Skip to content

R
rocksdb

kvdb / rocksdb
11 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. 14 7月, 2015 5 次提交
  3. 12 7月, 2015 1 次提交
  5. 11 7月, 2015 6 次提交
  7. 10 7月, 2015 8 次提交
  9. 09 7月, 2015 3 次提交
    • D
      All of these are in the new code added past 3.10 · d8586ab2
      由 Dmitri Smirnov 提交于 
           1) Crash in env_win.cc that prevented db_test run to completion and some new tests
     2) Fix new corruption tests in DBTest by allowing a shared trunction of files. Note that this is generally needed ONLY for tests.
     3) Close database so WAL is closed prior to inducing corruption similar to what we did within Corruption tests.
      d8586ab2
    • P
      Fix function name format according to google style · 4bed00a4
      由 Poornima Chozhiyath Raman 提交于 
      Summary: Change the naming style of getter and setters according to Google C++ style in compaction.h file

Test Plan: Compilation success

Reviewers: sdong

Reviewed By: sdong

Subscribers: leveldb, dhruba

Differential Revision: https://reviews.facebook.net/D41265
      4bed00a4
    • K
      Added multi WAL log testing to recovery tests. · e2e3d84b
      由 krad 提交于 
      Summary: Currently there is no test in the suite to test the case where
there are multiple WAL files and there is a corruption in one of them. We have
tests for single WAL file corruption scenarios. Added tests to mock
the scenarios for all combinations of recovery modes and corruption in
specified file locations.

Test Plan: Run make check

Reviewers: sdong igor

CC: leveldb@

Task ID: #7501229

Blame Rev:
      e2e3d84b
  11. 08 7月, 2015 13 次提交
  13. 07 7月, 2015 2 次提交
    • A
      Added tests for ExpandWhileOverlapping() · 58d7ab3c
      由 Andres Notzli 提交于 
      Summary:
This patch adds three test cases for ExpandWhileOverlapping()
to the compaction_picker_test test suite.
ExpandWhileOverlapping() only has an effect if the comparison
function for the internal keys allows for overlapping user
keys in different SST files on the same level. Thus, this
patch adds a comparator based on sequence numbers to
compaction_picker_test for the new test cases.

Test Plan:
- make compaction_picker_test && ./compaction_picker_test
  -> All tests pass
- Replace body of ExpandWhileOverlapping() with `return true`
  -> Compile and run ./compaction_picker_test as before
  -> New tests fail

Reviewers: sdong, yhchiang, rven, anthony, IslamAbdelRahman, kradhakrishnan, igor

Reviewed By: igor

Subscribers: dhruba, leveldb

Differential Revision: https://reviews.facebook.net/D41277
      58d7ab3c
    • I
      Fix compaction_job_test · 155ce60d
      由 Igor Canadi 提交于 
      Summary:
Two issues:
* the input keys to the compaction don't include sequence number.
* sequence number is set to max(seq_num), but it should be set to max(seq_num)+1, because the condition here is strictly-larger (i.e. we will only zero-out sequence number if the DB's sequence number is strictly greater than the key's sequence number): https://github.com/facebook/rocksdb/blob/master/db/compaction_job.cc#L830

Test Plan: make compaction_job_test && ./compaction_job_test

Reviewers: sdong, lovro

Reviewed By: lovro

Subscribers: dhruba, leveldb

Differential Revision: https://reviews.facebook.net/D41247
      155ce60d
  15. 06 7月, 2015 1 次提交
    • L
      Replace std::priority_queue in MergingIterator with custom heap · b6655a67
      由 lovro 提交于 
      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
      b6655a67
  17. 03 7月, 2015 1 次提交