Summary:
Based on my recent findings (posted in our internal group), if we use fallocate without KEEP_SIZE flag, we get superior performance of fdatasync() in append-only workloads.
This diff provides an option for user to not use KEEP_SIZE flag, thus optimizing his sync performance by up to 2x-3x.
At one point we also just called posix_fallocate instead of fallocate, which isn't very fast: http://code.woboq.org/userspace/glibc/sysdeps/posix/posix_fallocate.c.html (tl;dr it manually writes out zero bytes to allocate storage). This diff also fixes that, by first calling fallocate and then posix_fallocate if fallocate is not supported.
Test Plan: make check
Reviewers: dhruba, sdong, haobo, ljin
Reviewed By: dhruba
CC: leveldb
Differential Revision: https://reviews.facebook.net/D16761
Summary: Add a function to Env so that users can query the waiting queue length of each thread pool
Test Plan: add a test in env_test
Reviewers: haobo
Reviewed By: haobo
CC: dhruba, igor, yhchiang, ljin, nkg-, leveldb
Differential Revision: https://reviews.facebook.net/D16755
Summary:
Change to store the return value from ftruncate().
The reason is that ftruncate() has "warn_unused_result" attribute in some environment.
Signed-off-by: Yumikiyo Osanai <yumios.art@gmail.com>
Summary:
Blocks allocated with fallocate will take extra space on disk even if they are unused and the file is close.
Now we remove the extra blocks at the end of the file by calling `ftruncate`.
Test Plan: added a test to env_test
Reviewers: dhruba
Reviewed By: dhruba
CC: leveldb
Differential Revision: https://reviews.facebook.net/D16647
Summary:
This is not a generic thread local implementation in the sense that it
only takes pointer. But it does support multiple instances per thread
and lets user plugin function to perform cleanup when thread exits or an
instance gets destroyed.
Test Plan: unit test for now
Reviewers: haobo, igor, sdong, dhruba
Reviewed By: igor
CC: leveldb, kailiu
Differential Revision: https://reviews.facebook.net/D16131
Summary: Nothing major, just an extra return line and posibility of leaking fb in NewRandomRWFile
Test Plan: make check
Reviewers: kailiu, dhruba
Reviewed By: kailiu
CC: leveldb
Differential Revision: https://reviews.facebook.net/D15993
Summary:
@dhruba, I'm not sure where we need to sync the directory. I implemented the function in Env() and added the dir sync just after we close the newly created file in the builder.
Should I also add FsyncDir() to new files that get created by a compaction?
Test Plan: Confirmed that FsyncDir is returning Status::OK()
Reviewers: dhruba, haobo
Reviewed By: dhruba
CC: leveldb, dhruba
Differential Revision: https://reviews.facebook.net/D14751
This seems out of place as it's the only time RocksDB prints to stdout in the
normal course of operations. Thread IDs can still be retrieved from the LOG
file: cut -d ' ' -f2 LOG | sort | uniq | egrep -x '[0-9a-f]+'
Summary: The preprocessor does not follow normal rules of && evaluation, tries to evaluate __GLIBC_PREREQ(2, 12) even though the defined() check fails. This breaks the build if __GLIBC_PREREQ is absent.
Test Plan: Try adding #undef __GLIBC_PREREQ above the offending line, build no longer breaks
Reviewed By: igor
Blame Rev: 4c81383628
Summary: Makes it easier to monitor performance with top
Test Plan: ./manual_compaction_test with `top -H` running. Previously was two `manual_compacti`, now one shows `rocksdb:bg0`.
Reviewers: igor, dhruba
Reviewed By: igor
CC: leveldb
Differential Revision: https://reviews.facebook.net/D14367
Summary:
Fix a stupid bug I just introduced in b59d4d5a50, which I didn't even mean to include.
GCC might remove the munmap.
Test Plan: Run it and make sure munmap succeeds
Reviewers: haobo, kailiu
Reviewed By: kailiu
CC: dhruba, reconnect.grayhat, leveldb
Differential Revision: https://reviews.facebook.net/D14361
Summary:
A Simple plain table format. No block structure. When creating the table reader, scanning the full table to create indexes.
Test Plan:Add unit test
Reviewers:haobo,dhruba,kailiu
CC:
Task ID: #
Blame Rev:
Summary: This diff invoves some more complicated issues in the posix environment.
Test Plan: works under mac os. will need to verify dev box.
Reviewers: dhruba
Reviewed By: dhruba
CC: leveldb
Differential Revision: https://reviews.facebook.net/D14061
Summary:
Rocksdb can now support a uncompressed block cache, or a compressed
block cache or both. Lookups first look for a block in the
uncompressed cache, if it is not found only then it is looked up
in the compressed cache. If it is found in the compressed cache,
then it is uncompressed and inserted into the uncompressed cache.
It is possible that the same block resides in the compressed cache
as well as the uncompressed cache at the same time. Both caches
have their own individual LRU policy.
Test Plan: Unit test case attached.
Reviewers: kailiu, sdong, haobo, leveldb
Reviewed By: haobo
CC: xjin, haobo
Differential Revision: https://reviews.facebook.net/D12675
Summary: This might help with p99 performance, but does not solve the real problem. More discussion on #2947135
Test Plan: make check
Reviewers: dhruba, haobo
Reviewed By: dhruba
CC: leveldb
Differential Revision: https://reviews.facebook.net/D13809
Summary:
Create a new type of file on startup if it doesn't already exist called DBID.
This will store a unique number generated from boost library's uuid header file.
The use-case is to identify the case of a db losing all its data and coming back up either empty or from an image(backup/live replica's recovery)
the key point to note is that DBID is not stored in a backup or db snapshot
It's preferable to use Boost for uuid because:
1) A non-standard way of generating uuid is not good
2) /proc/sys/kernel/random/uuid generates a uuid but only on linux environments and the solution would not be clean
3) c++ doesn't have any direct way to get a uuid
4) Boost is a very good library that was already having linkage in rocksdb from third-party
Note: I had to update the TOOLCHAIN_REV in build files to get latest verison of boost from third-party as the older version had a bug.
I had to put Wno-uninitialized in Makefile because boost-1.51 has an unitialized variable and rocksdb would not comiple otherwise. Latet open-source for boost is 1.54 but is not there in third-party. I have notified the concerned people in fbcode about it.
@kailiu : While releasing to third-party, an additional dependency will need to be created for boost in TARGETS file. I can help identify.
Test Plan:
Expand db_test to test 2 cases
1) Restarting db with Id file present - verify that no change to Id
2)Restarting db with Id file deleted - verify that a different Id is there after reopen
Also run make all check
Reviewers: dhruba, haobo, kailiu, sdong
Reviewed By: dhruba
CC: leveldb
Differential Revision: https://reviews.facebook.net/D13587
Summary: I have implemented basic simple use case that I need for External Value Store I'm working on. There is a potential for making this prettier by refactoring/combining WritableFile and RandomAccessFile, avoiding some copypasta. However, I decided to implement just the basic functionality, so I can continue working on the other diff.
Test Plan: Added a unittest
Reviewers: dhruba, haobo, kailiu
Reviewed By: haobo
CC: leveldb
Differential Revision: https://reviews.facebook.net/D13365
Summary:
Change namespace from leveldb to rocksdb. This allows a single
application to link in open-source leveldb code as well as
rocksdb code into the same process.
Test Plan: compile rocksdb
Reviewers: emayanke
Reviewed By: emayanke
CC: leveldb
Differential Revision: https://reviews.facebook.net/D13287
Summary:
Added a new api to the Environment that allows clearing out not-needed
pages from the OS cache. This will be helpful when the compressed
block cache replaces the OS cache.
Test Plan: EnvPosixTest.InvalidateCache
Reviewers: haobo
Reviewed By: haobo
CC: leveldb
Differential Revision: https://reviews.facebook.net/D13041
Summary:
this is the ground work for separating memtable flush jobs to their own thread pool.
Both SetBackgroundThreads and Schedule take a third parameter Priority to indicate which thread pool they are working on. The names LOW and HIGH are just identifiers for two different thread pools, and does not indicate real difference in 'priority'. We can set number of threads in the pools independently.
The thread pool implementation is refactored.
Test Plan: make check
Reviewers: dhruba, emayanke
CC: leveldb
Differential Revision: https://reviews.facebook.net/D12885
Summary: The pupose of this diff is to expose per user-call level precise timing of block read, so that we can answer questions like: a Get() costs me 100ms, is that somehow related to loading blocks from file system, or sth else? We will answer that with EXACTLY how many blocks have been read, how much time was spent on transfering the bytes from os, how much time was spent on checksum verification and how much time was spent on block decompression, just for that one Get. A nano second stopwatch was introduced to track time with higher precision. The cost/precision of the stopwatch is also measured in unit-test. On my dev box, retrieving one time instance costs about 30ns, on average. The deviation of timing results is good enough to track 100ns-1us level events. And the overhead could be safely ignored for 100us level events (10000 instances/s), for example, a viewstate thrift call.
Test Plan: perf_context_test, also testing with viewstate shadow traffic.
Reviewers: dhruba
Reviewed By: dhruba
CC: leveldb, xjin
Differential Revision: https://reviews.facebook.net/D12351
Summary: Replace include/leveldb with include/rocksdb.
Test Plan:
make clean; make check
make clean; make release
Differential Revision: https://reviews.facebook.net/D12489
Summary: as title, also removed an incorrect assertion
Test Plan: make check; db_stress --mmap_read=1; db_stress --mmap_read=0
Reviewers: dhruba, emayanke
CC: leveldb
Differential Revision: https://reviews.facebook.net/D11367
Summary: This diff added an option to control the incremenal sync frequency. db_bench has a new flag bytes_per_sync for easy tuning exercise.
Test Plan: make check; db_bench
Reviewers: dhruba
CC: leveldb
Differential Revision: https://reviews.facebook.net/D11295
Summary:
During compaction, we sync the output files after they are fully written out. This causes unnecessary blocking of the compaction thread and burstiness of the write traffic.
This diff simply asks the OS to sync data incrementally as they are written, on the background. The hope is that, at the final sync, most of the data are already on disk and we would block less on the sync call. Thus, each compaction runs faster and we could use fewer number of compaction threads to saturate IO.
In addition, the write traffic will be smoothed out, hopefully reducing the IO P99 latency too.
Some quick tests show 10~20% improvement in per thread compaction throughput. Combined with posix advice on compaction read, just 5 threads are enough to almost saturate the udb flash bandwidth for 800 bytes write only benchmark.
What's more promising is that, with saturated IO, iostat shows average wait time is actually smoother and much smaller.
For the write only test 800bytes test:
Before the change: await occillate between 10ms and 3ms
After the change: await ranges 1-3ms
Will test against read-modify-write workload too, see if high read latency P99 could be resolved.
Will introduce a parameter to control the sync interval in a follow up diff after cleaning up EnvOptions.
Test Plan: make check; db_bench; db_stress
Reviewers: dhruba
CC: leveldb
Differential Revision: https://reviews.facebook.net/D11115
Summary:
This diff simplifies EnvOptions by treating it as POD, similar to Options.
- virtual functions are removed and member fields are accessed directly.
- StorageOptions is removed.
- Options.allow_readahead and Options.allow_readahead_compactions are deprecated.
- Unused global variables are removed: useOsBuffer, useFsReadAhead, useMmapRead, useMmapWrite
Test Plan: make check; db_stress
Reviewers: dhruba
CC: leveldb
Differential Revision: https://reviews.facebook.net/D11175
Summary:
Current posix advice implementation ties up the access pattern hint with the creation of a file.
It is not possible to apply different advice for different access (random get vs compaction read),
without keeping two open files for the same table. This patch extended the RandomeAccessFile interface
to accept new access hint at anytime. Particularly, we are able to set different access hint on the same
table file based on when/how the file is used.
Two options are added to set the access hint, after the file is first opened and after the file is being
compacted.
Test Plan: make check; db_stress; db_bench
Reviewers: dhruba
Reviewed By: dhruba
CC: MarkCallaghan, leveldb
Differential Revision: https://reviews.facebook.net/D10905
Summary:
This is initial version. A few ways in which this could
be extended in the future are:
(a) Killing from more places in source code
(b) Hashing stack and using that hash in determining whether to crash.
This is to avoid crashing more often at source lines that are executed
more often.
(c) Raising exceptions or returning errors instead of killing
Test Plan:
This whole thing is for testing.
Here is part of output:
python2.7 tools/db_crashtest2.py -d 600
Running db_stress
db_stress retncode -15 output LevelDB version : 1.5
Number of threads : 32
Ops per thread : 10000000
Read percentage : 50
Write-buffer-size : 4194304
Delete percentage : 30
Max key : 1000
Ratio #ops/#keys : 320000
Num times DB reopens: 0
Batches/snapshots : 1
Purge redundant % : 50
Num keys per lock : 4
Compression : snappy
------------------------------------------------
No lock creation because test_batches_snapshots set
2013/04/26-17:55:17 Starting database operations
Created bg thread 0x7fc1f07ff700
... finished 60000 ops
Running db_stress
db_stress retncode -15 output LevelDB version : 1.5
Number of threads : 32
Ops per thread : 10000000
Read percentage : 50
Write-buffer-size : 4194304
Delete percentage : 30
Max key : 1000
Ratio #ops/#keys : 320000
Num times DB reopens: 0
Batches/snapshots : 1
Purge redundant % : 50
Num keys per lock : 4
Compression : snappy
------------------------------------------------
Created bg thread 0x7ff0137ff700
No lock creation because test_batches_snapshots set
2013/04/26-17:56:15 Starting database operations
... finished 90000 ops
Revert Plan: OK
Task ID: #2252691
Reviewers: dhruba, emayanke
Reviewed By: emayanke
CC: leveldb, haobo
Differential Revision: https://reviews.facebook.net/D10581
Summary:
Mark's task description from #2316777
Env::Default() comes from util/env_posix.cc
This is a static global.
static PosixEnv default_env;
Env* Env::Default() {
return &default_env;
}
-----
These globals assume default_env was initialized first. I don't think that is safe or correct to do (http://stackoverflow.com/questions/1005685/c-static-initialization-order)
const string AutoRollLoggerTest::kTestDir(
test::TmpDir() + "/db_log_test");
const string AutoRollLoggerTest::kLogFile(
test::TmpDir() + "/db_log_test/LOG");
Env* AutoRollLoggerTest::env = Env::Default();
Test Plan:
run make clean && make && make check
But how can I know if it works in Ubuntu?
Reviewers: MarkCallaghan, chip
Reviewed By: chip
CC: leveldb, dhruba, haobo
Differential Revision: https://reviews.facebook.net/D10491
Summary:
RocksDB doesn't build on Ubuntu VM .. shoudl be fixed with this patch.
g++ --version
g++ (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3
util/env_posix.cc:68:24: sorry, unimplemented: non-static data member initializers
util/env_posix.cc:68:24: error: ISO C++ forbids in-class initialization of non-const static member ‘use_os_buffer’
util/env_posix.cc:113:24: sorry, unimplemented: non-static data member initializers
util/env_posix.cc:113:24: error: ISO C++ forbids in-class initialization of non-const static member ‘use_os_buffer
Test Plan: make check
Reviewers: sheki, leveldb
Reviewed By: sheki
Differential Revision: https://reviews.facebook.net/D10461
Summary:
The background compaction threads are never exitted and therefore caused
memory-leaks while running rpcksdb tests. Have changed the PosixEnv destructor to exit and join them and changed the tests likewise
The memory leaked has reduced from 320 bytes to 64 bytes in all the tests. The 64
bytes is relating to
pthread_exit, but still have to figure out why. The stack-trace right now with
table_test.cc = 64 bytes in 1 blocks are possibly lost in loss record 4 of 5
at 0x475D8C: malloc (jemalloc.c:914)
by 0x400D69E: _dl_map_object_deps (dl-deps.c:505)
by 0x4013393: dl_open_worker (dl-open.c:263)
by 0x400F015: _dl_catch_error (dl-error.c:178)
by 0x4013B2B: _dl_open (dl-open.c:569)
by 0x5D3E913: do_dlopen (dl-libc.c:86)
by 0x400F015: _dl_catch_error (dl-error.c:178)
by 0x5D3E9D6: __libc_dlopen_mode (dl-libc.c:47)
by 0x5048BF3: pthread_cancel_init (unwind-forcedunwind.c:53)
by 0x5048DC9: _Unwind_ForcedUnwind (unwind-forcedunwind.c:126)
by 0x5046D9F: __pthread_unwind (unwind.c:130)
by 0x50413A4: pthread_exit (pthreadP.h:289)
Test Plan: make all check
Reviewers: dhruba, sheki, haobo
Reviewed By: dhruba
CC: leveldb, chip
Differential Revision: https://reviews.facebook.net/D9573
Summary: as subject. This is causing problem in adsconv. Ideally, this flags should be set in open. But that is only supported in Linux kernel ≥2.6.23 and glibc ≥2.7.
Test Plan:
db_test
run db_test
Reviewers: dhruba, MarkCallaghan, haobo
Reviewed By: dhruba
CC: leveldb, chip
Differential Revision: https://reviews.facebook.net/D10089
Summary:
1. SetBackgroundThreads was not thread safe
2. queue_size_ does not seem necessary
3. moved condition signal after shared state change. Even though the original
order is in practice ok (because the mutex is still held), it looks fishy
and non-intuitive.
Test Plan: make check
Reviewers: dhruba
Reviewed By: dhruba
CC: leveldb, zshao
Differential Revision: https://reviews.facebook.net/D9825
Summary: This caused compilation problems on some gcc platforms during the third-partyrelease
Test Plan: make
Reviewers: sheki
Reviewed By: sheki
Differential Revision: https://reviews.facebook.net/D9627
Summary:
This patch allows an application to specify whether to use bufferedio,
reads-via-mmaps and writes-via-mmaps per database. Earlier, there
was a global static variable that was used to configure this functionality.
The default setting remains the same (and is backward compatible):
1. use bufferedio
2. do not use mmaps for reads
3. use mmap for writes
4. use readaheads for reads needed for compaction
I also added a parameter to db_bench to be able to explicitly specify
whether to do readaheads for compactions or not.
Test Plan: make check
Reviewers: sheki, heyongqiang, MarkCallaghan
Reviewed By: sheki
CC: leveldb
Differential Revision: https://reviews.facebook.net/D9429
Summary:
Ftruncate does not throw an error on disk-full. This causes Sig-bus in
the case where the database tries to issue a Put call on a full-disk.
Use posix_fallocate for allocation instead of truncate.
Add a check to use MMaped files only on ext4, xfs and tempfs, as
posix_fallocate is very slow on ext3 and older.
Test Plan: make all check
Reviewers: dhruba, chip
Reviewed By: dhruba
CC: adsharma, leveldb
Differential Revision: https://reviews.facebook.net/D9291
Summary:
Added function to `RandomAccessFile` to generate an unique ID for that file. Currently only `PosixRandomAccessFile` has this behaviour implemented and only on Linux.
Changed how key is generated in `Table::BlockReader`.
Added tests to check whether the unique ID is stable, unique and not a prefix of another unique ID. Added tests to see that `Table` uses the cache more efficiently.
Test Plan: make check
Reviewers: chip, vamsi, dhruba
Reviewed By: chip
CC: leveldb
Differential Revision: https://reviews.facebook.net/D8145
Summary: fallocate is linux only, so let's protect it with ifdef's
Test Plan: make
Reviewers: sheki, dhruba
Reviewed By: dhruba
CC: leveldb
Differential Revision: https://reviews.facebook.net/D8223
Summary:
On some filesystems, pre-allocation can be a considerable
amount of space. xfs in our production environment pre-allocates by
1GB, for instance. By using fallocate to inform the kernel of our
expected file sizes, we eliminate this wasteage (that isn't recovered
until the file is closed which, in the case of LOG files, can be a
considerable amount of time).
Test Plan:
created an xfs loopback filesystem, mounted with
allocsize=4M, and ran db_stress. LOG file without this change was 4M,
and with it it was 128k then grew to normal size.
Reviewers: dhruba
Reviewed By: dhruba
CC: adsharma, leveldb
Differential Revision: https://reviews.facebook.net/D7953
Summary:
Replace manual memory management with std::unique_ptr in a
number of places; not exhaustive, but this fixes a few leaks with file
handles as well as clarifies semantics of the ownership of file handles
with log classes.
Test Plan: db_stress, make check
Reviewers: dhruba
Reviewed By: dhruba
CC: zshao, leveldb, heyongqiang
Differential Revision: https://reviews.facebook.net/D8043
Summary:
clang is an alternate compiler based on llvm. It produces
nicer error messages and finds some bugs that gcc doesn't, such as the
size_t change in this file (which caused some write return values to be
misinterpreted!)
Clang isn't the default; to try it, do "USE_CLANG=1 make" or "export
USE_CLANG=1" then make as normal
Test Plan: "make check" and "USE_CLANG=1 make check"
Reviewers: dhruba
Reviewed By: dhruba
Differential Revision: https://reviews.facebook.net/D7899
Igor Canadi