Commit Graph

1209 Commits

Author SHA1 Message Date
Mark Callaghan
1f963305a8 Print MB per second compaction throughput separately for reads and writes
Summary:
From this line there used to be one column (MB/sec) that includes reads and writes. This change splits it and for real workloads the rd and wr rates might not match when keys are dropped.
2014/09/29-17:31:01.213162 7f929fbff700 (Original Log Time 2014/09/29-17:31:01.180025) [default] compacted to: files[2 5 0 0 0 0 0], MB/sec: 14.0 rd, 14.0 wr, level 1, files in(4, 0) out(5) MB in(8.5, 0.0) out(8.5), read-write-amplify(2.0) write-amplify(1.0) OK

Test Plan:
make check, grepped LOG

- begin *PUBLIC* platform impact section -
Bugzilla: #
- end platform impact -

Reviewers: igor

Differential Revision: https://reviews.facebook.net/D24237
2014-09-29 17:51:40 -07:00
mike@arpaia.co
f0f7955497 Fixing comile errors on OS X
Summary: Building master on OS X has some compile errors due to implicit type conversions which generate warnings which RocksDB's build settings raise as errors.

Test Plan: It compiles!

Reviewers: ljin, igor

Reviewed By: ljin

Differential Revision: https://reviews.facebook.net/D24135
2014-09-29 16:05:25 -07:00
Mark Callaghan
747523d241 Print per column family metrics in db_bench
Summary: see above

Test Plan:
make check, ran db_bench and looked at output

- begin *PUBLIC* platform impact section -
Bugzilla: #
- end platform impact -

Reviewers: igor

Differential Revision: https://reviews.facebook.net/D24189
2014-09-29 15:47:05 -07:00
erik
827e31c746 Make test use a compatible type in the size checks. 2014-09-29 14:52:16 -07:00
Lei Jin
fd5d80d55e CompactedDB: log using the correct info_log
Summary:
info_log from supplied Options can be nullptr. Using the one from
db_impl. Also call flush after that since no more loggging will happen
and LOG can contain partial output

Test Plan: verified with db_bench

Reviewers: igor, yhchiang, sdong

Reviewed By: sdong

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D24183
2014-09-29 12:45:04 -07:00
Lei Jin
2faf49d5f1 use GetContext to replace callback function pointer
Summary:
Intead of passing callback function pointer and its arg on Table::Get()
interface, passing GetContext. This makes the interface cleaner and
possible better perf. Also adding a fast pass for SaveValue()

Test Plan: make all check

Reviewers: igor, yhchiang, sdong

Reviewed By: sdong

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D24057
2014-09-29 11:09:09 -07:00
sdong
389edb6b1b universal compaction picker: use double for potential overflow
Summary: There is a possible overflow case in universal compaction picker. Use double to make the logic straight-forward

Test Plan: make all check

Reviewers: yhchiang, igor, MarkCallaghan, ljin

Reviewed By: ljin

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D23817
2014-09-26 16:17:05 -07:00
Lei Jin
fbd2dafc9f CompactedDBImpl::MultiGet() for better CuckooTable performance
Summary:
Add the MultiGet API to allow prefetching.
With file size of 1.5G, I configured it to have 0.9 hash ratio that can
fill With 115M keys and result in 2 hash functions, the lookup QPS is
~4.9M/s  vs. 3M/s for Get().
It is tricky to set the parameters right. Since files size is determined
by power-of-two factor, that means # of keys is fixed in each file. With
big file size (thus smaller # of files), we will have more chance to
waste lot of space in the last file - lower space utilization as a
result. Using smaller file size can improve the situation, but that
harms lookup speed.

Test Plan: db_bench

Reviewers: yhchiang, sdong, igor

Reviewed By: sdong

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D23673
2014-09-25 13:34:51 -07:00
Lei Jin
3c68006109 CompactedDBImpl
Summary:
Add a CompactedDBImpl that will enabled when calling OpenForReadOnly()
and the DB only has one level (>0) of files. As a performan comparison,
CuckooTable performs 2.1M/s with CompactedDBImpl vs. 1.78M/s with
ReadOnlyDBImpl.

Test Plan: db_bench

Reviewers: yhchiang, igor, sdong

Reviewed By: sdong

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D23553
2014-09-25 11:14:01 -07:00
Igor Canadi
f7375f39fd Fix double deletes
Summary: While debugging clients compaction issues, I noticed bunch of delete bugs: P16329995. MakeTableName returns sst file with "/" prefix. We also need "/" prefix when we get the files though GetChildren(), so that we can properly dedup the files.

Test Plan: none

Reviewers: sdong, yhchiang, ljin

Reviewed By: ljin

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D23457
2014-09-25 11:08:16 -07:00
Igor Canadi
21ddcf6e4f Remove allow_thread_local
Summary: See https://reviews.facebook.net/D19365

Test Plan: compiles

Reviewers: sdong, yhchiang, ljin

Reviewed By: ljin

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D23907
2014-09-24 13:12:16 -07:00
sdong
cdaf44f9ae Enlarge log size cap when printing file summary
Summary:
Now the file summary is too small for printing. Enlarge it.
To enable it, allow to pass a size to log buffer.

Test Plan:
Add a unit test.
make all check

Reviewers: ljin, yhchiang

Reviewed By: yhchiang

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D21723
2014-09-23 16:56:34 -07:00
sdong
d0de413f4d WriteBatchWithIndex to allow different Comparators for different column families
Summary:
Previously, one single column family is given to WriteBatchWithIndex to index keys for all column families. An extra map from column family ID to comparator is maintained which can override the default comparator given in the constructor. A WriteBatchWithIndex::SetComparatorForCF() is added for user to add comparators per column family.

Also move more codes into anonymous namespace.

Test Plan: Add a unit test

Reviewers: ljin, igor

Reviewed By: igor

Subscribers: dhruba, leveldb, yhchiang

Differential Revision: https://reviews.facebook.net/D23355
2014-09-22 13:47:39 -07:00
Lei Jin
57a32f147f change target_file_size_base to uint64_t
Summary: It contrains the file size to be 4G max with int

Test Plan:
tried to grep instance and made sure other related variables are also
uint64

Reviewers: sdong, yhchiang, igor

Reviewed By: igor

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D23697
2014-09-22 11:15:03 -07:00
Lei Jin
5e6aee4325 dont create backup_input if compaction filter v2 is not used
Summary:
Compaction creates backup_input iterator even though it only needed
when compaction filter v2 is enabled

Test Plan: make all check

Reviewers: sdong, yhchiang, igor

Reviewed By: igor

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D23769
2014-09-22 10:36:53 -07:00
Venkatesh Radhakrishnan
f44594743f RocksDB: Format uint64 using PRIu64 in db_impl.cc
Summary: Use PRIu64 to format uint64 in a portable manner

Test Plan: Run "make all check"

Reviewers: sdong

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D23595
2014-09-18 22:19:41 -07:00
Igor Canadi
90b8c07b48 Fix unit tests errors
Summary: Those were introduced with 2fb1fea30f because the flushing behavior changed when max_background_flushes is > 0.

Test Plan: make check

Reviewers: ljin, yhchiang, sdong

Reviewed By: sdong

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D23577
2014-09-18 13:32:44 -07:00
Lei Jin
51af7c326c CuckooTable: add one option to allow identity function for the first hash function
Summary:
MurmurHash becomes expensive when we do millions Get() a second in one
thread. Add this option to allow the first hash function to use identity
function as hash function. It results in QPS increase from 3.7M/s to
~4.3M/s. I did not observe improvement for end to end RocksDB
performance. This may be caused by other bottlenecks that I will address
in a separate diff.

Test Plan:
```
[ljin@dev1964 rocksdb] ./cuckoo_table_reader_test --enable_perf --file_dir=/dev/shm --write --identity_as_first_hash=0
==== Test CuckooReaderTest.WhenKeyExists
==== Test CuckooReaderTest.WhenKeyExistsWithUint64Comparator
==== Test CuckooReaderTest.CheckIterator
==== Test CuckooReaderTest.CheckIteratorUint64
==== Test CuckooReaderTest.WhenKeyNotFound
==== Test CuckooReaderTest.TestReadPerformance
With 125829120 items, utilization is 93.75%, number of hash functions: 2.
Time taken per op is 0.272us (3.7 Mqps) with batch size of 0, # of found keys 125829120
With 125829120 items, utilization is 93.75%, number of hash functions: 2.
Time taken per op is 0.138us (7.2 Mqps) with batch size of 10, # of found keys 125829120
With 125829120 items, utilization is 93.75%, number of hash functions: 2.
Time taken per op is 0.142us (7.1 Mqps) with batch size of 25, # of found keys 125829120
With 125829120 items, utilization is 93.75%, number of hash functions: 2.
Time taken per op is 0.142us (7.0 Mqps) with batch size of 50, # of found keys 125829120
With 125829120 items, utilization is 93.75%, number of hash functions: 2.
Time taken per op is 0.144us (6.9 Mqps) with batch size of 100, # of found keys 125829120

With 104857600 items, utilization is 78.12%, number of hash functions: 2.
Time taken per op is 0.201us (5.0 Mqps) with batch size of 0, # of found keys 104857600
With 104857600 items, utilization is 78.12%, number of hash functions: 2.
Time taken per op is 0.121us (8.3 Mqps) with batch size of 10, # of found keys 104857600
With 104857600 items, utilization is 78.12%, number of hash functions: 2.
Time taken per op is 0.123us (8.1 Mqps) with batch size of 25, # of found keys 104857600
With 104857600 items, utilization is 78.12%, number of hash functions: 2.
Time taken per op is 0.121us (8.3 Mqps) with batch size of 50, # of found keys 104857600
With 104857600 items, utilization is 78.12%, number of hash functions: 2.
Time taken per op is 0.112us (8.9 Mqps) with batch size of 100, # of found keys 104857600

With 83886080 items, utilization is 62.50%, number of hash functions: 2.
Time taken per op is 0.251us (4.0 Mqps) with batch size of 0, # of found keys 83886080
With 83886080 items, utilization is 62.50%, number of hash functions: 2.
Time taken per op is 0.107us (9.4 Mqps) with batch size of 10, # of found keys 83886080
With 83886080 items, utilization is 62.50%, number of hash functions: 2.
Time taken per op is 0.099us (10.1 Mqps) with batch size of 25, # of found keys 83886080
With 83886080 items, utilization is 62.50%, number of hash functions: 2.
Time taken per op is 0.100us (10.0 Mqps) with batch size of 50, # of found keys 83886080
With 83886080 items, utilization is 62.50%, number of hash functions: 2.
Time taken per op is 0.116us (8.6 Mqps) with batch size of 100, # of found keys 83886080

With 73400320 items, utilization is 54.69%, number of hash functions: 2.
Time taken per op is 0.189us (5.3 Mqps) with batch size of 0, # of found keys 73400320
With 73400320 items, utilization is 54.69%, number of hash functions: 2.
Time taken per op is 0.095us (10.5 Mqps) with batch size of 10, # of found keys 73400320
With 73400320 items, utilization is 54.69%, number of hash functions: 2.
Time taken per op is 0.096us (10.4 Mqps) with batch size of 25, # of found keys 73400320
With 73400320 items, utilization is 54.69%, number of hash functions: 2.
Time taken per op is 0.098us (10.2 Mqps) with batch size of 50, # of found keys 73400320
With 73400320 items, utilization is 54.69%, number of hash functions: 2.
Time taken per op is 0.105us (9.5 Mqps) with batch size of 100, # of found keys 73400320

[ljin@dev1964 rocksdb] ./cuckoo_table_reader_test --enable_perf --file_dir=/dev/shm --write --identity_as_first_hash=1
==== Test CuckooReaderTest.WhenKeyExists
==== Test CuckooReaderTest.WhenKeyExistsWithUint64Comparator
==== Test CuckooReaderTest.CheckIterator
==== Test CuckooReaderTest.CheckIteratorUint64
==== Test CuckooReaderTest.WhenKeyNotFound
==== Test CuckooReaderTest.TestReadPerformance
With 125829120 items, utilization is 93.75%, number of hash functions: 2.
Time taken per op is 0.230us (4.3 Mqps) with batch size of 0, # of found keys 125829120
With 125829120 items, utilization is 93.75%, number of hash functions: 2.
Time taken per op is 0.086us (11.7 Mqps) with batch size of 10, # of found keys 125829120
With 125829120 items, utilization is 93.75%, number of hash functions: 2.
Time taken per op is 0.088us (11.3 Mqps) with batch size of 25, # of found keys 125829120
With 125829120 items, utilization is 93.75%, number of hash functions: 2.
Time taken per op is 0.083us (12.1 Mqps) with batch size of 50, # of found keys 125829120
With 125829120 items, utilization is 93.75%, number of hash functions: 2.
Time taken per op is 0.083us (12.1 Mqps) with batch size of 100, # of found keys 125829120

With 104857600 items, utilization is 78.12%, number of hash functions: 2.
Time taken per op is 0.159us (6.3 Mqps) with batch size of 0, # of found keys 104857600
With 104857600 items, utilization is 78.12%, number of hash functions: 2.
Time taken per op is 0.078us (12.8 Mqps) with batch size of 10, # of found keys 104857600
With 104857600 items, utilization is 78.12%, number of hash functions: 2.
Time taken per op is 0.080us (12.6 Mqps) with batch size of 25, # of found keys 104857600
With 104857600 items, utilization is 78.12%, number of hash functions: 2.
Time taken per op is 0.080us (12.5 Mqps) with batch size of 50, # of found keys 104857600
With 104857600 items, utilization is 78.12%, number of hash functions: 2.
Time taken per op is 0.082us (12.2 Mqps) with batch size of 100, # of found keys 104857600

With 83886080 items, utilization is 62.50%, number of hash functions: 2.
Time taken per op is 0.154us (6.5 Mqps) with batch size of 0, # of found keys 83886080
With 83886080 items, utilization is 62.50%, number of hash functions: 2.
Time taken per op is 0.077us (13.0 Mqps) with batch size of 10, # of found keys 83886080
With 83886080 items, utilization is 62.50%, number of hash functions: 2.
Time taken per op is 0.077us (12.9 Mqps) with batch size of 25, # of found keys 83886080
With 83886080 items, utilization is 62.50%, number of hash functions: 2.
Time taken per op is 0.078us (12.8 Mqps) with batch size of 50, # of found keys 83886080
With 83886080 items, utilization is 62.50%, number of hash functions: 2.
Time taken per op is 0.079us (12.6 Mqps) with batch size of 100, # of found keys 83886080

With 73400320 items, utilization is 54.69%, number of hash functions: 2.
Time taken per op is 0.218us (4.6 Mqps) with batch size of 0, # of found keys 73400320
With 73400320 items, utilization is 54.69%, number of hash functions: 2.
Time taken per op is 0.083us (12.0 Mqps) with batch size of 10, # of found keys 73400320
With 73400320 items, utilization is 54.69%, number of hash functions: 2.
Time taken per op is 0.085us (11.7 Mqps) with batch size of 25, # of found keys 73400320
With 73400320 items, utilization is 54.69%, number of hash functions: 2.
Time taken per op is 0.086us (11.6 Mqps) with batch size of 50, # of found keys 73400320
With 73400320 items, utilization is 54.69%, number of hash functions: 2.
Time taken per op is 0.078us (12.8 Mqps) with batch size of 100, # of found keys 73400320
```

Reviewers: sdong, igor, yhchiang

Reviewed By: igor

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D23451
2014-09-18 11:00:48 -07:00
Igor Canadi
2fb1fea30f Fix syncronization issues 2014-09-18 10:42:54 -07:00
Lei Jin
a062e1f2c4 SetOptions() for memtable related options
Summary: as title

Test Plan:
make all check
I will think a way to set up stress test for this

Reviewers: sdong, yhchiang, igor

Reviewed By: igor

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D23055
2014-09-17 12:49:13 -07:00
Igor Canadi
60a4aa175e Test use_mmap_reads
Summary: We currently don't test mmap reads as part of db_test. Piggyback it on kWalDir test config.

Test Plan: make check

Reviewers: ljin, sdong, yhchiang

Reviewed By: yhchiang

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D23337
2014-09-17 12:31:53 -07:00
Igor Canadi
4a27a2f193 Don't sync manifest when disableDataSync = true
Summary: As we discussed offline

Test Plan: compiles

Reviewers: yhchiang, sdong, ljin, dhruba

Reviewed By: sdong

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D22989
2014-09-15 11:32:01 -07:00
Igor Canadi
04ce1b25f3 Fix #284 2014-09-13 14:14:10 -07:00
Igor Canadi
dee91c259d WriteThread
Summary: This diff just moves the write thread control out of the DBImpl. I will need this as I will control column family data concurrency by only accessing some data in the write thread. That way, we won't have to lock our accesses to column family hash table (mappings from IDs to CFDs).

Test Plan: make check

Reviewers: sdong, yhchiang, ljin

Reviewed By: ljin

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D23301
2014-09-12 16:23:58 -07:00
Igor Canadi
540a257f2c Fix WAL synced
Summary: Uhm...

Test Plan: nope

Reviewers: sdong, yhchiang, tnovak, ljin

Reviewed By: ljin

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D23343
2014-09-12 16:15:29 -07:00
Chilledheart
49fe329e5e Fix build issue under macosx 2014-09-13 05:05:22 +08:00
Feng Zhu
0352a9fa91 add_wrapped_bloom_test
Summary:
1. wrap a filter policy like what fbcode/multifeed/rocksdb/MultifeedRocksDbKey.h
   to ensure that rocksdb works fine after filterpolicy interface change

Test Plan: 1. valgrind ./bloom_test

Reviewers: ljin, igor, yhchiang, dhruba, sdong

Reviewed By: sdong

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D23229
2014-09-11 16:33:46 -07:00
Igor Canadi
9c0e66ce98 Don't run background jobs (flush, compactions) when bg_error_ is set
Summary:
If bg_error_ is set, that means that we mark DB read only. However, current behavior still continues the flushes and compactions, even though bg_error_ is set.

On the other hand, if bg_error_ is set, we will return Status::OK() from CompactRange(), although the compaction didn't actually succeed.

This is clearly not desired behavior. I found this when I was debugging t5132159, although I'm pretty sure these aren't related.

Also, when we're shutting down, it's dangerous to exit RunManualCompaction(), since that will destruct ManualCompaction object. Background compaction job might still hold a reference to manual_compaction_ and this will lead to undefined behavior. I changed the behavior so that we only exit RunManualCompaction when manual compaction job is marked done.

Test Plan: make check

Reviewers: sdong, ljin, yhchiang

Reviewed By: yhchiang

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D23223
2014-09-11 16:24:16 -07:00
Igor Canadi
a9639bda84 Fix valgrind test
Summary: Get valgrind to stop complaining about uninitialized value

Test Plan: valgrind not complaining anymore

Reviewers: sdong, yhchiang, ljin

Reviewed By: ljin

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D23289
2014-09-11 15:36:30 -07:00
Igor Canadi
d1f24dc7ee Relax FlushSchedule test
Summary: The test makes sure that we don't call flush too often. For that, it's ok to check if we have less than 10 table files. Otherwise, the test is flaky because it's hard to estimate number of entries in the memtable before it gets flushed (any ideas?)

Test Plan: Still works, but hopefully less flaky.

Reviewers: ljin, sdong, yhchiang

Reviewed by: yhchiang

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D23241
2014-09-11 11:00:45 -07:00
Igor Canadi
3d9e6f7759 Push model for flushing memtables
Summary:
When memtable is full it calls the registered callback. That callback then registers column family as needing the flush. Every write checks if there are some column families that need to be flushed. This completely eliminates the need for MakeRoomForWrite() function and simplifies our Write code-path.

There is some complexity with the concurrency when the column family is dropped. I made it a bit less complex by dropping the column family from the write thread in https://reviews.facebook.net/D22965. Let me know if you want to discuss this.

Test Plan: make check works. I'll also run db_stress with creating and dropping column families for a while.

Reviewers: yhchiang, sdong, ljin

Reviewed By: ljin

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D23067
2014-09-10 18:46:09 -07:00
Igor Canadi
059e584dd3 [unit test] CompactRange should fail if we don't have space
Summary:
See t5106397.

Also, few more changes:
1. in unit tests, the assumption is that writes will be dropped when there is no space left on device. I changed the wording around it.
2. InvalidArgument() errors are only when user-provided arguments are invalid. When the file is corrupted, we need to return Status::Corruption

Test Plan: make check

Reviewers: sdong, ljin

Reviewed By: ljin

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D23145
2014-09-10 17:00:00 -07:00
Igor Canadi
a52cecb56c Fix Mac compile 2014-09-09 18:42:35 -07:00
Jonah Cohen
092f97e219 Fix comments and typos
Summary: Correct some comments and typos in RocksDB.

Test Plan: Inspection

Reviewers: sdong, igor

Reviewed By: igor

Differential Revision: https://reviews.facebook.net/D23133
2014-09-09 15:20:49 -07:00
Igor Canadi
0a42295a24 Fix SimpleWriteTimeoutTest
Summary:
In column family's SanitizeOptions() [1], we make sure that min_write_buffer_number_to_merge is normal value. However, this test depended on the fact that setting min_write_buffer_number_to_merge to be bigger than max_write_buffer_number will cause a deadlock. I'm not sure how it worked before.

This diff fixes it by scheduling sleeping background task, which will actually block any attempts of flushing.

[1] https://github.com/facebook/rocksdb/blob/master/db/column_family.cc#L104

Test Plan: the test works now

Reviewers: yhchiang, sdong, ljin

Reviewed By: ljin

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D23103
2014-09-09 11:50:05 -07:00
sdong
06d986252a Always pass MergeContext as pointer, not reference
Summary: To follow the coding convention and make sure when passing reference as a parameter it is also const, pass MergeContext as a pointer to mem tables.

Test Plan: make all check

Reviewers: ljin, igor

Reviewed By: igor

Subscribers: leveldb, dhruba, yhchiang

Differential Revision: https://reviews.facebook.net/D23085
2014-09-09 11:37:32 -07:00
Stanislau Hlebik
d343c3fe46 Improve db recovery
Summary: Avoid creating unnecessary sst files while db opening

Test Plan: make all check

Reviewers: sdong, igor

Reviewed By: igor

Subscribers: zagfox, yhchiang, ljin, leveldb

Differential Revision: https://reviews.facebook.net/D20661
2014-09-09 11:18:50 -07:00
Lei Jin
52311463e9 MemTableOptions
Summary: removed reference to options in WriteBatch and DBImpl::Get()

Test Plan: make all check

Reviewers: yhchiang, igor, sdong

Reviewed By: sdong

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D23049
2014-09-08 18:46:52 -07:00
Lei Jin
171d4ff4a2 remove TailingIterator reference in db_impl.h
Summary: as title

Test Plan: make release

Reviewers: igor

Differential Revision: https://reviews.facebook.net/D23073
2014-09-08 15:39:53 -07:00
Lei Jin
9b0f7ffa1c rename version_set options_ to db_options_ to avoid confusion
Summary: as title

Test Plan: make release

Reviewers: sdong, yhchiang, igor

Reviewed By: igor

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D23007
2014-09-08 15:25:01 -07:00
Igor Canadi
2d57828d0e Check stop level trigger-0 before slowdown level-0 trigger
Summary: ...

Test Plan: Can't repro the test failure, but let's see what jenkins says

Reviewers: zagfox, sdong, ljin

Reviewed By: sdong, ljin

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D23061
2014-09-08 15:23:58 -07:00
Lei Jin
659d2d50c3 move compaction_filter to immutable_options
Summary:
all shared_ptrs are in immutable_options now. This will also make
options assignment a little cheaper

Test Plan: make release

Reviewers: sdong, yhchiang, igor

Reviewed By: igor

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D23001
2014-09-08 15:09:25 -07:00
Lei Jin
048560a642 reduce references to cfd->options() in DBImpl
Summary:
I found it is almost impossible to get rid of this function in a single
batch. I will take a step by step approach

Test Plan: make release

Reviewers: sdong, yhchiang, igor

Reviewed By: igor

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D22995
2014-09-08 15:04:34 -07:00
sdong
011241bb99 DB::Flush() Do not wait for background threads when there is nothing in mem table
Summary:
When we have multiple column families, users can issue Flush() on every column families to make sure everything is flushes, even if some of them might be empty. By skipping the waiting for empty cases, it can be greatly speed up.

Still wait for people's comments before writing unit tests for it.

Test Plan: Will write a unit test to make sure it is correct.

Reviewers: ljin, yhchiang, igor

Reviewed By: igor

Subscribers: leveldb, dhruba

Differential Revision: https://reviews.facebook.net/D22953
2014-09-08 13:40:42 -07:00
Igor Canadi
a2bb7c3c33 Push- instead of pull-model for managing Write stalls
Summary:
Introducing WriteController, which is a source of truth about per-DB write delays. Let's define an DB epoch as a period where there are no flushes and compactions (i.e. new epoch is started when flush or compaction finishes). Each epoch can either:
* proceed with all writes without delay
* delay all writes by fixed time
* stop all writes

The three modes are recomputed at each epoch change (flush, compaction), rather than on every write (which is currently the case).

When we have a lot of column families, our current pull behavior adds a big overhead, since we need to loop over every column family for every write. With new push model, overhead on Write code-path is minimal.

This is just the start. Next step is to also take care of stalls introduced by slow memtable flushes. The final goal is to eliminate function MakeRoomForWrite(), which currently needs to be called for every column family by every write.

Test Plan: make check for now. I'll add some unit tests later. Also, perf test.

Reviewers: dhruba, yhchiang, MarkCallaghan, sdong, ljin

Reviewed By: ljin

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D22791
2014-09-08 11:20:25 -07:00
Feng Zhu
0af157f9bf Implement full filter for block based table.
Summary:
1. Make filter_block.h a base class. Derive block_based_filter_block and full_filter_block. The previous one is the traditional filter block. The full_filter_block is newly added. It would generate a filter block that contain all the keys in SST file.

2. When querying a key, table would first check if full_filter is available. If not, it would go to the exact data block and check using block_based filter.

3. User could choose to use full_filter or tradional(block_based_filter). They would be stored in SST file with different meta index name. "filter.filter_policy" or "full_filter.filter_policy". Then, Table reader is able to know the fllter block type.

4. Some optimizations have been done for full_filter_block, thus it requires a different interface compared to the original one in filter_policy.h.

5. Actual implementation of filter bits coding/decoding is placed in util/bloom_impl.cc

Benchmark: base commit 1d23b5c470
Command:
db_bench --db=/dev/shm/rocksdb --num_levels=6 --key_size=20 --prefix_size=20 --keys_per_prefix=0 --value_size=100 --write_buffer_size=134217728 --max_write_buffer_number=2 --target_file_size_base=33554432 --max_bytes_for_level_base=1073741824 --verify_checksum=false --max_background_compactions=4 --use_plain_table=0 --memtablerep=prefix_hash --open_files=-1 --mmap_read=1 --mmap_write=0 --bloom_bits=10 --bloom_locality=1 --memtable_bloom_bits=500000 --compression_type=lz4 --num=393216000 --use_hash_search=1 --block_size=1024 --block_restart_interval=16 --use_existing_db=1 --threads=1 --benchmarks=readrandom —disable_auto_compactions=1
Read QPS increase for about 30% from 2230002 to 2991411.

Test Plan:
make all check
valgrind db_test
db_stress --use_block_based_filter = 0
./auto_sanity_test.sh

Reviewers: igor, yhchiang, ljin, sdong

Reviewed By: sdong

Subscribers: dhruba, leveldb

Differential Revision: https://reviews.facebook.net/D20979
2014-09-08 10:37:05 -07:00
Igor Canadi
9360cc690e Fix valgrind issue 2014-09-08 08:01:25 -07:00
Igor Canadi
9f1c80b556 Drop column family from write thread
Summary: If we drop column family only from (single) write thread, we can be sure that nobody will drop the column family while we're writing (and our mutex is released). This greatly simplifies my patch that's getting rid of MakeRoomForWrite().

Test Plan: make check, but also running stress test

Reviewers: ljin, sdong

Reviewed By: sdong

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D22965
2014-09-05 15:20:05 -07:00
Igor Canadi
8de151bb99 Add db_bench with lots of column families to regression tests
Summary:
That way we can see when this graph goes up and be happy.

Couple of changes:
1. title
2. fix db_bench to delete column families before deleting the DB. this was asserting when compiled in debug mode
3. don't sync manifest when disableDataSync. We discussed this offline. I can move it to separate diff if you'd like

Test Plan: ran it

Reviewers: sdong, yhchiang, ljin

Reviewed By: ljin

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D22815
2014-09-05 14:20:18 -07:00
Lei Jin
c9e419ccb6 rename options_ to db_options_ in DBImpl to avoid confusion
Summary: as title

Test Plan: make release

Reviewers: sdong, igor

Reviewed By: igor

Subscribers: leveldb

Differential Revision: https://reviews.facebook.net/D22935
2014-09-05 11:48:17 -07:00