Summary:
While trying to reuse PinData() / ReleasePinnedData() .. to optimize away some memcpys I realized that there is a significant overhead for using PinData() / ReleasePinnedData if they were called many times.
This diff refactor the pinning logic by introducing PinnedIteratorsManager a centralized component that will be created once and will be notified whenever we need to Pin an Iterator. This implementation have much less overhead than the original implementation
Test Plan:
make check -j64
COMPILE_WITH_ASAN=1 make check -j64
Reviewers: yhchiang, sdong, andrewkr
Reviewed By: andrewkr
Subscribers: andrewkr, dhruba
Differential Revision: https://reviews.facebook.net/D56493
Summary: DBImpl::GetLatestSequenceForKey() can do memcpy's to load a value that will never be used. This can be optimized by changing all the Get() functions called to optionally not fetch the value (and only fetch the sequencenumber).
Test Plan: optimistic_transaction_test and transaction_test
Reviewers: anthony
Reviewed By: anthony
Subscribers: leveldb, dhruba, hermanlee4
Differential Revision: https://reviews.facebook.net/D52227
Summary:
This diff adds support for concurrent adds to the skiplist memtable
implementations. Memory allocation is made thread-safe by the addition of
a spinlock, with small per-core buffers to avoid contention. Concurrent
memtable writes are made via an additional method and don't impose a
performance overhead on the non-concurrent case, so parallelism can be
selected on a per-batch basis.
Write thread synchronization is an increasing bottleneck for higher levels
of concurrency, so this diff adds --enable_write_thread_adaptive_yield
(default off). This feature causes threads joining a write batch
group to spin for a short time (default 100 usec) using sched_yield,
rather than going to sleep on a mutex. If the timing of the yield calls
indicates that another thread has actually run during the yield then
spinning is avoided. This option improves performance for concurrent
situations even without parallel adds, although it has the potential to
increase CPU usage (and the heuristic adaptation is not yet mature).
Parallel writes are not currently compatible with
inplace updates, update callbacks, or delete filtering.
Enable it with --allow_concurrent_memtable_write (and
--enable_write_thread_adaptive_yield). Parallel memtable writes
are performance neutral when there is no actual parallelism, and in
my experiments (SSD server-class Linux and varying contention and key
sizes for fillrandom) they are always a performance win when there is
more than one thread.
Statistics are updated earlier in the write path, dropping the number
of DB mutex acquisitions from 2 to 1 for almost all cases.
This diff was motivated and inspired by Yahoo's cLSM work. It is more
conservative than cLSM: RocksDB's write batch group leader role is
preserved (along with all of the existing flush and write throttling
logic) and concurrent writers are blocked until all memtable insertions
have completed and the sequence number has been advanced, to preserve
linearizability.
My test config is "db_bench -benchmarks=fillrandom -threads=$T
-batch_size=1 -memtablerep=skip_list -value_size=100 --num=1000000/$T
-level0_slowdown_writes_trigger=9999 -level0_stop_writes_trigger=9999
-disable_auto_compactions --max_write_buffer_number=8
-max_background_flushes=8 --disable_wal --write_buffer_size=160000000
--block_size=16384 --allow_concurrent_memtable_write" on a two-socket
Xeon E5-2660 @ 2.2Ghz with lots of memory and an SSD hard drive. With 1
thread I get ~440Kops/sec. Peak performance for 1 socket (numactl
-N1) is slightly more than 1Mops/sec, at 16 threads. Peak performance
across both sockets happens at 30 threads, and is ~900Kops/sec, although
with fewer threads there is less performance loss when the system has
background work.
Test Plan:
1. concurrent stress tests for InlineSkipList and DynamicBloom
2. make clean; make check
3. make clean; DISABLE_JEMALLOC=1 make valgrind_check; valgrind db_bench
4. make clean; COMPILE_WITH_TSAN=1 make all check; db_bench
5. make clean; COMPILE_WITH_ASAN=1 make all check; db_bench
6. make clean; OPT=-DROCKSDB_LITE make check
7. verify no perf regressions when disabled
Reviewers: igor, sdong
Reviewed By: sdong
Subscribers: MarkCallaghan, IslamAbdelRahman, anthony, yhchiang, rven, sdong, guyg8, kradhakrishnan, dhruba
Differential Revision: https://reviews.facebook.net/D50589
Summary:
This patch update the Iterator API to introduce new functions that allow users to keep the Slices returned by key() valid as long as the Iterator is not deleted
ReadOptions::pin_data : If true keep loaded blocks in memory as long as the iterator is not deleted
Iterator::IsKeyPinned() : If true, this mean that the Slice returned by key() is valid as long as the iterator is not deleted
Also add a new option BlockBasedTableOptions::use_delta_encoding to allow users to disable delta_encoding if needed.
Benchmark results (using https://phabricator.fb.com/P20083553)
```
// $ du -h /home/tec/local/normal.4K.Snappy/db10077
// 6.1G /home/tec/local/normal.4K.Snappy/db10077
// $ du -h /home/tec/local/zero.8K.LZ4/db10077
// 6.4G /home/tec/local/zero.8K.LZ4/db10077
// Benchmarks for shard db10077
// _build/opt/rocks/benchmark/rocks_copy_benchmark \
// --normal_db_path="/home/tec/local/normal.4K.Snappy/db10077" \
// --zero_db_path="/home/tec/local/zero.8K.LZ4/db10077"
// First run
// ============================================================================
// rocks/benchmark/RocksCopyBenchmark.cpp relative time/iter iters/s
// ============================================================================
// BM_StringCopy 1.73s 576.97m
// BM_StringPiece 103.74% 1.67s 598.55m
// ============================================================================
// Match rate : 1000000 / 1000000
// Second run
// ============================================================================
// rocks/benchmark/RocksCopyBenchmark.cpp relative time/iter iters/s
// ============================================================================
// BM_StringCopy 611.99ms 1.63
// BM_StringPiece 203.76% 300.35ms 3.33
// ============================================================================
// Match rate : 1000000 / 1000000
```
Test Plan: Unit tests
Reviewers: sdong, igor, anthony, yhchiang, rven
Reviewed By: rven
Subscribers: dhruba, lovro, adsharma
Differential Revision: https://reviews.facebook.net/D48999
Summary:
Separate a new class InternalIterator from class Iterator, when the look-up is done internally, which also means they operate on key with sequence ID and type.
This change will enable potential future optimizations but for now InternalIterator's functions are still the same as Iterator's.
At the same time, separate the cleanup function to a separate class and let both of InternalIterator and Iterator inherit from it.
Test Plan: Run all existing tests.
Reviewers: igor, yhchiang, anthony, kradhakrishnan, IslamAbdelRahman, rven
Reviewed By: rven
Subscribers: leveldb, dhruba
Differential Revision: https://reviews.facebook.net/D48549
Summary:
hit and miss bloom filter stats for memtable and SST
stats added to perf_context struct
key matches and prefix matches combined into one stat
Test Plan: unit test veryfing the functionality added, see BloomStatsTest in db_test.cc for details
Reviewers: yhchiang, igor, sdong
Reviewed By: sdong
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D47859
Summary:
This patch fixes#7460559. It introduces SingleDelete as a new database
operation. This operation can be used to delete keys that were never
overwritten (no put following another put of the same key). If an overwritten
key is single deleted the behavior is undefined. Single deletion of a
non-existent key has no effect but multiple consecutive single deletions are
not allowed (see limitations).
In contrast to the conventional Delete() operation, the deletion entry is
removed along with the value when the two are lined up in a compaction. Note:
The semantics are similar to @igor's prototype that allowed to have this
behavior on the granularity of a column family (
https://reviews.facebook.net/D42093 ). This new patch, however, is more
aggressive when it comes to removing tombstones: It removes the SingleDelete
together with the value whenever there is no snapshot between them while the
older patch only did this when the sequence number of the deletion was older
than the earliest snapshot.
Most of the complex additions are in the Compaction Iterator, all other changes
should be relatively straightforward. The patch also includes basic support for
single deletions in db_stress and db_bench.
Limitations:
- Not compatible with cuckoo hash tables
- Single deletions cannot be used in combination with merges and normal
deletions on the same key (other keys are not affected by this)
- Consecutive single deletions are currently not allowed (and older version of
this patch supported this so it could be resurrected if needed)
Test Plan: make all check
Reviewers: yhchiang, sdong, rven, anthony, yoshinorim, igor
Reviewed By: igor
Subscribers: maykov, dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D43179
Summary:
In some cases, equality comparisons can be done more efficiently than three-way
comparisons. There are quite a few places in the code where we only care about
equality. This patch adds an Equal() method that defaults to using the
Compare() method.
Test Plan: make clean all check
Reviewers: rven, anthony, yhchiang, igor, sdong
Reviewed By: igor
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D46233
Summary:
Flushes in listener_test happened to early when ROCKSDB_MALLOC_USABLE_SIZE was
active (e.g. when compiling with ROCKSDB_FBCODE_BUILD_WITH_481=1) due to
malloc_usable_size() reporting a better estimate (similar to
https://reviews.facebook.net/D43317 ). This patch grows the write buffer size
slightly to compensate for this.
Test Plan: ROCKSDB_FBCODE_BUILD_WITH_481=1 make listener_test && ./listener_test
Reviewers: rven, anthony, yhchiang, igor, sdong
Reviewed By: igor
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D45921
Summary:
While working on supporting mixing merge operators with
single deletes ( https://reviews.facebook.net/D43179 ),
I realized that returning and dealing with merge results
can be made simpler. Submitting this as a separate diff
because it is not directly related to single deletes.
Before, callers of merge helper had to retrieve the merge
result in one of two ways depending on whether the merge
was successful or not (success = result of merge was single
kTypeValue). For successful merges, the caller could query
the resulting key/value pair and for unsuccessful merges,
the result could be retrieved in the form of two deques of
keys and values. However, with single deletes, a successful merge
does not return a single key/value pair (if merge
operands are merged with a single delete, we have to generate
a value and keep the original single delete around to make
sure that we are not accidentially producing a key overwrite).
In addition, the two existing call sites of the merge
helper were taking the same actions independently from whether
the merge was successful or not, so this patch simplifies that.
Test Plan: make clean all check
Reviewers: rven, sdong, yhchiang, anthony, igor
Reviewed By: igor
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D43353
Summary:
Add an option in GetApproximateSize() so that the result will include estimated sizes in mem tables.
To implement it, implement an estimated count from the beginning to a key in skip list. The approach is to count to find the entry, how many Next() is issued from each level, and sum them with a weight that is <branching factor> ^ <level>.
Test Plan: Add a test case
Subscribers: leveldb, dhruba
Differential Revision: https://reviews.facebook.net/D40119
Summary: Optimistic transactions supporting begin/commit/rollback semantics. Currently relies on checking the memtable to determine if there are any collisions at commit time. Not yet implemented would be a way of enuring the memtable has some minimum amount of history so that we won't fail to commit when the memtable is empty. You should probably start with transaction.h to get an overview of what is currently supported.
Test Plan: Added a new test, but still need to look into stress testing.
Reviewers: yhchiang, igor, rven, sdong
Reviewed By: sdong
Subscribers: adamretter, MarkCallaghan, leveldb, dhruba
Differential Revision: https://reviews.facebook.net/D33435
Summary:
We have addded new stats and perf_context for measuring the merge and filter operation time consumption.
We have bounded all the merge operations within the GUARD statment and collected the total time for these operations in the DB.
Test Plan: WIP
Reviewers: rven, yhchiang, kradhakrishnan, igor, sdong
Reviewed By: sdong
Subscribers: dhruba
Differential Revision: https://reviews.facebook.net/D34377
Summary: Add a DB property for number of deletions in memtables. It can sometimes help people debug slowness because of too many deletes.
Test Plan: Add test cases.
Reviewers: rven, yhchiang, kradhakrishnan, igor
Reviewed By: igor
Subscribers: leveldb, dhruba, yoshinorim
Differential Revision: https://reviews.facebook.net/D35247
Summary:
When using latest clang (3.6 or 3.7/trunck) rocksdb is failing with many errors. Almost all of them are missing override errors. This diff adds missing override keyword. No manual changes.
Prerequisites: bear and clang 3.5 build with extra tools
```lang=bash
% USE_CLANG=1 bear make all # generate a compilation database http://clang.llvm.org/docs/JSONCompilationDatabase.html
% clang-modernize -p . -include . -add-override
% make format
```
Test Plan:
Make sure all tests are passing.
```lang=bash
% #Use default fb code clang.
% make check
```
Verify less error and no missing override errors.
```lang=bash
% # Have trunk clang present in path.
% ROCKSDB_NO_FBCODE=1 CC=clang CXX=clang++ make
```
Reviewers: igor, kradhakrishnan, rven, meyering, sdong
Reviewed By: sdong
Subscribers: dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D34077
Summary:
Introduces a new class for managing write buffer memory across column
families. We supplement ColumnFamilyOptions::write_buffer_size with
ColumnFamilyOptions::write_buffer, a shared pointer to a WriteBuffer
instance that enforces memory limits before flushing out to disk.
Test Plan: Added SharedWriteBuffer unit test to db_test.cc
Reviewers: sdong, rven, ljin, igor
Reviewed By: igor
Subscribers: tnovak, yhchiang, dhruba, xjin, MarkCallaghan, yoshinorim
Differential Revision: https://reviews.facebook.net/D22581
Summary:
We need to turn on -Wshorten-64-to-32 for mobile. See D1671432 (internal phabricator) for details.
This diff turns on the warning flag and fixes all the errors. There were also some interesting errors that I might call bugs, especially in plain table. Going forward, I think it makes sense to have this flag turned on and be very very careful when converting 64-bit to 32-bit variables.
Test Plan: compiles
Reviewers: ljin, rven, yhchiang, sdong
Reviewed By: yhchiang
Subscribers: bobbaldwin, dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D28689
Summary:
Make inplace_update_support and inplace_update_num_locks dynamic.
inplace_callback becomes immutable
We are almost free of references to cfd->options() in db_impl
Test Plan: unit test
Reviewers: igor, yhchiang, rven, sdong
Reviewed By: sdong
Subscribers: leveldb
Differential Revision: https://reviews.facebook.net/D25293
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
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
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
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
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
Summary:
Simply code by removing code path which does not use Arena
from NewInternalIterator
Test Plan:
make all check
make valgrind_check
Reviewers: sdong
Reviewed By: sdong
Subscribers: leveldb
Differential Revision: https://reviews.facebook.net/D22395
This eliminates the need to remember to call PERF_TIMER_STOP when a section has
been timed. This allows more useful design with the perf timers and enables
possible return value optimizations. Simplistic example:
class Foo {
public:
Foo(int v) : m_v(v);
private:
int m_v;
}
Foo makeFrobbedFoo(int *errno)
{
*errno = 0;
return Foo();
}
Foo bar(int *errno)
{
PERF_TIMER_GUARD(some_timer);
return makeFrobbedFoo(errno);
}
int main(int argc, char[] argv)
{
Foo f;
int errno;
f = bar(&errno);
if (errno)
return -1;
return 0;
}
After bar() is called, perf_context.some_timer would be incremented as if
Stop(&perf_context.some_timer) was called at the end, and the compiler is still
able to produce optimizations on the return value from makeFrobbedFoo() through
to main().
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
Summary:
Fixed the crash when merge_operator is not properly set after reopen
and added two test cases for this.
Test Plan:
make merge_test
./merge_test
Reviewers: igor, ljin, sdong
Reviewed By: sdong
Subscribers: benj, mvikjord, leveldb
Differential Revision: https://reviews.facebook.net/D20793
Summary:
Bloomfilter and hashskiplist's buckets_ allocated by memtable's arena
DynamicBloom: pass arena via constructor, allocate space in SetTotalBits
HashSkipListRep: allocate space of buckets_ using arena.
do not delete it in deconstructor because arena would take care of it.
Several test files are changed.
Test Plan:
make all check
Reviewers: ljin, haobo, yhchiang, sdong
Reviewed By: sdong
Subscribers: igor, dhruba
Differential Revision: https://reviews.facebook.net/D19335
Summary: It seems to me that when ever function MemTableRep::GetIterator(const Slice& slice) is used, we can use MemTableRep::GetDynamicPrefixIterator() instead. Just delete it to simplify the codes.
Test Plan: make all check
Reviewers: yhchiang, ljin
Reviewed By: ljin
Subscribers: xjin, dhruba, haobo, leveldb
Differential Revision: https://reviews.facebook.net/D19281
Some platforms, particularly Windows, do not have a single method that can
release both a held reader lock and a held writer lock; instead, a
separate method (ReleaseSRWLockShared or ReleaseSRWLockExclusive) must be
called in each case.
This may also be necessary to back MutexRW with a shared_mutex in C++14;
the current language proposal includes both an unlock() and a
shared_unlock() method.
Summary:
In this patch, try to allocate the whole iterator tree starting from DBIter from an arena
1. ArenaWrappedDBIter is created when serves as the entry point of an iterator tree, with an arena in it.
2. Add an option to create iterator from arena for following iterators: DBIter, MergingIterator, MemtableIterator, all mem table's iterators, all table reader's iterators and two level iterator.
3. MergeIteratorBuilder is created to incrementally build the tree of internal iterators. It is passed to mem table list and version set and add iterators to it.
Limitations:
(1) Only DB::NewIterator() without tailing uses the arena. Other cases, including readonly DB and compactions are still from malloc
(2) Two level iterator itself is allocated in arena, but not iterators inside it.
Test Plan: make all check
Reviewers: ljin, haobo
Reviewed By: haobo
Subscribers: leveldb, dhruba, yhchiang, igor
Differential Revision: https://reviews.facebook.net/D18513
Summary:
TLB page allocation errors are now logged to info logs, instead of stderr.
In order to do that, mem table rep's factory functions take a info logger now.
Test Plan: make all check
Reviewers: haobo, igor, yhchiang
Reviewed By: yhchiang
CC: leveldb, yhchiang, dhruba
Differential Revision: https://reviews.facebook.net/D18471
Summary: Add an option to allocate a piece of memory from huge page TLB. Add options to trigger it in dynamic bloom, plain table indexes andhash linked list hash table.
Test Plan: make all check
Reviewers: haobo, ljin
Reviewed By: haobo
CC: nkg-, dhruba, leveldb, igor, yhchiang
Differential Revision: https://reviews.facebook.net/D18357
Summary:
= Major Changes =
* Add a new mem-table representation, HashCuckooRep, which is based cuckoo hash.
Cuckoo hash uses multiple hash functions. This allows each key to have multiple
possible locations in the mem-table.
- Put: When insert a key, it will try to find whether one of its possible
locations is vacant and store the key. If none of its possible
locations are available, then it will kick out a victim key and
store at that location. The kicked-out victim key will then be
stored at a vacant space of its possible locations or kick-out
another victim. In this diff, the kick-out path (known as
cuckoo-path) is found using BFS, which guarantees to be the shortest.
- Get: Simply tries all possible locations of a key --- this guarantees
worst-case constant time complexity.
- Time complexity: O(1) for Get, and average O(1) for Put if the
fullness of the mem-table is below 80%.
- Default using two hash functions, the number of hash functions used
by the cuckoo-hash may dynamically increase if it fails to find a
short-enough kick-out path.
- Currently, HashCuckooRep does not support iteration and snapshots,
as our current main purpose of this is to optimize point access.
= Minor Changes =
* Add IsSnapshotSupported() to DB to indicate whether the current DB
supports snapshots. If it returns false, then DB::GetSnapshot() will
always return nullptr.
Test Plan:
Run existing tests. Will develop a test specifically for cuckoo hash in
the next diff.
Reviewers: sdong, haobo
Reviewed By: sdong
CC: leveldb, dhruba, igor
Differential Revision: https://reviews.facebook.net/D16155
Summary:
also add an override option total_order_iteration if you want to use full
iterator with prefix_extractor
Test Plan: make all check
Reviewers: igor, haobo, sdong, yhchiang
Reviewed By: haobo
CC: leveldb, dhruba
Differential Revision: https://reviews.facebook.net/D17805
Summary: In this patch, two new DB properties are defined: rocksdb.num-immutable-mem-table and rocksdb.num-entries-imm-mem-tables, from where number of entries in mem tables can be exposed to users
Test Plan:
Cover the codes in db_test
make all check
Reviewers: haobo, ljin, igor
Reviewed By: igor
CC: nkg-, igor, yhchiang, dhruba, leveldb
Differential Revision: https://reviews.facebook.net/D18207
Summary: Get rid of the devil. Probably won't impact anything on the perf side.
Test Plan: make all check
Reviewers: igor, haobo, sdong, yhchiang
Reviewed By: haobo
CC: leveldb, dhruba
Differential Revision: https://reviews.facebook.net/D18153
Summary: This will allow us to disable them completely for iOS or for better performance
Test Plan: will run make all check
Reviewers: igor, haobo, dhruba
Reviewed By: haobo
CC: leveldb
Differential Revision: https://reviews.facebook.net/D17511
Summary: per sdong's request, this will help processor prefetch on n->key case.
Test Plan: make all check
Reviewers: sdong, haobo, igor
Reviewed By: sdong
CC: leveldb
Differential Revision: https://reviews.facebook.net/D17415
Summary: In one of the perf, I shows 10%-25% CPU costs of MemTableIterator.Seek(), when doing dynamic prefix seek, are spent on checking whether we need to do bloom filter check or finding out the prefix extractor. Seems that more level of pointer checking makes CPU cache miss more likely. This patch makes things slightly simpler by copying pointer of bloom of prefix extractor into the iterator.
Test Plan: make all check
Reviewers: haobo, ljin
Reviewed By: ljin
CC: igor, dhruba, yhchiang, leveldb
Differential Revision: https://reviews.facebook.net/D17247
Summary:
By constraining the probes within cache line(s), we can improve the
cache miss rate thus performance. This probably only makes sense for
in-memory workload so defaults the option to off.
Numbers and comparision can be found in wiki:
https://our.intern.facebook.com/intern/wiki/index.php/Ljin/rocksdb_perf/2014_03_17#Bloom_Filter_Study
Test Plan: benchmarked this change substantially. Will run make all check as well
Reviewers: haobo, igor, dhruba, sdong, yhchiang
Reviewed By: haobo
CC: leveldb
Differential Revision: https://reviews.facebook.net/D17133
