Skip to content

R
rocksdb

kvdb / rocksdb
12 个月 前同步成功

通知 0
Star 0
Fork 0
R
rocksdb

体验新版 GitCode，发现更多精彩内容 >>

切换分支/标签
  1. 13 1月, 2017 1 次提交
  3. 12 1月, 2017 2 次提交
    • A
      direct reads refactor · dc2584ee
      由 Aaron Gao 提交于 
      Summary:
direct IO reads refactoring
remove unnecessary classes and unified interfaces
tested with db_bench

need more change for options and ON/OFF for different files.
Since disabled is default, it should be fine now
Closes https://github.com/facebook/rocksdb/pull/1636

Differential Revision: D4307189

Pulled By: lightmark

fbshipit-source-id: 6991e22
      dc2584ee
    • A
      Guarding extra fallocate call with TRAVIS because its not working pro… · 62384ebe
      由 Anirban Rahut 提交于 
      Summary:
…perly on travis

 There is some old code in PosixWritableFile::Close(), which
truncates the file to the measured size and then does an extra fallocate
with KEEP_SIZE. This is commented as a failsafe because in some
cases ftruncate doesn't do the right job (I don't know of an instance of
this btw). However doing an fallocate with KEEP_SIZE should not increase
the file size. However on Travis Worker which is Docker (likely AUFS )
its not working. There are comments on web that show that the AUFS
author had initially not implemented fallocate, and then did it later.
So not sure what is the quality of the implementation.
Closes https://github.com/facebook/rocksdb/pull/1765

Differential Revision: D4401340

Pulled By: anirbanr-fb

fbshipit-source-id: e2d8100
      62384ebe
  5. 11 1月, 2017 1 次提交
    • A
      Allow incrementing refcount on cache handles · fe395fb6
      由 Andrew Kryczka 提交于 
      Summary:
Previously the only way to increment a handle's refcount was to invoke Lookup(), which (1) did hash table lookup to get cache handle, (2) incremented that handle's refcount. For a future DeleteRange optimization, I added a function, Ref(), for when the caller already has a cache handle and only needs to do (2).
Closes https://github.com/facebook/rocksdb/pull/1761

Differential Revision: D4397114

Pulled By: ajkr

fbshipit-source-id: 9addbe5
      fe395fb6
  7. 10 1月, 2017 1 次提交
    • D
      Fix Windows environment issues · 3c233ca4
      由 Dmitri Smirnov 提交于 
      Summary:
Enable directIO on WritableFileImpl::Append
     with offset being current length of the file.
     Enable UniqueID tests on Windows, disable others but
     leeting them to compile. Unique tests are valuable to
     detect failures on different filesystems and upcoming
     ReFS.
     Clear output in WinEnv Getchildren.This is different from
     previous strategy, do not touch output on failure.
     Make sure DBTest.OpenWhenOpen works with windows error message
Closes https://github.com/facebook/rocksdb/pull/1746

Differential Revision: D4385681

Pulled By: IslamAbdelRahman

fbshipit-source-id: c07b702
      3c233ca4
  9. 09 1月, 2017 2 次提交
    • M
      Revert "PinnableSlice" · d0ba8ec8
      由 Maysam Yabandeh 提交于 
      Summary:
This reverts commit 54d94e9c.

The pull request was landed by mistake.
Closes https://github.com/facebook/rocksdb/pull/1755

Differential Revision: D4391678

Pulled By: maysamyabandeh

fbshipit-source-id: 36d5149
      d0ba8ec8
    • M
      PinnableSlice · 54d94e9c
      由 Maysam Yabandeh 提交于 
      Summary:
Currently the point lookup values are copied to a string provided by the user.
This incures an extra memcpy cost. This patch allows doing point lookup
via a PinnableSlice which pins the source memory location (instead of
copying their content) and releases them after the content is consumed
by the user. The old API of Get(string) is translated to the new API
underneath.

 Here is the summary for improvements:
 1. value 100 byte: 1.8%  regular, 1.2% merge values
 2. value 1k   byte: 11.5% regular, 7.5% merge values
 3. value 10k byte: 26% regular,    29.9% merge values

 The improvement for merge could be more if we extend this approach to
 pin the merge output and delay the full merge operation until the user
 actually needs it. We have put that for future work.

PS:
Sometimes we observe a small decrease in performance when switching from
t5452014 to this patch but with the old Get(string) API. The difference
is a little and could be noise. More importantly it is safely
cancelled
Closes https://github.com/facebook/rocksdb/pull/1732

Differential Revision: D4374613

Pulled By: maysamyabandeh

fbshipit-source-id: a077f1a
      54d94e9c
  11. 07 1月, 2017 1 次提交
  13. 04 1月, 2017 1 次提交
  15. 29 12月, 2016 1 次提交
    • S
      Always fsync the file after file copying · 17a4b75c
      由 Siying Dong 提交于 
      Summary:
File copying happens when creating checkpoints and bulkloading files from different FS partition. We should fsync the files when copying them to guarantee durability. A side effect will be that the dirty pages in file system buffers won't grow too large.
Closes https://github.com/facebook/rocksdb/pull/1728

Differential Revision: D4371083

Pulled By: siying

fbshipit-source-id: 579e14c
      17a4b75c
  17. 23 12月, 2016 2 次提交
    • Y
      Print cache options to info log · ab48c165
      由 Yi Wu 提交于 
      Summary:
Improve cache options logging to info log.
Also print the value of
cache_index_and_filter_blocks_with_high_priority.
Closes https://github.com/facebook/rocksdb/pull/1709

Differential Revision: D4358776

Pulled By: yiwu-arbug

fbshipit-source-id: 8f030a0
      ab48c165
    • A
      direct io write support · 972f96b3
      由 Aaron Gao 提交于 
      Summary:
rocksdb direct io support

```
[gzh@dev11575.prn2 ~/rocksdb] ./db_bench -benchmarks=fillseq --num=1000000
Initializing RocksDB Options from the specified file
Initializing RocksDB Options from command-line flags
RocksDB:    version 5.0
Date:       Wed Nov 23 13:17:43 2016
CPU:        40 * Intel(R) Xeon(R) CPU E5-2660 v2 @ 2.20GHz
CPUCache:   25600 KB
Keys:       16 bytes each
Values:     100 bytes each (50 bytes after compression)
Entries:    1000000
Prefix:    0 bytes
Keys per prefix:    0
RawSize:    110.6 MB (estimated)
FileSize:   62.9 MB (estimated)
Write rate: 0 bytes/second
Compression: Snappy
Memtablerep: skip_list
Perf Level: 1
WARNING: Assertions are enabled; benchmarks unnecessarily slow
------------------------------------------------
Initializing RocksDB Options from the specified file
Initializing RocksDB Options from command-line flags
DB path: [/tmp/rocksdbtest-112628/dbbench]
fillseq      :       4.393 micros/op 227639 ops/sec;   25.2 MB/s

[gzh@dev11575.prn2 ~/roc
Closes https://github.com/facebook/rocksdb/pull/1564

Differential Revision: D4241093

Pulled By: lightmark

fbshipit-source-id: 98c29e3
      972f96b3
  19. 22 12月, 2016 1 次提交
  21. 17 12月, 2016 2 次提交
  23. 15 12月, 2016 3 次提交
  25. 14 12月, 2016 3 次提交
    • D
      util/logging.cc: buffer of insufficient size (gcc-7 -Werror=format-length) · e097222e
      由 Daniel Black 提交于 
      Summary:
util/logging.cc:100:13: error: output may be truncated before the last format character [-Werror=format-length=]
 std::string NumberToHumanString(int64_t num) {
             ^~~~~~~~~~~~~~~~~~~
util/logging.cc:106:59: note: format output between 3 and 19 bytes into a destination of size 16
     snprintf(buf, sizeof(buf), "%" PRIi64 "K", num / 1000);
Closes https://github.com/facebook/rocksdb/pull/1653

Differential Revision: D4318687

Pulled By: yiwu-arbug

fbshipit-source-id: 3a5c931
      e097222e
    • D
      Gcc 7 error expansion to defined · bfbcec23
      由 Daniel Black 提交于 
      Summary:
sorry if these gcc-7/clang-4 cleanups are getting tedious.
Closes https://github.com/facebook/rocksdb/pull/1658

Differential Revision: D4318792

Pulled By: yiwu-arbug

fbshipit-source-id: 8e85891
      bfbcec23
    • D
      util/histogram.cc: HistogramStat::toString buffer insufficient · c3e5ee71
      由 Daniel Black 提交于 
      Summary:
Increased buffer size to 1650.

util/histogram.cc: In member function 'std::__cxx11::string rocksdb::HistogramStat::ToString() const':
util/histogram.cc:189:13: error: '%.2f' directive output truncated writing between 4 and 313 bytes into a region of size 0 [-Werror=format-length=]
 std::string HistogramStat::ToString() const {
             ^~~~~~~~~~~~~
util/histogram.cc:205:30: note: format output between 69 and 1614 bytes into a destination of size 200
            Percentile(99.99));
                              ^
cc1plus: all warnings being treated as errors
Makefile:1521: recipe for target 'util/histogram.o' failed
Closes https://github.com/facebook/rocksdb/pull/1660

Differential Revision: D4318820

Pulled By: yiwu-arbug

fbshipit-source-id: 45ae6ea
      c3e5ee71
  27. 13 12月, 2016 1 次提交
    • A
      Return finer-granularity status from Env::GetChildren* · f0c509e2
      由 Andrew Kryczka 提交于 
      Summary:
It'd be nice to use the error status type to distinguish
between user error and system error. For example, GetChildren can fail
listing a backup directory's contents either because a bad path was provided
(user error) or because an operation failed, e.g., a remote storage service
call failed (system error). In the former case, we want to continue and treat
the backup directory as empty; in the latter case, we want to immediately
propagate the error to the caller.

This diff uses NotFound to indicate user error and IOError to indicate
system error. Previously IOError indicated both.
Closes https://github.com/facebook/rocksdb/pull/1644

Differential Revision: D4312157

Pulled By: ajkr

fbshipit-source-id: 51b4f24
      f0c509e2
  29. 06 12月, 2016 2 次提交
    • M
      Implement non-exclusive locks · 2005c88a
      由 Manuel Ung 提交于 
      Summary:
This is an implementation of non-exclusive locks for pessimistic transactions. It is relatively simple and does not prevent starvation (ie. it's possible that request for exclusive access will never be granted if there are always threads holding shared access). It is done by changing `KeyLockInfo` to hold an set a transaction ids, instead of just one, and adding a flag specifying whether this lock is currently held with exclusive access or not.

Some implementation notes:
- Some lock diagnostic functions had to be updated to return a set of transaction ids for a given lock, eg. `GetWaitingTxn` and `GetLockStatusData`.
- Deadlock detection is a bit more complicated since a transaction can now wait on multiple other transactions. A BFS is done in this case, and deadlock detection depth is now just a limit on the number of transactions we visit.
- Expirable transactions do not work efficiently with shared locks at the moment, but that's okay for now.
Closes https://github.com/facebook/rocksdb/pull/1573

Differential Revision: D4239097

Pulled By: lth

fbshipit-source-id: da7c074
      2005c88a
    • A
      Made delete_obsolete_files_period_micros option dynamic · 9053fe2a
      由 Anton Safonov 提交于 
      Summary:
Made delete_obsolete_files_period_micros option dynamic. It can be updating using DB::SetDBOptions().
Closes https://github.com/facebook/rocksdb/pull/1595

Differential Revision: D4246569

Pulled By: tonek

fbshipit-source-id: d23f560
      9053fe2a
  31. 02 12月, 2016 2 次提交
    • I
      Cache heap::downheap() root comparison (optimize heap cmp call) · 4a21b140
      由 Islam AbdelRahman 提交于 
      Summary:
Reduce number of comparisons in heap by caching which child node in the first level is smallest (left_child or right_child)
So next time we can compare directly against the smallest child

I see that the total number of calls to comparator drops significantly when using this optimization

Before caching (~2mil key comparison for iterating the DB)
```
$ DEBUG_LEVEL=0 make db_bench -j64 && ./db_bench --benchmarks="readseq" --db="/dev/shm/heap_opt" --use_existing_db --disable_auto_compactions --cache_size=1000000000  --perf_level=2
readseq      :       0.338 micros/op 2959201 ops/sec;  327.4 MB/s user_key_comparison_count = 2000008
```
After caching (~1mil key comparison for iterating the DB)
```
$ DEBUG_LEVEL=0 make db_bench -j64 && ./db_bench --benchmarks="readseq" --db="/dev/shm/heap_opt" --use_existing_db --disable_auto_compactions --cache_size=1000000000 --perf_level=2
readseq      :       0.309 micros/op 3236801 ops/sec;  358.1 MB/s user_key_comparison_count = 1000011
```

It also improves
Closes https://github.com/facebook/rocksdb/pull/1600

Differential Revision: D4256027

Pulled By: IslamAbdelRahman

fbshipit-source-id: 76fcc66
      4a21b140
    • I
      Fix travis (compile for clang < 3.9) · e39d0808
      由 Islam AbdelRahman 提交于 
      Summary:
Travis fail because it uses clang 3.6 which don't recognize
`__attribute__((__no_sanitize__("undefined")))`
Closes https://github.com/facebook/rocksdb/pull/1601

Differential Revision: D4257175

Pulled By: IslamAbdelRahman

fbshipit-source-id: fb4d1ab
      e39d0808
  33. 29 11月, 2016 3 次提交
  35. 24 11月, 2016 1 次提交
    • S
      Improve Write Stalling System · cd7c4143
      由 Siying Dong 提交于 
      Summary:
Current write stalling system has the problem of lacking of positive feedback if the restricted rate is already too low. Users sometimes stack in very low slowdown value. With the diff, we add a positive feedback (increasing the slowdown value) if we recover from slowdown state back to normal. To avoid the positive feedback to keep the slowdown value to be to high, we add issue a negative feedback every time we are close to the stop condition. Experiments show it is easier to reach a relative balance than before.

Also increase level0_stop_writes_trigger default from 24 to 32. Since level0_slowdown_writes_trigger default is 20, stop trigger 24 only gives four files as the buffer time to slowdown writes. In order to avoid stop in four files while 20 files have been accumulated, the slowdown value must be very low, which is amost the same as stop. It also doesn't give enough time for the slowdown value to converge. Increase it to 32 will smooth out the system.
Closes https://github.com/facebook/rocksdb/pull/1562

Differential Revision: D4218519

Pulled By: siying

fbshipit-source-id: 95e4088
      cd7c4143
  37. 22 11月, 2016 3 次提交
  39. 21 11月, 2016 1 次提交
    • C
      Fix deadlock when calling getMergedHistogram · a0deec96
      由 Changli Gao 提交于 
      Summary:
When calling StatisticsImpl::HistogramInfo::getMergedHistogram(), if
there is a dying thread, which is calling
ThreadLocalPtr::StaticMeta::OnThreadExit() to merge its thread values to
HistogramInfo, deadlock will occur. Because the former try to hold
merge_lock then ThreadMeta::mutex_, but the later try to hold
ThreadMeta::mutex_ then merge_lock. In short, the locking order isn't
the same.

This patch addressed this issue by releasing merge_lock before folding
thread values.
Closes https://github.com/facebook/rocksdb/pull/1552

Differential Revision: D4211942

Pulled By: ajkr

fbshipit-source-id: ef89bcb
      a0deec96
  41. 20 11月, 2016 1 次提交
    • M
      Use more efficient hash map for deadlock detection · e63350e7
      由 Manuel Ung 提交于 
      Summary:
Currently, deadlock cycles are held in std::unordered_map. The problem with it is that it allocates/deallocates memory on every insertion/deletion. This limits throughput since we're doing this expensive operation while holding a global mutex. Fix this by using a vector which caches memory instead.

Running the deadlock stress test, this change increased throughput from 39k txns/s -> 49k txns/s. The effect is more noticeable in MyRocks.
Closes https://github.com/facebook/rocksdb/pull/1545

Differential Revision: D4205662

Pulled By: lth

fbshipit-source-id: ff990e4
      e63350e7
  43. 19 11月, 2016 2 次提交
  45. 17 11月, 2016 1 次提交
  47. 16 11月, 2016 1 次提交
  49. 15 11月, 2016 1 次提交