API, instead it do its own sequence number checking before beginning the transaction. A normal write can sneak in after the sequence number check and overwrite the key, and the GC will delete or relocate the old version of the key by mistake. Update the code to property use GetForUpdate() to check the existing index entry.

After the patch the sequence number store with each blob record is useless, So I'm considering remove the sequence number from blob record, in another patch.
Closes https://github.com/facebook/rocksdb/pull/2703

Differential Revision: D5589178

Pulled By: yiwu-arbug

fbshipit-source-id: 8dc960cd5f4e61b36024ba7c32d05584ce149c24

Summary: Closes https://github.com/facebook/rocksdb/pull/2724

Differential Revision: D5613416

Pulled By: ajkr

fbshipit-source-id: ed55fb66ab1b41dfdfe765fe3264a1c87a8acb00

Summary:
It would be super helpful to not have to recompile rocksdb to get this performance tweak for mechanical disks.

I have signed the CLA.
Closes https://github.com/facebook/rocksdb/pull/2718

Differential Revision: D5606994

Pulled By: yiwu-arbug

fbshipit-source-id: c05e92bad0d03bd38211af1e1ced0d0d1e02f634

Summary:
Right now, if direct I/O is enabled, prefetching the last 512KB cannot be applied, except compaction inputs or readahead is enabled for iterators. This can create a lot of I/O for HDD cases. To solve the problem, the 512KB is prefetched in block based table if direct I/O is enabled. The prefetched buffer is passed in totegher with random access file reader, so that we try to read from the buffer before reading from the file. This can be extended in the future to support flexible user iterator readahead too.
Closes https://github.com/facebook/rocksdb/pull/2708

Differential Revision: D5593091

Pulled By: siying

fbshipit-source-id: ee36ff6d8af11c312a2622272b21957a7b5c81e7

Summary:
This reverting #2699 to fix clang build.
Closes https://github.com/facebook/rocksdb/pull/2723

Differential Revision: D5610207

Pulled By: yiwu-arbug

fbshipit-source-id: 6857f4556d6d18f17b74cf81fa936d1dc0bd364c

Summary:
TSAN shows error when we grab too many locks at the same time. In TSAN crash test, make one shard key cover 2^22 keys so that no many keys will be hold at the same time.
Closes https://github.com/facebook/rocksdb/pull/2719

Differential Revision: D5609035

Pulled By: siying

fbshipit-source-id: 930e5d63fff92dbc193dc154c4c615efbdf06c6a

Summary:
(#2564)
Closes https://github.com/facebook/rocksdb/pull/2669

Differential Revision: D5594151

Pulled By: yiwu-arbug

fbshipit-source-id: 67ae9446342f3323d6ecad8e811f4158da194270

Summary:
Error was:

utilities/persistent_cache/block_cache_tier.cc: In instantiation of ‘void rocksdb::Add(std::map<std::__cxx11::basic_string<char>, double>*, const string&, const T&) [with T = double; std::__cxx11::string = std::__cxx11::basic_string<char>]’:
utilities/persistent_cache/block_cache_tier.cc:147:40:   required from here
utilities/persistent_cache/block_cache_tier.cc:141:23: error: type qualifiers ignored on cast result type [-Werror=ignored-qualifiers]

   stats->insert({key, static_cast<const double>(t)});

fixing like #2562
Closes https://github.com/facebook/rocksdb/pull/2603

Differential Revision: D5600910

Pulled By: yiwu-arbug

fbshipit-source-id: 891a5ec7e451d2dec6ad1b6b7fac545657f87363

Summary: Closes https://github.com/facebook/rocksdb/pull/2655

Differential Revision: D5600858

Pulled By: yiwu-arbug

fbshipit-source-id: cf52f9104e348438bf168dc6bf7af3837faf12ef

Summary:
- moved the max call for numeric limits into paranthesis so that max wont be called as macro when including <Windows.h>
Closes https://github.com/facebook/rocksdb/pull/2709

Differential Revision: D5600773

Pulled By: yiwu-arbug

fbshipit-source-id: fd28b6f7c10ddce21bad4030f2db06f965bb08da

Summary:
Signed-off-by: Ntang.jin <tang.jin@istuary.com>
Closes https://github.com/facebook/rocksdb/pull/2644

Differential Revision: D5600861

Pulled By: yiwu-arbug

fbshipit-source-id: 9516636cb6e77b09fe0ebef78953adf4b7e88cc8

Summary:
Commit 4f81ab38 has the test wrong.

clang doesn't support a -dumpversion option. By lucky coincidence
clang/gcc --version both place a version number at the same output location
when --verison is passed.

Example output (1st line only).

    $ clang --version
    clang version 3.9.1 (tags/RELEASE_391/final)

    $ gcc --version
    gcc (GCC) 6.4.1 20170727 (Red Hat 6.4.1-1)

During the test of the compiler we ensure that a minimum version is met
as Makefile doesn't support patterns.

Also xcode9 doesn't seem affected by https://github.com/facebook/rocksdb/issues/2672
and also doesn't have "clang" as the first part of its output so the
fix implemented here also is Apple clang friendly.

    $ clang --version
    Apple LLVM version 9.0.0 (clang-900.0.31)
Signed-off-by: NDaniel Black <daniel.black@au.ibm.com>
Closes https://github.com/facebook/rocksdb/pull/2699

Differential Revision: D5600818

Pulled By: yiwu-arbug

fbshipit-source-id: 3b0f2751becb53c1c35468bf29f3f828e7cf2c2a

Summary:
We need a tool to check any sst file corruption in the db.
It will check all the sst files in current version and read all the blocks (data, meta, index) with checksum verification. If any verification fails, the function will return non-OK status.
Closes https://github.com/facebook/rocksdb/pull/2498

Differential Revision: D5324269

Pulled By: lightmark

fbshipit-source-id: 6f8a272008b722402a772acfc804524c9d1a483b

Summary: Closes https://github.com/facebook/rocksdb/pull/2702

Differential Revision: D5585389

Pulled By: ajkr

fbshipit-source-id: cb54041eb481d0d759c440f82a8a2c5b34534173

Summary:
We should search wal_dir in Repairer::FindFiles function, and avoid use
LogFileNmae(dbname, number) to get WAL file's name, which will get a wrong
WAL filename. as following:

```
[WARN] [/home/liuchang/Workspace/rocksdb/db/repair.cc:310] Log #3: ignoring conversion error: IO error: While opening a file for sequentially reading: /tmp/rocksdbtest-1000/repair_test/000003.log: No such file or directory
```
  I have added a new test case to repair_test.cc, which try to repair db with all WAL options.
Signed-off-by: NChang Liu <liuchang0812@gmail.com>
Closes https://github.com/facebook/rocksdb/pull/2692

Differential Revision: D5575888

Pulled By: ajkr

fbshipit-source-id: 5b93e9f85cddc01663ccecd87631fa723ac466a3

Summary:
Markup i386 code paths resolving compilation
failure under i386 with gcc-7.
Signed-off-by: NJames Page <james.page@ubuntu.com>
Closes https://github.com/facebook/rocksdb/pull/2700

Differential Revision: D5583047

Pulled By: maysamyabandeh

fbshipit-source-id: fe31bcfeaf7cd2d3f51b55f5ae0b3b0cb3788fbc

Summary:
This patch splits Commit and Prepare into lock-related logic and db-write-related logic. It moves lock-related logic to PessimisticTransaction to be reused by all children classes and movies the existing impl of db-write-related to PrepareInternal, CommitSingleInternal, and CommitInternal in WriteCommittedTxnImpl.
Closes https://github.com/facebook/rocksdb/pull/2691

Differential Revision: D5569464

Pulled By: maysamyabandeh

fbshipit-source-id: d1b8698e69801a4126c7bc211745d05c636f5325

Summary: Closes https://github.com/facebook/rocksdb/pull/2697

Differential Revision: D5573894

Pulled By: maysamyabandeh

fbshipit-source-id: 8fc03ea8ea6f3f3bc0f68b64cf90243a70562dc4

Summary:
I'd like to propose adding my company as a RocksDB user
Closes https://github.com/facebook/rocksdb/pull/2694

Differential Revision: D5572113

Pulled By: ajkr

fbshipit-source-id: 646143b955e3efddee56691cce912d7badaa6e8b

Summary:
This opens space for the new implementations of TransactionDBImpl such as WritePreparedTxnDBImpl that has a different policy of how to write to DB.
Closes https://github.com/facebook/rocksdb/pull/2689

Differential Revision: D5568918

Pulled By: maysamyabandeh

fbshipit-source-id: f7eac866e175daf3793ae79da108f65cc7dc7b25

Summary:
In the condition:
```
if (range_del_agg != nullptr &&
    range_del_agg->ShouldDelete(
        iter->key(),
        RangeDelAggregator::RangePositioningMode::kForwardTraversal) &&
    filter != CompactionFilter::Decision::kRemoveAndSkipUntil) {
...
}
```
it could be possible that all the work done in `range_del_agg->ShouldDelete` is wasted due to not having the right `filter` value later on.
Instead, check `filter` value before even calling `range_del_agg->ShouldDelete`, which is a much more involved function.
Closes https://github.com/facebook/rocksdb/pull/2690

Differential Revision: D5568931

Pulled By: sagar0

fbshipit-source-id: 17512d52360425c7ae9de7675383f5d7bc3dad58

Summary:
The FsyncFiles background job call Fsync() periodically for blob files. However it can access WritableFileWriter concurrently with a Put() or Write(). And WritableFileWriter does not support concurrent access. It will lead to WritableFileWriter buffer being flush with same content twice, and blob file end up corrupted. Fixing by simply let FsyncFiles hold write_mutex_.
Closes https://github.com/facebook/rocksdb/pull/2685

Differential Revision: D5561908

Pulled By: yiwu-arbug

fbshipit-source-id: f0bb5bcab0e05694e053b8c49eab43640721e872

Summary:
fixes #2555
Closes https://github.com/facebook/rocksdb/pull/2684

Differential Revision: D5560527

Pulled By: maysamyabandeh

fbshipit-source-id: 6e1d874ae0b4e699a77203d9d52d0bb8f59013b0

Summary:
The background thread pools' sizes weren't easily configurable by `max_background_compactions` and `max_background_flushes` in multi-instance setups. Introduced separate arguments for their sizes.
Closes https://github.com/facebook/rocksdb/pull/2680

Differential Revision: D5550675

Pulled By: ajkr

fbshipit-source-id: bab5f0a7bc5db63bb084d0c10facbe437096367d

Summary:
maysamyabandeh IslamAbdelRahman PTAL

Fix https://github.com/facebook/rocksdb/issues/2672Signed-off-by: NCholerae Hu <huyingqian@pingcap.com>
Closes https://github.com/facebook/rocksdb/pull/2681

Differential Revision: D5561515

Pulled By: ajkr

fbshipit-source-id: 676187802ebd8a87a6c051bb565818a1bf89d0a9

Summary:
The current blob db implementation use mix of int32_t, uint32_t and uint64_t for TTL and expiration. Update all timestamps to uint64_t for consistency.
Closes https://github.com/facebook/rocksdb/pull/2683

Differential Revision: D5557103

Pulled By: yiwu-arbug

fbshipit-source-id: e4eab2691629a755e614e8cf1eed9c3a681d0c42

Summary:
"/bin/bash" is a Linuxism.  "/usr/bin/env bash" is portable.
Closes https://github.com/facebook/rocksdb/pull/2646

Differential Revision: D5556259

Pulled By: ajkr

fbshipit-source-id: cbffd38ecdbfffb2438969ec007ab345ed893ccb

Summary:
When we had a single thread pool for compactions, a thread could be busy for a long time (minutes) executing a compaction involving the bottom level. In multi-instance setups, the entire thread pool could be consumed by such bottom-level compactions. Then, top-level compactions (e.g., a few L0 files) would be blocked for a long time ("head-of-line blocking"). Such top-level compactions are critical to prevent compaction stalls as they can quickly reduce number of L0 files / sorted runs.

This diff introduces a bottom-priority queue for universal compactions including the bottom level. This alleviates the head-of-line blocking situation for fast, top-level compactions.

- Added `Env::Priority::BOTTOM` thread pool. This feature is only enabled if user explicitly configures it to have a positive number of threads.
- Changed `ThreadPoolImpl`'s default thread limit from one to zero. This change is invisible to users as we call `IncBackgroundThreadsIfNeeded` on the low-pri/high-pri pools during `DB::Open` with values of at least one. It is necessary, though, for bottom-pri to start with zero threads so the feature is disabled by default.
- Separated `ManualCompaction` into two parts in `PrepickedCompaction`. `PrepickedCompaction` is used for any compaction that's picked outside of its execution thread, either manual or automatic.
- Forward universal compactions involving last level to the bottom pool (worker thread's entry point is `BGWorkBottomCompaction`).
- Track `bg_bottom_compaction_scheduled_` so we can wait for bottom-level compactions to finish. We don't count them against the background jobs limits. So users of this feature will get an extra compaction for free.
Closes https://github.com/facebook/rocksdb/pull/2580

Differential Revision: D5422916

Pulled By: ajkr

fbshipit-source-id: a74bd11f1ea4933df3739b16808bb21fcd512333

Summary:
I'm going with brute-force solution, just letting Put() and Write() holding a mutex before writing. May improve concurrent writing with finer granularity locking later.
Closes https://github.com/facebook/rocksdb/pull/2682

Differential Revision: D5552690

Pulled By: yiwu-arbug

fbshipit-source-id: 039abd675b5d274a7af6428198d1733cafecef4c

Summary:
Fix the bug where if blob db garbage collection revmoe keys with newer version. It shouldn't delete the key from base db when sequence number in base db is not equal to the one in blob log.
Closes https://github.com/facebook/rocksdb/pull/2678

Differential Revision: D5549752

Pulled By: yiwu-arbug

fbshipit-source-id: abb8649260963b5c389748023970fd746279d227

Summary:
https://github.com/facebook/rocksdb/issues/2559 reports an overflow in AwaitState. nbronson has debugged the issue and presented the fix, which is applied to this patch. Moreover this patch adds more comments to clarify the logic in AwaitState.

I tried with both 16 and 64 threads on update benchmark. The fix lowers cpu usage by 1.6 but also lowers the throughput by 1.6 and 2% respectively. Apparently the bug had favored using the spinning more often.

Benchmarks:
TEST_TMPDIR=/dev/shm/tmpdb time ./db_bench --benchmarks="fillrandom" --threads=16 --num=2000000
TEST_TMPDIR=/dev/shm/tmpdb time ./db_bench --use_existing_db=1 --benchmarks="updaterandom[X3]" --threads=16 --num=2000000
TEST_TMPDIR=/dev/shm/tmpdb time ./db_bench --use_existing_db=1 --benchmarks="updaterandom[X3]" --threads=64 --num=200000

Results
$ cat update-16t-bug.txt | tail -4
updaterandom [AVG    3 runs] : 234117 ops/sec;   51.8 MB/sec
updaterandom [MEDIAN 3 runs] : 233581 ops/sec;   51.7 MB/sec
3896.42user 1539.12system 6:50.61elapsed 1323%CPU (0avgtext+0avgdata 331308maxresident)k
0inputs+0outputs (0major+1281001minor)pagefaults 0swaps
$ cat update-16t-fixed.txt | tail -4
updaterandom [AVG    3 runs] : 230364 ops/sec;   51.0 MB/sec
updaterandom [MEDIAN 3 runs] : 226169 ops/sec;   50.0 MB/sec
3865.46user 1568.32system 6:57.63elapsed 1301%CPU (0avgtext+0avgdata 315012maxresident)k
0inputs+0outputs (0major+1342568minor)pagefaults 0swaps

$ cat update-64t-bug.txt | tail -4
updaterandom [AVG    3 runs] : 261878 ops/sec;   57.9 MB/sec
updaterandom [MEDIAN 3 runs] : 262859 ops/sec;   58.2 MB/sec
926.27user 578.06system 2:27.46elapsed 1020%CPU (0avgtext+0avgdata 475480maxresident)k
0inputs+0outputs (0major+1058728minor)pagefaults 0swaps
$ cat update-64t-fixed.txt | tail -4
updaterandom [AVG    3 runs] : 256699 ops/sec;   56.8 MB/sec
updaterandom [MEDIAN 3 runs] : 256380 ops/sec;   56.7 MB/sec
933.47user 575.37system 2:30.41elapsed 1003%CPU (0avgtext+0avgdata 482340maxresident)k
0inputs+0outputs (0major+1078557minor)pagefaults 0swaps
Closes https://github.com/facebook/rocksdb/pull/2679

Differential Revision: D5553732

Pulled By: maysamyabandeh

fbshipit-source-id: 98b72dc3a8e0f22ea29d4f7c7790af10c369c5bb

Summary:
This patch refactors TransactionImpl by separating the logic for pessimistic concurrency control from the implementation of how to write the data to rocksdb. The existing implementation is named WriteCommittedTxnImpl as it writes committed data to the db. A template named WritePreparedTxnImpl is also added which will be later completed to provide a an alternative implementation.
Closes https://github.com/facebook/rocksdb/pull/2676

Differential Revision: D5549998

Pulled By: maysamyabandeh

fbshipit-source-id: 16298e86b43ca4849324c1f35c731913c6d17bec

Summary:
Move an option necessary for running db_bench on multiple CFs into the general initialization area, so it works with both flag-based init and OPTIONS-based init.
Closes https://github.com/facebook/rocksdb/pull/2675

Differential Revision: D5541378

Pulled By: ajkr

fbshipit-source-id: 169926cb4ae95c17974f744faf7cc794d41e5c0a