Summary:
In case of non-TransactionDB and avoid_flush_during_recovery = true, RocksDB won't
flush the data from WAL to L0 for all column families if possible. As a
result, not all column families can increase their log_numbers, and
min_log_number_to_keep won't change.
For transaction DB (.allow_2pc), even with the flush, there may be old WAL files that it must not delete because they can contain data of uncommitted transactions and min_log_number_to_keep won't change.
If we persist a new MANIFEST with
advanced log_numbers for some column families, then during a second
crash after persisting the MANIFEST, RocksDB will see some column
families' log_numbers larger than the corrupted WAL, and the "column family inconsistency" error will be hit, causing recovery to fail.
This PR update unit tests to emulate the errors and tests are failing without a fix.
Error:
```
[ RUN ] CorruptionTest/CrashDuringRecoveryWithCorruptionTest.CrashDuringRecovery/0
db/corruption_test.cc:1190: Failure
DB::Open(options, dbname_, cf_descs, &handles, &db_)
Corruption: SST file is ahead of WALs in CF test_cf
[ FAILED ] CorruptionTest/CrashDuringRecoveryWithCorruptionTest.CrashDuringRecovery/0, where GetParam() = (true, false) (91 ms)
[ RUN ] CorruptionTest/CrashDuringRecoveryWithCorruptionTest.CrashDuringRecovery/1
db/corruption_test.cc:1190: Failure
DB::Open(options, dbname_, cf_descs, &handles, &db_)
Corruption: SST file is ahead of WALs in CF test_cf
[ FAILED ] CorruptionTest/CrashDuringRecoveryWithCorruptionTest.CrashDuringRecovery/1, where GetParam() = (false, false) (92 ms)
[ RUN ] CorruptionTest/CrashDuringRecoveryWithCorruptionTest.CrashDuringRecovery/2
db/corruption_test.cc:1190: Failure
DB::Open(options, dbname_, cf_descs, &handles, &db_)
Corruption: SST file is ahead of WALs in CF test_cf
[ FAILED ] CorruptionTest/CrashDuringRecoveryWithCorruptionTest.CrashDuringRecovery/2, where GetParam() = (true, true) (95 ms)
[ RUN ] CorruptionTest/CrashDuringRecoveryWithCorruptionTest.CrashDuringRecovery/3
db/corruption_test.cc:1190: Failure
DB::Open(options, dbname_, cf_descs, &handles, &db_)
Corruption: SST file is ahead of WALs in CF test_cf
[ FAILED ] CorruptionTest/CrashDuringRecoveryWithCorruptionTest.CrashDuringRecovery/3, where GetParam() = (false, true) (92 ms)
[ RUN ] CorruptionTest/CrashDuringRecoveryWithCorruptionTest.TxnDbCrashDuringRecovery/0
db/corruption_test.cc:1354: Failure
TransactionDB::Open(options, txn_db_opts, dbname_, cf_descs, &handles, &txn_db)
Corruption: SST file is ahead of WALs in CF default
[ FAILED ] CorruptionTest/CrashDuringRecoveryWithCorruptionTest.TxnDbCrashDuringRecovery/0, where GetParam() = (true, false) (94 ms)
[ RUN ] CorruptionTest/CrashDuringRecoveryWithCorruptionTest.TxnDbCrashDuringRecovery/1
db/corruption_test.cc:1354: Failure
TransactionDB::Open(options, txn_db_opts, dbname_, cf_descs, &handles, &txn_db)
Corruption: SST file is ahead of WALs in CF default
[ FAILED ] CorruptionTest/CrashDuringRecoveryWithCorruptionTest.TxnDbCrashDuringRecovery/1, where GetParam() = (false, false) (97 ms)
[ RUN ] CorruptionTest/CrashDuringRecoveryWithCorruptionTest.TxnDbCrashDuringRecovery/2
db/corruption_test.cc:1354: Failure
TransactionDB::Open(options, txn_db_opts, dbname_, cf_descs, &handles, &txn_db)
Corruption: SST file is ahead of WALs in CF default
[ FAILED ] CorruptionTest/CrashDuringRecoveryWithCorruptionTest.TxnDbCrashDuringRecovery/2, where GetParam() = (true, true) (94 ms)
[ RUN ] CorruptionTest/CrashDuringRecoveryWithCorruptionTest.TxnDbCrashDuringRecovery/3
db/corruption_test.cc:1354: Failure
TransactionDB::Open(options, txn_db_opts, dbname_, cf_descs, &handles, &txn_db)
Corruption: SST file is ahead of WALs in CF default
[ FAILED ] CorruptionTest/CrashDuringRecoveryWithCorruptionTest.TxnDbCrashDuringRecovery/3, where GetParam() = (false, true) (91 ms)
[ RUN ] CorruptionTest/CrashDuringRecoveryWithCorruptionTest.CrashDuringRecoveryWithFlush/0
db/corruption_test.cc:1483: Failure
DB::Open(options, dbname_, cf_descs, &handles, &db_)
Corruption: SST file is ahead of WALs in CF default
[ FAILED ] CorruptionTest/CrashDuringRecoveryWithCorruptionTest.CrashDuringRecoveryWithFlush/0, where GetParam() = (true, false) (93 ms)
[ RUN ] CorruptionTest/CrashDuringRecoveryWithCorruptionTest.CrashDuringRecoveryWithFlush/1
db/corruption_test.cc:1483: Failure
DB::Open(options, dbname_, cf_descs, &handles, &db_)
Corruption: SST file is ahead of WALs in CF default
[ FAILED ] CorruptionTest/CrashDuringRecoveryWithCorruptionTest.CrashDuringRecoveryWithFlush/1, where GetParam() = (false, false) (94 ms)
[ RUN ] CorruptionTest/CrashDuringRecoveryWithCorruptionTest.CrashDuringRecoveryWithFlush/2
db/corruption_test.cc:1483: Failure
DB::Open(options, dbname_, cf_descs, &handles, &db_)
Corruption: SST file is ahead of WALs in CF default
[ FAILED ] CorruptionTest/CrashDuringRecoveryWithCorruptionTest.CrashDuringRecoveryWithFlush/2, where GetParam() = (true, true) (90 ms)
[ RUN ] CorruptionTest/CrashDuringRecoveryWithCorruptionTest.CrashDuringRecoveryWithFlush/3
db/corruption_test.cc:1483: Failure
DB::Open(options, dbname_, cf_descs, &handles, &db_)
Corruption: SST file is ahead of WALs in CF default
[ FAILED ] CorruptionTest/CrashDuringRecoveryWithCorruptionTest.CrashDuringRecoveryWithFlush/3, where GetParam() = (false, true) (93 ms)
[----------] 12 tests from CorruptionTest/CrashDuringRecoveryWithCorruptionTest (1116 ms total)
```
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9942
Test Plan: Not needed
Reviewed By: riversand963
Differential Revision: D36324112
Pulled By: akankshamahajan15
fbshipit-source-id: cab2075ac4ebe48f5ef93a6ea162558aa4fc334d
Summary:
Condense down to 8 groups rather than 20+ for ease of browsing
pages like
https://app.circleci.com/pipelines/github/facebook/rocksdb?branch=main&filter=all
Also, run nightly builds at 1AM or 2AM Pacific (depending on daylight
time) rather than 4PM or 5PM Pacific, so that they actually use each
day's landed changes.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9981
Test Plan:
CI
And manually inspected
```
grep -Eo 'build-[^: ]*' .circleci/config.yml | sort | uniq -c | less
```
to ensure I didn't orphan anything
Reviewed By: jay-zhuang
Differential Revision: D36317634
Pulled By: pdillinger
fbshipit-source-id: 1c10d29d6b5d60ce3dd1364cd91f175380075ff3
Summary:
Example failure when compiling on sufficiently new hardware and built-in headers:
```
In file included from /usr/local/lib/gcc/x86_64-linux-gnu/12.1.0/include/immintrin.h:49,
from ./util/bloom_impl.h:21,
from table/block_based/filter_policy.cc:31,
from unity.cc:167:
In function '__m512i _mm512_shuffle_epi32(__m512i, _MM_PERM_ENUM)',
inlined from 'void XXH3_accumulate_512_avx512(void*, const void*, const void*)' at util/xxhash.h:3605:58,
inlined from 'void XXH3_accumulate(xxh_u64*, const xxh_u8*, const xxh_u8*, size_t, XXH3_f_accumulate_512)' at util/xxhash.h:4229:17,
inlined from 'void XXH3_hashLong_internal_loop(xxh_u64*, const xxh_u8*, size_t, const xxh_u8*, size_t, XXH3_f_accumulate_512, XXH3_f_scrambleAcc)' at util/xxhash.h:4251:24,
inlined from 'XXH128_hash_t XXH3_hashLong_128b_internal(const void*, size_t, const xxh_u8*, size_t, XXH3_f_accumulate_512, XXH3_f_scrambleAcc)' at util/xxhash.h:5065:32,
inlined from 'XXH128_hash_t XXH3_hashLong_128b_withSecret(const void*, size_t, XXH64_hash_t, const void*, size_t)' at util/xxhash.h:5104:39:
/usr/local/lib/gcc/x86_64-linux-gnu/12.1.0/include/avx512fintrin.h:4459:50: error: '__Y' may be used uninitialized [-Werror=maybe-uninitialized]
```
https://app.circleci.com/pipelines/github/facebook/rocksdb/13295/workflows/1695fb5c-40c1-423b-96b4-45107dc3012d/jobs/360416
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9978
Test Plan:
I was able to re-run in CircleCI with ssh, see the failure, ssh in and
verify that adding -fno-avx512f fixed the failure. Will watch build-linux-unity-and-headers
Reviewed By: riversand963
Differential Revision: D36296028
Pulled By: pdillinger
fbshipit-source-id: ba5955cf2ac730f57d1d18c2f517e92f34be77a3
Summary:
CircleCI was using a soft open file limit of 1024 which would
frequently be exceeded during test runs. Now using
```
ulimit -S -n `ulimit -H -n`
```
to set soft limit up to the hard limit (524288 in my test). I've also
applied this same idiom to existing applicable MacOS configurations to
reduce hard-coding numbers.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9972
Test Plan: CI
Reviewed By: riversand963
Differential Revision: D36262943
Pulled By: pdillinger
fbshipit-source-id: 86320cdf9b68a97fdb73531a7b4a59b4c2d2f73f
Summary:
The new microbenchmarks, DBGetMergeOperandsInMemtable and DBGetMergeOperandsInSstFile, correspond to the two different LSMs tested: all data in one memtable and all data in one SST file, respectively. Both cases are parameterized by thread count (1 or 8) and merge operands per key (1, 32, or 1024). The SST file case is additionally parameterized by whether data is in block cache or mmap'd memory.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9971
Test Plan:
```
$ TEST_TMPDIR=/dev/shm/db_basic_bench/ ./db_basic_bench --benchmark_filter=DBGetMergeOperands
The number of inputs is very large. DBGet will be repeated at least 192 times.
The number of inputs is very large. DBGet will be repeated at least 192 times.
2022-05-09T13:15:40-07:00
Running ./db_basic_bench
Run on (36 X 2570.91 MHz CPU s)
CPU Caches:
L1 Data 32 KiB (x18)
L1 Instruction 32 KiB (x18)
L2 Unified 1024 KiB (x18)
L3 Unified 25344 KiB (x1)
Load Average: 4.50, 4.33, 4.37
----------------------------------------------------------------------------------------------------------------------------
Benchmark Time CPU Iterations UserCounters...
----------------------------------------------------------------------------------------------------------------------------
DBGetMergeOperandsInMemtable/entries_per_key:1/threads:1 846 ns 846 ns 849893 db_size=0
DBGetMergeOperandsInMemtable/entries_per_key:32/threads:1 2436 ns 2436 ns 305779 db_size=0
DBGetMergeOperandsInMemtable/entries_per_key:1024/threads:1 77226 ns 77224 ns 8152 db_size=0
DBGetMergeOperandsInMemtable/entries_per_key:1/threads:8 116 ns 929 ns 779368 db_size=0
DBGetMergeOperandsInMemtable/entries_per_key:32/threads:8 330 ns 2644 ns 280824 db_size=0
DBGetMergeOperandsInMemtable/entries_per_key:1024/threads:8 12466 ns 99718 ns 7200 db_size=0
DBGetMergeOperandsInSstFile/entries_per_key:1/mmap:0/threads:1 1640 ns 1640 ns 461262 db_size=21.7826M
DBGetMergeOperandsInSstFile/entries_per_key:1/mmap:1/threads:1 1693 ns 1693 ns 439936 db_size=21.7826M
DBGetMergeOperandsInSstFile/entries_per_key:32/mmap:0/threads:1 3999 ns 3999 ns 172881 db_size=19.6981M
DBGetMergeOperandsInSstFile/entries_per_key:32/mmap:1/threads:1 5544 ns 5543 ns 135657 db_size=19.6981M
DBGetMergeOperandsInSstFile/entries_per_key:1024/mmap:0/threads:1 78767 ns 78761 ns 8395 db_size=19.6389M
DBGetMergeOperandsInSstFile/entries_per_key:1024/mmap:1/threads:1 157242 ns 157238 ns 4495 db_size=19.6389M
DBGetMergeOperandsInSstFile/entries_per_key:1/mmap:0/threads:8 231 ns 1848 ns 347768 db_size=21.7826M
DBGetMergeOperandsInSstFile/entries_per_key:1/mmap:1/threads:8 214 ns 1715 ns 393312 db_size=21.7826M
DBGetMergeOperandsInSstFile/entries_per_key:32/mmap:0/threads:8 596 ns 4767 ns 142088 db_size=19.6981M
DBGetMergeOperandsInSstFile/entries_per_key:32/mmap:1/threads:8 720 ns 5757 ns 118200 db_size=19.6981M
DBGetMergeOperandsInSstFile/entries_per_key:1024/mmap:0/threads:8 11613 ns 92460 ns 7344 db_size=19.6389M
DBGetMergeOperandsInSstFile/entries_per_key:1024/mmap:1/threads:8 19989 ns 159908 ns 4440 db_size=19.6389M
```
Reviewed By: jay-zhuang
Differential Revision: D36258861
Pulled By: ajkr
fbshipit-source-id: 04b733e1cc3a4a70ed9baa894c50fdf96c0d6064
Summary:
https://github.com/facebook/rocksdb/issues/9961 broke format_compatible check because of `make clean`
referencing TEST_TMPDIR. The Makefile behavior seems reasonable to me,
so here's a fix in check_format_compatible.sh
Apparently I also included removing a redundant part of our CircleCI config.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9970
Test Plan: manual run: SHORT_TEST=1 ./tools/check_format_compatible.sh
Reviewed By: riversand963
Differential Revision: D36258172
Pulled By: pdillinger
fbshipit-source-id: d46507f04614e888b414ff23b88d040ae2b5c294
Summary:
Removing unnecessary checks around conversion from int/long to double as it does not lose information (see https://docs.oracle.com/javase/specs/jls/se9/html/jls-5.html#jls-5.1.2).
For example, `value > Double.MAX_VALUE` is always false when value is long or int.
Can you please have a look adamretter? Also fixed some other minor issues (do you prefer a separate PR?)
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9194
Reviewed By: ajkr
Differential Revision: D36221694
fbshipit-source-id: bf327c07386560b87ddc0c98039e8d6e8f2f1e82
Summary:
- For entry charge, we should only calculate the value size instead of including key size in LRUCache
- The capacity of string could show the memory usage precisely
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9337
Reviewed By: ajkr
Differential Revision: D36219855
fbshipit-source-id: 393e48ca419d230dc552ae62dd0eb1cc9f45961d
Summary:
Improve memkind library detection in build_detect_platform:
- The current position of -lmemkind does not work with all versions of gcc
- LDFLAGS allows specifying non-standard library path through EXTRA_LDFLAGS
After the change, the options match TBB detection.
This is a follow-up to https://github.com/facebook/rocksdb/issues/6214.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9134
Reviewed By: ajkr, mrambacher
Differential Revision: D32192028
fbshipit-source-id: 115fafe8d93f1fe6aaf80afb32b2cb67aad074c7
Summary:
Just fixing a very minor typo in the doc block :) Hope it will help anyway 😊
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9331
Reviewed By: riversand963
Differential Revision: D34339823
fbshipit-source-id: b76104bc3efbc9d1f38cbf5c6dd7648dc909ced3
Summary:
Having all of TMPD, TMPDIR and TEST_TMPDIR as configuration
parameters is confusing. This change simplifies a number of things by
standardizing on TEST_TMPDIR, while still recognizing the old names
also. In detail:
* crash_test.mk also needs to use TEST_TMPDIR for crash test, so put in
shared common.mk (an upgrade of python.mk)
* Always exporting TEST_TMPDIR eliminates the need to propagate TMPD or
export TEST_TMPDIR in selective places.
* Use --tmpdir option to gnu_parallel so that it doesn't need TMPDIR
environment variable
* Remove obsolete parloop and parallel_check Makefile targets
* Remove undefined, unused function ResetTmpDirForDirectIO()
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9961
Test Plan: manual + CI
Reviewed By: riversand963
Differential Revision: D36212178
Pulled By: pdillinger
fbshipit-source-id: b76c1876c4f4d38b37789c2779eaa7c3026824dd
Summary:
ToString() is created as some platform doesn't support std::to_string(). However, we've already used std::to_string() by mistake for 16 months (in db/db_info_dumper.cc). This commit just remove ToString().
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9955
Test Plan: Watch CI tests
Reviewed By: riversand963
Differential Revision: D36176799
fbshipit-source-id: bdb6dcd0e3a3ab96a1ac810f5d0188f684064471
Summary:
Previously all fault injection was ignored in release mode. This PR adds it back except for read fault injection (`--read_fault_one_in > 0`) since its dependency (`IGNORE_STATUS_IF_ERROR`) is unavailable in release mode.
Other notable changes include:
- Moved `EnableWriteErrorInjection()` for `--write_fault_one_in > 0` so it's after `DB::Open()` without depending on `SyncPoint`
- Made `--read_fault_one_in > 0` return an error in release mode
- Updated `db_crashtest.py` to always set `--read_fault_one_in=0` in release mode
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9957
Test Plan:
```
$ DEBUG_LEVEL=0 make -j24 db_stress
$ DEBUG_LEVEL=0 TEST_TMPDIR=/dev/shm python3 tools/db_crashtest.py blackbox
```
Reviewed By: anand1976
Differential Revision: D36193830
Pulled By: ajkr
fbshipit-source-id: 0b97946b4e3f06e3e0f6e7833c2763da08ec5321
Summary:
dont -> don't
refered -> referred
This is a re-run of PR#7785 and acc9679 since these typos keep coming back.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9653
Reviewed By: jay-zhuang
Differential Revision: D34879593
fbshipit-source-id: d7631fb779ea0129beae92abfb838038e60790f8
Summary:
`db_stress` already tracks expected state history to verify prefix-recoverability when `sync_fault_injection` is enabled. This PR enables `sync_fault_injection` in `db_crashtest.py`.
Previously enabling `sync_fault_injection` would cause whole unsynced files to be dropped. This PR adds a more interesting case of losing only the tail of unsynced data by implementing `TestFSWritableFile::RangeSync()` and enabling `{wal_,}bytes_per_sync`.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9947
Test Plan:
- regular blackbox, blackbox --simple
- various commands to stress this new case, such as `TEST_TMPDIR=/dev/shm python3 tools/db_crashtest.py blackbox --max_key=100000 --write_buffer_size=2097152 --avoid_flush_during_recovery=1 --disable_wal=0 --interval=10 --db_write_buffer_size=0 --sync_fault_injection=1 --wal_compression=none --delpercent=0 --delrangepercent=0 --prefixpercent=0 --iterpercent=0 --writepercent=100 --readpercent=0 --wal_bytes_per_sync=131072 --duration=36000 --sync=0 --open_write_fault_one_in=16`
Reviewed By: riversand963
Differential Revision: D36152775
Pulled By: ajkr
fbshipit-source-id: 44b68a7fad0a4cf74af9fe1f39be01baab8141d8
Summary:
Right now we still don't fully use std::numeric_limits but use a macro, mainly for supporting VS 2013. Right now we only support VS 2017 and up so it is not a problem. The code comment claims that MinGW still needs it. We don't have a CI running MinGW so it's hard to validate. since we now require C++17, it's hard to imagine MinGW would still build RocksDB but doesn't support std::numeric_limits<>.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9954
Test Plan: See CI Runs.
Reviewed By: riversand963
Differential Revision: D36173954
fbshipit-source-id: a35a73af17cdcae20e258cdef57fcf29a50b49e0
Summary:
PR 9929 adds a new CompactionFilter::Decision, i.e.
kRemoveWithSingleDelete so that CompactionFilter can indicate to
CompactionIterator that a PUT can only be removed with SD. However, how
CompactionIterator handles such a key is implementation detail which
should not be implied in the public API. In fact,
such a PUT can just be dropped. This is an optimization which we will apply in the near future.
Discussion thread: https://github.com/facebook/rocksdb/pull/9929#discussion_r863198964
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9951
Test Plan: make check
Reviewed By: ajkr
Differential Revision: D36156590
Pulled By: riversand963
fbshipit-source-id: 7b7d01f47bba4cad7d9cca6ca52984f27f88b372
Summary:
Right now in DumpDBFileSummary, IO error isn't printed out, but they are sometimes helpful. Print it out instead.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9940
Test Plan: Watch existing tests to pass.
Reviewed By: riversand963
Differential Revision: D36113016
fbshipit-source-id: 13002080fa4dc76589e2c1c5a1079df8a3c9391c
Summary:
This is inspired by debugging a regression test that runs for ~0.05 seconds and the short
running time makes it prone to variance. While db_bench ran for ~60 seconds, 59.95 seconds
was spent opening 128 databases (and doing recovery). So it was harder to notice that the
benchmark only ran for 0.05 seconds.
Normally I add output to the end of the line to make life easier for existing tools that parse it
but in this case the output near the end of the line has two optional parts and one of the optional
parts adds an extra newline.
This is for https://github.com/facebook/rocksdb/issues/9856
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9886
Test Plan:
./db_bench --benchmarks=overwrite,readrandom --num=1000000 --threads=4
old output:
DB path: [/tmp/rocksdbtest-2260/dbbench]
overwrite : 14.108 micros/op 283338 ops/sec; 31.3 MB/s
DB path: [/tmp/rocksdbtest-2260/dbbench]
readrandom : 7.994 micros/op 496788 ops/sec; 55.0 MB/s (1000000 of 1000000 found)
new output:
DB path: [/tmp/rocksdbtest-2260/dbbench]
overwrite : 14.117 micros/op 282862 ops/sec 14.141 seconds 4000000 operations; 31.3 MB/s
DB path: [/tmp/rocksdbtest-2260/dbbench]
readrandom : 8.649 micros/op 458475 ops/sec 8.725 seconds 4000000 operations; 49.8 MB/s (981548 of 1000000 found)
Reviewed By: ajkr
Differential Revision: D36102269
Pulled By: mdcallag
fbshipit-source-id: 5cd8a9e11f5cbe2a46809571afd83335b6b0caa0
Summary:
When a memtable is flushed and the flush would lead to a 0 byte .sst
file being created, RocksDB does not write out the empty .sst file to
disk.
However it still calls Env::DeleteFile() on the file as part of some
cleanup procedure at the end of BuildTable().
Because the to-be-deleted file does not exist, this requires
implementors of the DeleteFile() API to check if the file exists on
their own code, or otherwise risk running into PathNotFound errors when
DeleteFile is invoked on non-existing files.
This PR fixes the situation so that when no .sst file is created,
Deletefile will not be called either.
TableFileCreationStarted() will still be called as before.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9920
Reviewed By: ajkr
Differential Revision: D36107102
Pulled By: riversand963
fbshipit-source-id: 15881ba3fa3192dd448f906280a1cfc7a68a114a
Summary:
If the DB path is specified, the user would expect ldb loads the
options from the path, but it's not:
```
$ ldb list_live_files_metadata --db=`pwd`
```
Default `try_load_options` to true in that case. The user can still
disable that by:
```
$ ldb list_live_files_metadata --db=`pwd` --try_load_options=false
```
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9937
Test Plan:
`ldb list_live_files_metadata --db=`pwd`` is able to work for
a db generated with different options.num_levels.
Reviewed By: ajkr
Differential Revision: D36106708
Pulled By: jay-zhuang
fbshipit-source-id: 2732fdc027a4d172436b2c9b6a9787b56b10c710
Summary:
This PR solves the problem discussed in https://github.com/facebook/rocksdb/issues/7149. By storing the pointer of InternalKeyComparator as icmp_ in BlockIter, the object size remains the same. And for each call to CompareCurrentKey, there is no need to create Comparator objects. One can use icmp_ directly or use the "user_comparator" from the icmp_.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9611
Test Plan:
with https://github.com/facebook/rocksdb/issues/9903,
```
$ TEST_TMPDIR=/dev/shm python3.6 ../benchmark/tools/compare.py benchmarks ./db_basic_bench ../rocksdb-pr9611/db_basic_bench --benchmark_filter=DBGet/comp_style:0/max_data:134217728/per_key_size:256/enable_statistics:0/negative_query:0/enable_filter:0/mmap:1/iterations:262144/threads:1 --benchmark_repetitions=50
...
Comparing ./db_basic_bench to ../rocksdb-pr9611/db_basic_bench
Benchmark Time CPU Time Old Time New CPU Old CPU New
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
...
DBGet/comp_style:0/max_data:134217728/per_key_size:256/enable_statistics:0/negative_query:0/enable_filter:0/mmap:1/iterations:262144/threads:1_pvalue 0.0001 0.0001 U Test, Repetitions: 50 vs 50
DBGet/comp_style:0/max_data:134217728/per_key_size:256/enable_statistics:0/negative_query:0/enable_filter:0/mmap:1/iterations:262144/threads:1_mean -0.0483 -0.0483 3924 3734 3924 3734
DBGet/comp_style:0/max_data:134217728/per_key_size:256/enable_statistics:0/negative_query:0/enable_filter:0/mmap:1/iterations:262144/threads:1_median -0.0713 -0.0713 3971 3687 3970 3687
DBGet/comp_style:0/max_data:134217728/per_key_size:256/enable_statistics:0/negative_query:0/enable_filter:0/mmap:1/iterations:262144/threads:1_stddev -0.0342 -0.0344 225 217 225 217
DBGet/comp_style:0/max_data:134217728/per_key_size:256/enable_statistics:0/negative_query:0/enable_filter:0/mmap:1/iterations:262144/threads:1_cv +0.0148 +0.0146 0 0 0 0
OVERALL_GEOMEAN -0.0483 -0.0483 0 0 0 0
```
Reviewed By: akankshamahajan15
Differential Revision: D35882037
Pulled By: ajkr
fbshipit-source-id: 9e5337bbad8f1239dff7aa9f6549020d599bfcdf
Summary:
It confused users and use that with options.allow_mmap_reads = true, CPU is high with checksum verification. Add a comment to explain it.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9936
Reviewed By: anand1976
Differential Revision: D36106529
fbshipit-source-id: 3d723bd686f96a84c694c8b2d91ad28d9ccfd979
Summary:
Use `unique_ptr<DB>` to make sure the DB object is deleted. Previously it was not, which led to accumulating file descriptors for deleted directories because a `DBImpl::db_dir_` from each test remained alive.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9939
Test Plan: run `lsof -p $(pidof db_basic_bench)` while benchmark runs; verify no FDs for deleted directories.
Reviewed By: jay-zhuang
Differential Revision: D36108761
Pulled By: ajkr
fbshipit-source-id: cfe02646b038a445af7d5db8989eb1f40d658359
Summary:
To support a project to prototype and evaluate algorithmic
enhancments and alternatives to LRUCache, here I have separated out
LRUCache into internal-only "FastLRUCache" and cut it down to
essentials, so that details like secondary cache handling and
priorities do not interfere with prototyping. These can be
re-integrated later as needed, along with refactoring to minimize code
duplication (which would slow down prototyping for now).
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9917
Test Plan:
unit tests updated to ensure basic functionality has (likely)
been preserved
Reviewed By: anand1976
Differential Revision: D35995554
Pulled By: pdillinger
fbshipit-source-id: d67b20b7ada3b5d3bfe56d897a73885894a1d9db
Summary:
When compaction filter determines that a key should be removed, it updates the internal key's type
to `Delete`. If this internal key is preserved in current compaction but seen by a later compaction
together with `SingleDelete`, it will cause compaction iterator to return Corruption.
To fix the issue, compaction filter should return more information in addition to the intention of removing
a key. Therefore, we add a new `kRemoveWithSingleDelete` to `CompactionFilter::Decision`. Seeing
`kRemoveWithSingleDelete`, compaction iterator will update the op type of the internal key to `kTypeSingleDelete`.
In addition, I updated db_stress_shared_state.[cc|h] so that `no_overwrite_ids_` becomes `const`. It is easier to
reason about thread-safety if accessed from multiple threads. This information is passed to `PrepareTxnDBOptions()`
when calling from `Open()` so that we can set up the rollback deletion type callback for transactions.
Finally, disable compaction filter for multiops_txn because the key removal logic of `DbStressCompactionFilter` does
not quite work with `MultiOpsTxnsStressTest`.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9929
Test Plan:
make check
make crash_test
make crash_test_with_txn
Reviewed By: anand1976
Differential Revision: D36069678
Pulled By: riversand963
fbshipit-source-id: cedd2f1ba958af59ad3916f1ba6f424307955f92
Summary:
`VerifyChecksum()` does not specify `largest_seqno` when creating a `TableReader`. As a result, the `TableReader` uses the `TableReaderOptions` default value (0) for `largest_seqno`. This causes the following error when the file has a nonzero global seqno in its properties:
```
Corruption: An external sst file with version 2 have global seqno property with value , while largest seqno in the file is 0
```
This PR fixes this by specifying `largest_seqno` in `VerifyChecksumInternal` with `largest_seqno` from the file metadata.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9919
Test Plan: `make check`
Reviewed By: ajkr
Differential Revision: D36028824
Pulled By: cbi42
fbshipit-source-id: 428d028a79386f46ef97bb6b6051dc76c83e1f2b
Summary:
Enforce the contract of SingleDelete so that they are not mixed with
Delete for the same key. Otherwise, it will lead to undefined behavior.
See https://github.com/facebook/rocksdb/wiki/Single-Delete#notes.
Also fix unit tests and write-unprepared.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9888
Test Plan: make check
Reviewed By: ajkr
Differential Revision: D35837817
Pulled By: riversand963
fbshipit-source-id: acd06e4dcba8cb18df92b44ed18c57e10e5a7635
Summary:
Adds more coverage to `MultiOpsTxnsStressTest` with a focus on write-prepared transactions.
1. Add a hack to manually evict commit cache entries. We currently cannot assign small values to `wp_commit_cache_bits` because it requires a prepared transaction to commit within a certain range of sequence numbers, otherwise it will throw.
2. Add coverage for commit-time-write-batch. If write policy is write-prepared, we need to set `use_only_the_last_commit_time_batch_for_recovery` to true.
3. After each flush/compaction, verify data consistency. This is possible since data size can be small: default numbers of primary/secondary keys are just 1000.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9829
Test Plan:
```
TEST_TMPDIR=/dev/shm/rocksdb_crashtest_blackbox/ make blackbox_crash_test_with_multiops_wp_txn
```
Reviewed By: pdillinger
Differential Revision: D35806678
Pulled By: riversand963
fbshipit-source-id: d7fde7a29fda0fb481a61f553e0ca0c47da93616
Summary:
The current locktree implementation stores the address of the
PessimisticTransactions object as the TXNID. However, when a transaction
is blocked on a lock, it records the list of waitees with conflicting
locks using the rocksdb assigned TransactionID. This is performed by
calling GetID() on PessimisticTransactions objects of the waitees,
and then recorded in the waiter's list.
However, there is no guarantee the objects are valid when recording the
waitee list during the conflict callbacks because the waitee
could have released the lock and freed the PessimisticTransactions
object.
The waitee/txnid values are only valid PessimisticTransaction objects
while the mutex for the root of the locktree is held.
The simplest fix for this problem is to use the address of the
PessimisticTransaction as the TransactionID so that it is consistent
with its usage in the locktree. The TXNID is only converted back to a
PessimisticTransaction for the report_wait callbacks. Since
these callbacks are now all made within the critical section where the
lock_request queue mutx is held, these conversions will be safe.
Otherwise, only the uint64_t TXNID of the waitee is registerd
with the waiter transaction. The PessimisitcTransaction object of the
waitee is never referenced.
The main downside of this approach is the TransactionID will not change
if the PessimisticTransaction object is reused for new transactions.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9898
Test Plan:
Add a new test case and run unit tests.
Also verified with MyRocks workloads using range locks that the
crash no longer happens.
Reviewed By: riversand963
Differential Revision: D35950376
Pulled By: hermanlee
fbshipit-source-id: 8c9cae272e23e487fc139b6a8ed5b8f8f24b1570
Summary:
When MultiGet() determines that multiple query keys can be
served by examining the same data block in block cache (one Lookup()),
each PinnableSlice referring to data in that data block needs to hold
on to the block in cache so that they can be released at arbitrary
times by the API user. Historically this is accomplished with extra
calls to Ref() on the Handle from Lookup(), with each PinnableSlice
cleanup calling Release() on the Handle, but this creates extra
contention on the block cache for the extra Ref()s and Release()es,
especially because they hit the same cache shard repeatedly.
In the case of merge operands (possibly more cases?), the problem was
compounded by doing an extra Ref()+eventual Release() for each merge
operand for a key reusing a block (which could be the same key!), rather
than one Ref() per key. (Note: the non-shared case with `biter` was
already one per key.)
This change optimizes MultiGet not to rely on these extra, contentious
Ref()+Release() calls by instead, in the shared block case, wrapping
the cache Release() cleanup in a refcounted object referenced by the
PinnableSlices, such that after the last wrapped reference is released,
the cache entry is Release()ed. Relaxed atomic refcounts should be
much faster than mutex-guarded Ref() and Release(), and much less prone
to a performance cliff when MultiGet() does a lot of block sharing.
Note that I did not use std::shared_ptr, because that would require an
extra indirection object (shared_ptr itself new/delete) in order to
associate a ref increment/decrement with a Cleanable cleanup entry. (If
I assumed it was the size of two pointers, I could do some hackery to
make it work without the extra indirection, but that's too fragile.)
Some details:
* Fixed (removed) extra block cache tracing entries in cases of cache
entry reuse in MultiGet, but it's likely that in some other cases traces
are missing (XXX comment inserted)
* Moved existing implementations for cleanable.h from iterator.cc to
new cleanable.cc
* Improved API comments on Cleanable
* Added a public SharedCleanablePtr class to cleanable.h in case others
could benefit from the same pattern (potentially many Cleanables and/or
smart pointers referencing a shared Cleanable)
* Add a typedef for MultiGetContext::Mask
* Some variable renaming for clarity
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9899
Test Plan:
Added unit tests for SharedCleanablePtr.
Greatly enhanced ability of existing tests to detect cache use-after-free.
* Release PinnableSlices from MultiGet as they are read rather than in
bulk (in db_test_util wrapper).
* In ASAN build, default to using a trivially small LRUCache for block_cache
so that entries are immediately erased when unreferenced. (Updated two
tests that depend on caching.) New ASAN testsuite running time seems
OK to me.
If I introduce a bug into my implementation where we skip the shared
cleanups on block reuse, ASAN detects the bug in
`db_basic_test *MultiGet*`. If I remove either of the above testing
enhancements, the bug is not detected.
Consider for follow-up work: manipulate or randomize ordering of
PinnableSlice use and release from MultiGet db_test_util wrapper. But in
typical cases, natural ordering gives pretty good functional coverage.
Performance test:
In the extreme (but possible) case of MultiGetting the same or adjacent keys
in a batch, throughput can improve by an order of magnitude.
`./db_bench -benchmarks=multireadrandom -db=/dev/shm/testdb -readonly -num=5 -duration=10 -threads=20 -multiread_batched -batch_size=200`
Before ops/sec, num=5: 1,384,394
Before ops/sec, num=500: 6,423,720
After ops/sec, num=500: 10,658,794
After ops/sec, num=5: 16,027,257
Also note that previously, with high parallelism, having query keys
concentrated in a single block was worse than spreading them out a bit. Now
concentrated in a single block is faster than spread out, which is hopefully
consistent with natural expectation.
Random query performance: with num=1000000, over 999 x 10s runs running before & after simultaneously (each -threads=12):
Before: multireadrandom [AVG 999 runs] : 1088699 (± 7344) ops/sec; 120.4 (± 0.8 ) MB/sec
After: multireadrandom [AVG 999 runs] : 1090402 (± 7230) ops/sec; 120.6 (± 0.8 ) MB/sec
Possibly better, possibly in the noise.
Reviewed By: anand1976
Differential Revision: D35907003
Pulled By: pdillinger
fbshipit-source-id: bbd244d703649a8ca12d476f2d03853ed9d1a17e
Summary:
I tried evaluating https://github.com/facebook/rocksdb/issues/9611 using DBGet microbenchmarks but mostly found the change is well within the noise even for hundreds of repetitions; meanwhile, the InternalKeyComparator CPU it saves is 1-2% according to perf so it should be measurable. In this PR I tried adding a mmap mode that will bypass compression/checksum/block cache/file read to focus more on the block lookup paths, and also increased the Get() count.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/9903
Reviewed By: jay-zhuang, riversand963
Differential Revision: D35907375
Pulled By: ajkr
fbshipit-source-id: 69490d5040ef0863e1ce296724104d0aa7667215
Akanksha Mahajan