- 27 8月, 2014 7 次提交
-
-
由 Igor Canadi 提交于
-
由 Igor Canadi 提交于
-
由 Radheshyam Balasundaram 提交于
Summary: In DBImpl::Recover method, while loading memtables, also check if memtables are empty. Use this in DBImplReadonly to determine whether to lookup memtable or not. Test Plan: db_test make check all Reviewers: sdong, yhchiang, ljin, igor Reviewed By: ljin Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D22281
-
由 Stanislau Hlebik 提交于
Summary: Add property 'rocksdb.is-file-deletions-enable' which equals disable_delete_obsole_file_ Test Plan: make all check Reviewers: sdong Reviewed By: sdong Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D22119
-
由 Lei Jin 提交于
Summary: also fix HISTORY.md Test Plan: make all check Reviewers: sdong, yhchiang, igor Reviewed By: igor Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D22437
-
由 Igor Canadi 提交于
Summary: See https://github.com/facebook/rocksdb/issues/244#issuecomment-53372297 Also see this: https://github.com/facebook/rocksdb/blob/master/util/env_posix.cc#L1075 Test Plan: compiles Reviewers: yhchiang, ljin, sdong Reviewed By: ljin Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D22419
-
由 Lei Jin 提交于
Summary: It was creating BlockBasedTableOptions object in a loop without calling destroy() Test Plan: valgrind ./c_test --leak-check=full --show-reachable=yes Reviewers: sdong, igor Reviewed By: igor Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D22431
-
- 26 8月, 2014 5 次提交
-
-
由 Igor Canadi 提交于
Summary: Two things: 1. Use hash-based index for data column family 2. Use Get() instead of Iterator Seek() when DB is opened read-only Test Plan: added read-only test in unit test Reviewers: yinwang Reviewed By: yinwang Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D22323
-
由 Lei Jin 提交于
ReadOptions.total_order_seek to allow total order seek for block-based table when hash index is enabled Summary: as title Test Plan: table_test Reviewers: igor, yhchiang, sdong Reviewed By: sdong Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D22239
-
由 Lei Jin 提交于
Summary: Add a virtual function in table factory that will print table options Test Plan: make release Reviewers: igor, yhchiang, sdong Reviewed By: sdong Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D22149
-
由 Lei Jin 提交于
Summary: as title Test Plan: tested on my mac make rocksdbjava make jtest Reviewers: sdong, igor, yhchiang Reviewed By: yhchiang Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D21963
-
由 Lei Jin 提交于
Summary: I will move compression related options in a separate diff since this diff is already pretty lengthy. I guess I will also need to change JNI accordingly :( Test Plan: make all check Reviewers: yhchiang, igor, sdong Reviewed By: igor Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D21915
-
- 24 8月, 2014 2 次提交
-
-
由 Igor Canadi 提交于
Add missing include to use std::unique_ptr
-
由 Andrew Bonventre 提交于
This was causing issues when including this header from another file.
-
- 23 8月, 2014 1 次提交
-
-
由 Igor Canadi 提交于
Summary: I am currently working on a project that uses RocksDB. While debugging some perf issues, I came up across interesting compaction concurrency issue. Namely, I had 15 idle threads and a good comapction to do, but CompactionPicker returned "Compaction nothing to do". Here's how Internal stats looked: 2014/08/22-08:08:04.551982 7fc7fc3f5700 ------- DUMPING STATS ------- 2014/08/22-08:08:04.552000 7fc7fc3f5700 ** Compaction Stats [default] ** Level Files Size(MB) Score Read(GB) Rn(GB) Rnp1(GB) Write(GB) Wnew(GB) RW-Amp W-Amp Rd(MB/s) Wr(MB/s) Rn(cnt) Rnp1(cnt) Wnp1(cnt) Wnew(cnt) Comp(sec) Comp(cnt) Avg(sec) Stall(sec) Stall(cnt) Avg(ms) ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ L0 7/5 353 1.0 0.0 0.0 0.0 2.3 2.3 0.0 0.0 0.0 9.4 0 0 0 0 247 46 5.359 8.53 1 8526.25 L1 2/2 86 1.3 2.6 1.9 0.7 2.6 1.9 2.7 1.3 24.3 24.0 39 19 71 52 109 11 9.938 0.00 0 0.00 L2 26/0 833 1.3 5.7 1.7 4.0 5.2 1.2 6.3 3.0 15.6 14.2 47 112 147 35 373 44 8.468 0.00 0 0.00 L3 12/0 505 0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 0 0 0 0 0.000 0.00 0 0.00 Sum 47/7 1778 0.0 8.3 3.6 4.6 10.0 5.4 8.1 4.4 11.6 14.1 86 131 218 87 728 101 7.212 8.53 1 8526.25 Int 0/0 0 0.0 2.4 0.8 1.6 2.7 1.2 11.5 6.1 12.0 13.6 20 43 63 20 203 23 8.845 0.00 0 0.00 Flush(GB): accumulative 2.266, interval 0.444 Stalls(secs): 0.000 level0_slowdown, 0.000 level0_numfiles, 8.526 memtable_compaction, 0.000 leveln_slowdown_soft, 0.000 leveln_slowdown_hard Stalls(count): 0 level0_slowdown, 0 level0_numfiles, 1 memtable_compaction, 0 leveln_slowdown_soft, 0 leveln_slowdown_hard ** DB Stats ** Uptime(secs): 336.8 total, 60.4 interval Cumulative writes: 61584000 writes, 6480589 batches, 9.5 writes per batch, 1.39 GB user ingest Cumulative WAL: 0 writes, 0 syncs, 0.00 writes per sync, 0.00 GB written Interval writes: 11235257 writes, 1175050 batches, 9.6 writes per batch, 259.9 MB user ingest Interval WAL: 0 writes, 0 syncs, 0.00 writes per sync, 0.00 MB written To see what happened, go here: https://github.com/facebook/rocksdb/blob/47b452cfcf9b1487d41f886a98bc0d6f95587e90/db/compaction_picker.cc#L430 * The for loop started with level 1, because it has the worst score. * PickCompactionBySize on L429 returned nullptr because all files were being compacted * ExpandWhileOverlapping(c) returned true (because that's what it does when it gets nullptr!?) * for loop break-ed, never trying compactions for level 2 :( :( This bug was present at least since January. I have no idea how we didn't find this sooner. Test Plan: Unit testing compaction picker is hard. I tested this by running my service and observing L0->L1 and L2->L3 compactions in parallel. However, for long-term, I opened the task #4968469. @yhchiang is currently refactoring CompactionPicker, hopefully the new version will be unit-testable ;) Here's how my compactions look like after the patch: 2014/08/22-08:50:02.166699 7f3400ffb700 ------- DUMPING STATS ------- 2014/08/22-08:50:02.166722 7f3400ffb700 ** Compaction Stats [default] ** Level Files Size(MB) Score Read(GB) Rn(GB) Rnp1(GB) Write(GB) Wnew(GB) RW-Amp W-Amp Rd(MB/s) Wr(MB/s) Rn(cnt) Rnp1(cnt) Wnp1(cnt) Wnew(cnt) Comp(sec) Comp(cnt) Avg(sec) Stall(sec) Stall(cnt) Avg(ms) ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ L0 8/5 404 1.5 0.0 0.0 0.0 4.3 4.3 0.0 0.0 0.0 9.6 0 0 0 0 463 88 5.260 0.00 0 0.00 L1 2/2 60 0.9 4.8 3.9 0.8 4.7 3.9 2.4 1.2 23.9 23.6 80 23 131 108 204 19 10.747 0.00 0 0.00 L2 23/3 697 1.0 11.6 3.5 8.1 10.9 2.8 6.4 3.1 17.7 16.6 95 242 317 75 669 92 7.268 0.00 0 0.00 L3 58/14 2207 0.3 6.2 1.6 4.6 5.9 1.3 7.4 3.6 14.6 13.9 43 121 159 38 436 36 12.106 0.00 0 0.00 Sum 91/24 3368 0.0 22.5 9.1 13.5 25.8 12.4 11.2 6.0 13.0 14.9 218 386 607 221 1772 235 7.538 0.00 0 0.00 Int 0/0 0 0.0 3.2 0.9 2.3 3.6 1.3 15.3 8.0 12.4 13.7 24 66 89 23 266 27 9.838 0.00 0 0.00 Flush(GB): accumulative 4.336, interval 0.444 Stalls(secs): 0.000 level0_slowdown, 0.000 level0_numfiles, 0.000 memtable_compaction, 0.000 leveln_slowdown_soft, 0.000 leveln_slowdown_hard Stalls(count): 0 level0_slowdown, 0 level0_numfiles, 0 memtable_compaction, 0 leveln_slowdown_soft, 0 leveln_slowdown_hard ** DB Stats ** Uptime(secs): 577.7 total, 60.1 interval Cumulative writes: 116960736 writes, 11966220 batches, 9.8 writes per batch, 2.64 GB user ingest Cumulative WAL: 0 writes, 0 syncs, 0.00 writes per sync, 0.00 GB written Interval writes: 11643735 writes, 1206136 batches, 9.7 writes per batch, 269.2 MB user ingest Interval WAL: 0 writes, 0 syncs, 0.00 writes per sync, 0.00 MB written Yay for concurrent L0->L1 and L2->L3 compactions! Reviewers: sdong, yhchiang, ljin Reviewed By: yhchiang Subscribers: yhchiang, leveldb Differential Revision: https://reviews.facebook.net/D22305
-
- 22 8月, 2014 2 次提交
-
-
由 Siying Dong 提交于
Fix compilation issue on OSX
-
由 Shao Yu Zhang 提交于
-
- 21 8月, 2014 7 次提交
-
-
由 Radheshyam Balasundaram 提交于
Summary: - Implement Prepare method - Rewrite performance tests in cuckoo_table_reader_test to write new file only if one doesn't already exist. - Add performance tests for batch lookup along with prefetching. Test Plan: ./cuckoo_table_reader_test --enable_perf Results (We get better results if we used int64 comparator instead of string comparator (TBD in future diffs)): With 100000000 items and hash table ratio 0.500000, number of hash functions used: 2. Time taken per op is 0.208us (4.8 Mqps) with batch size of 0 With 100000000 items and hash table ratio 0.500000, number of hash functions used: 2. Time taken per op is 0.182us (5.5 Mqps) with batch size of 10 With 100000000 items and hash table ratio 0.500000, number of hash functions used: 2. Time taken per op is 0.161us (6.2 Mqps) with batch size of 25 With 100000000 items and hash table ratio 0.500000, number of hash functions used: 2. Time taken per op is 0.161us (6.2 Mqps) with batch size of 50 With 100000000 items and hash table ratio 0.500000, number of hash functions used: 2. Time taken per op is 0.163us (6.1 Mqps) with batch size of 100 With 100000000 items and hash table ratio 0.600000, number of hash functions used: 3. Time taken per op is 0.252us (4.0 Mqps) with batch size of 0 With 100000000 items and hash table ratio 0.600000, number of hash functions used: 3. Time taken per op is 0.192us (5.2 Mqps) with batch size of 10 With 100000000 items and hash table ratio 0.600000, number of hash functions used: 3. Time taken per op is 0.195us (5.1 Mqps) with batch size of 25 With 100000000 items and hash table ratio 0.600000, number of hash functions used: 3. Time taken per op is 0.191us (5.2 Mqps) with batch size of 50 With 100000000 items and hash table ratio 0.600000, number of hash functions used: 3. Time taken per op is 0.194us (5.1 Mqps) with batch size of 100 With 100000000 items and hash table ratio 0.750000, number of hash functions used: 3. Time taken per op is 0.228us (4.4 Mqps) with batch size of 0 With 100000000 items and hash table ratio 0.750000, number of hash functions used: 3. Time taken per op is 0.185us (5.4 Mqps) with batch size of 10 With 100000000 items and hash table ratio 0.750000, number of hash functions used: 3. Time taken per op is 0.186us (5.4 Mqps) with batch size of 25 With 100000000 items and hash table ratio 0.750000, number of hash functions used: 3. Time taken per op is 0.189us (5.3 Mqps) with batch size of 50 With 100000000 items and hash table ratio 0.750000, number of hash functions used: 3. Time taken per op is 0.188us (5.3 Mqps) with batch size of 100 With 100000000 items and hash table ratio 0.900000, number of hash functions used: 3. Time taken per op is 0.325us (3.1 Mqps) with batch size of 0 With 100000000 items and hash table ratio 0.900000, number of hash functions used: 3. Time taken per op is 0.196us (5.1 Mqps) with batch size of 10 With 100000000 items and hash table ratio 0.900000, number of hash functions used: 3. Time taken per op is 0.199us (5.0 Mqps) with batch size of 25 With 100000000 items and hash table ratio 0.900000, number of hash functions used: 3. Time taken per op is 0.196us (5.1 Mqps) with batch size of 50 With 100000000 items and hash table ratio 0.900000, number of hash functions used: 3. Time taken per op is 0.209us (4.8 Mqps) with batch size of 100 Reviewers: sdong, yhchiang, igor, ljin Reviewed By: ljin Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D22167
-
由 Yueh-Hsuan Chiang 提交于
Summary: Fix the error of c_test.c Test Plan: make c_test ./c_test
-
由 Yueh-Hsuan Chiang 提交于
Summary: Add missing implementaiton of SanitizeDBOptions in simple_table_db_test.cc Test Plan: make simple_table_db_test.cc
-
由 Yueh-Hsuan Chiang 提交于
Summary: Currently, PlainTable must use mmap_reads. When PlainTable is used but allow_mmap_reads is not set, rocksdb will fail in flush. This diff improve Options sanitization and add MmapReadRequired() to TableFactory. Test Plan: export ROCKSDB_TESTS=PlainTableOptionsSanitizeTest make db_test -j32 ./db_test Reviewers: sdong, ljin Reviewed By: ljin Subscribers: you, leveldb Differential Revision: https://reviews.facebook.net/D21939
-
由 Jonah Cohen 提交于
Summary: ManifestDumpCommand::DoCommand was allocating a VersionSet and never freeing it. Test Plan: make Reviewers: igor Reviewed By: igor Differential Revision: https://reviews.facebook.net/D22221
-
由 sdong 提交于
Summary: A previous change triggered a change by mistake: DestroyDB() will keep info logs under DB directory. Revert the unintended change. Test Plan: Add a unit test case to verify it. Reviewers: ljin, yhchiang, igor Reviewed By: igor Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D22209
-
由 Igor Canadi 提交于
Summary: We need to start compression at level 1, while OptimizeForLevelComapaction() only sets up rocksdb to start compressing at level 2. I also adjusted some other things. Test Plan: compiles Reviewers: yinwang Reviewed By: yinwang Differential Revision: https://reviews.facebook.net/D22203
-
- 20 8月, 2014 1 次提交
-
-
由 sdong 提交于
Summary: Make table_reader_bench cover all the three table formats. Test Plan: Run it using three options Reviewers: radheshyamb, ljin Reviewed By: ljin Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D22137
-
- 19 8月, 2014 9 次提交
-
-
由 Igor Canadi 提交于
-
由 Igor Canadi 提交于
Summary: I was checking some functions in coding.h and coding.cc when I noticed these unused functions. Let's remove them. Test Plan: compiles Reviewers: sdong, ljin, yhchiang, dhruba Reviewed By: dhruba Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D22077
-
由 Radheshyam Balasundaram 提交于
Summary: Adding num_column_families flag. Adding support for column families in DoWrite and ReadRandom methods. [Igor, please let me know if this approach sounds good. I shall add it to other methods too.] Test Plan: Ran fillseq on 1M keys and 10 Column families and ran readrandom. Reviewers: sdong, yhchiang, igor, ljin Reviewed By: ljin Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D21387
-
由 sdong 提交于
Summary: Add WriteBatchWithIndex so that a user can query data out of a WriteBatch, to support MongoDB's read-its-own-write. WriteBatchWithIndex uses a skiplist to store the binary index. The index stores the offset of the entry in the write batch. When searching for a key, the key for the entry is read by read the entry from the write batch from the offset. Define a new iterator class for querying data out of WriteBatchWithIndex. A user can create an iterator of the write batch for one column family, seek to a key and keep calling Next() to see next entries. I will add more unit tests if people are OK about this API. Test Plan: make all check Add unit tests. Reviewers: yhchiang, igor, MarkCallaghan, ljin Reviewed By: ljin Subscribers: dhruba, leveldb, xjin Differential Revision: https://reviews.facebook.net/D21381
-
由 sdong 提交于
Summary: N/A Test Plan: N/A Reviewers: yhchiang, ljin Reviewed By: ljin Subscribers: dhruba, leveldb, igor Differential Revision: https://reviews.facebook.net/D22053
-
由 sdong 提交于
Summary: Bump up version after we've cut 3.4 Test Plan: N/A Reviewers: yhchiang, ljin Reviewed By: ljin Subscribers: leveldb, dhruba, igor Differential Revision: https://reviews.facebook.net/D22047
-
由 Radheshyam Balasundaram 提交于
Summary: Adding flags to use cuckoo table SST in db_bench.cc Test Plan: Ran benchmark with fillseq and readrandom Reviewers: sdong, ljin Reviewed By: ljin Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D21729
-
由 Igor Canadi 提交于
-
由 Lei Jin 提交于
Summary: auto_roll_logger_test fails from time to time. I wasn't able to repro the issue but by looking at the code, it seems like the initial ctime_ value can be set to the boundary of the second so it may still have a chance to get rolled when interval is set to 1 second. ``` util/auto_roll_logger_test.cc:120: failed: 118 > 708 ==19470== Syscall param msync(start) points to unaddressable byte(s) ==19470== at 0x4E46CE0: __msync_nocancel (in /usr/local/fbcode/gcc-4.8.1-glibc-2.17/lib/libpthread-2.17.so) ==19470== by 0x584EFB: access_mem (Ginit.c:137) ==19470== by 0x5834E3: _ULx86_64_access_reg (libunwind_i.h:162) ==19470== by 0x585601: apply_reg_state (Gparser.c:742) ==19470== by 0x5866BE: _ULx86_64_dwarf_find_save_locs (Gparser.c:883) ==19470== by 0x584550: _ULx86_64_dwarf_step (Gstep.c:34) ==19470== by 0x583653: _ULx86_64_step (Gstep.c:71) ==19470== by 0x583FD2: _ULx86_64_tdep_trace (Gtrace.c:217) ==19470== by 0x5831C3: backtrace (backtrace.c:69) Test Plan: ./auto_roll_logger_test Reviewers: sdong, yhchiang, igor Reviewed By: igor Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D21951
-
- 18 8月, 2014 1 次提交
-
-
由 Igor Canadi 提交于
Pass parsed user key to prefix extractor in V2 compaction
-
- 16 8月, 2014 1 次提交
-
-
由 Yueh-Hsuan Chiang 提交于
Summary: Avoid retrying to read property block from a table when it does not exist in updating stats for compensating deletion entries. In addition, ReadTableProperties() now returns Status::NotFound instead of Status::Corruption when table properties does not exist in the file. Test Plan: make db_test -j32 export ROCKSDB_TESTS=CompactionDeleteionTrigger ./db_test Reviewers: ljin, sdong Reviewed By: sdong Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D21867
-
- 15 8月, 2014 4 次提交
-
-
由 Lei Jin 提交于
fix compile error under Mac OS X
-
由 Jonah Cohen 提交于
Test Plan: Inspection Reviewers: sdong Reviewed By: sdong Differential Revision: https://reviews.facebook.net/D21891
-
由 Jonah Cohen 提交于
Summary: The prefix and postfix operators were mixed up in the autovector class. Test Plan: Inspection Reviewers: sdong, kailiu Reviewed By: kailiu Differential Revision: https://reviews.facebook.net/D21873
-
由 sdong 提交于
Summary: 1. Support purging info logs from a separate paths from DB path. Refactor the codes of generating info log prefixes so that it can be called when generating new files and scanning log directory. 2. Fix the bug of not scanning multiple DB paths (should only impact multiple DB paths) Test Plan: Add unit test for generating and parsing info log files Add end-to-end test in db_test Reviewers: yhchiang, ljin Reviewed By: ljin Subscribers: leveldb, igor, dhruba Differential Revision: https://reviews.facebook.net/D21801
-