Summary:
**Summary**:
2 new statistics counters are added to RocksDB: `MEMTABLE_PAYLOAD_BYTES_AT_FLUSH` and `MEMTABLE_GARBAGE_BYTES_AT_FLUSH`. The former tracks how many raw bytes of useful data are present on the memtable at flush time, whereas the latter is tracks how many of these raw bytes are considered garbage, meaning that they ended up not being imported on the SSTables resulting from the flush operations.
**Unit test**: run `make db_flush_test -j$(nproc); ./db_flush_test` to run the unit test.
This executable includes 3 tests, that test support and correct stat calculations for workloads with inserts, deletes, and DeleteRanges. The parameters are set such that the workloads are performed on a single memtable, and a single SSTable is created as a result of the flush operation. The flush operation is manually called in the test file. The tests verify that the values of these 2 statistics counters introduced in this PR can be exactly predicted, showing that we have a full understanding of the underlying operations.
**Performance testing**:
`./db_bench -statistics -benchmarks=fillrandom -num=10000000` repeated 10 times.
Timing done using "date" function in a bash script.
_Results_:
Original Rocksdb fork: mean 66.6 sec, std 1.18 sec.
This feature branch: mean 67.4 sec, std 1.35 sec.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/8411
Reviewed By: akankshamahajan15
Differential Revision: D29150629
Pulled By: bjlemaire
fbshipit-source-id: 7b3c2e86d50c6aa34fa50fd134282eacb543a5b1
Summary:
The ImmutableCFOptions contained a bunch of fields that belonged to the ImmutableDBOptions. This change cleans that up by introducing an ImmutableOptions struct. Following the pattern of Options struct, this class inherits from the DB and CFOption structs (of the Immutable form).
Only one structural change (the ImmutableCFOptions::fs was changed to a shared_ptr from a raw one) is in this PR. All of the other changes involve moving the member variables from the ImmutableCFOptions into the ImmutableOptions and changing member variables or function parameters as required for compilation purposes.
Follow-on PRs may do a further clean-up of the code, such as renaming variables (such as "ImmutableOptions cf_options") and potentially eliminating un-needed function parameters (there is no longer a need to pass both an ImmutableDBOptions and an ImmutableOptions to a function).
Pull Request resolved: https://github.com/facebook/rocksdb/pull/8262
Reviewed By: pdillinger
Differential Revision: D28226540
Pulled By: mrambacher
fbshipit-source-id: 18ae71eadc879dedbe38b1eb8e6f9ff5c7147dbf
Summary:
In a distributed environment, a file `rename()` operation can succeed on server (remote)
side, but the client can somehow return non-ok status to RocksDB. Possible reasons include
network partition, connection issue, etc. This happens in `rocksdb::SetCurrentFile()`, which
can be called in `LogAndApply() -> ProcessManifestWrites()` if RocksDB tries to switch to a
new MANIFEST. We currently always delete the new MANIFEST if an error occurs.
This is problematic in distributed world. If the server-side successfully updates the CURRENT
file via renaming, then a subsequent `DB::Open()` will try to look for the new MANIFEST and fail.
As a fix, we can track the execution result of IO operations on the new MANIFEST.
- If IO operations on the new MANIFEST fail, then we know the CURRENT must point to the original
MANIFEST. Therefore, it is safe to remove the new MANIFEST.
- If IO operations on the new MANIFEST all succeed, but somehow we end up in the clean up
code block, then we do not know whether CURRENT points to the new or old MANIFEST. (For local
POSIX-compliant FS, it should still point to old MANIFEST, but it does not matter if we keep the
new MANIFEST.) Therefore, we keep the new MANIFEST.
- Any future `LogAndApply()` will switch to a new MANIFEST and update CURRENT.
- If process reopens the db immediately after the failure, then the CURRENT file can point
to either the new MANIFEST or the old one, both of which exist. Therefore, recovery can
succeed and ignore the other.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/8192
Test Plan: make check
Reviewed By: zhichao-cao
Differential Revision: D27804648
Pulled By: riversand963
fbshipit-source-id: 9c16f2a5ce41bc6aadf085e48449b19ede8423e4
Summary:
Added `TableProperties::{fast,slow}_compression_estimated_data_size`.
These properties are present in block-based tables when
`ColumnFamilyOptions::sample_for_compression > 0` and the necessary
compression library is supported when the file is generated. They
contain estimates of what `TableProperties::data_size` would be if the
"fast"/"slow" compression library had been used instead. One
limitation is we do not record exactly which "fast" (ZSTD or Zlib)
or "slow" (LZ4 or Snappy) compression library produced the result.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/8139
Test Plan:
- new unit test
- ran `db_bench` with `sample_for_compression=1`; verified the `data_size` property matches the `{slow,fast}_compression_estimated_data_size` when the same compression type is used for the output file compression and the sampled compression
Reviewed By: riversand963
Differential Revision: D27454338
Pulled By: ajkr
fbshipit-source-id: 9529293de93ddac7f03b2e149d746e9f634abac4
Summary:
Which should return 2 long instead of an array.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/8098
Reviewed By: mrambacher
Differential Revision: D27308741
Pulled By: jay-zhuang
fbshipit-source-id: 44beea2bd28cf6779b048bebc98f2426fe95e25c
Summary:
This is a small fix to what I think is a mistype in two comments in `DBOptionsInterface.java`. If it was not an error, feel free to close.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/8086
Reviewed By: ajkr
Differential Revision: D27260488
Pulled By: mrambacher
fbshipit-source-id: 469daadaf6039d5b5187132b8e0c7c3672842f21
Summary:
Add statistics and info log for error handler: counters for bg error, bg io error, bg retryable io error, auto resume, auto resume total retry, and auto resume sucess; Histogram for auto resume retry count in each recovery call.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/8050
Test Plan: make check and add test to error_handler_fs_test
Reviewed By: anand1976
Differential Revision: D26990565
Pulled By: zhichao-cao
fbshipit-source-id: 49f71e8ea4e9db8b189943976404205b56ab883f
Summary:
support getUsage and getPinnedUsage in JavaAPI for Cache
also fix a typo in LRUCacheTest.java that the highPriPoolRatio is not valid(set 5, I guess it means 0.05)
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7925
Reviewed By: mrambacher
Differential Revision: D26900241
Pulled By: ajkr
fbshipit-source-id: 735d1e40a16fa8919c89c7c7154ba7f81208ec33
Summary:
Fixes 3 minor Javadoc copy-paste errors in the `RocksDB#newIterator()` and `Transaction#getIterator()` variants that take a column family handle but are talking about iterating over "the database" or "the default column family".
Pull Request resolved: https://github.com/facebook/rocksdb/pull/8034
Reviewed By: jay-zhuang
Differential Revision: D26877667
Pulled By: mrambacher
fbshipit-source-id: 95dd95b667c496e389f221acc9a91b340e4b63bf
Summary:
The variable `byteCompressionType` is only assigned values of primitive type and is never 'null', but it is declared with the boxed type 'Byte'.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7981
Reviewed By: ajkr
Differential Revision: D26546600
Pulled By: jay-zhuang
fbshipit-source-id: 07b579cdfcfc2262a448ca3626e216416fd05892
Summary:
Haven't seen any production issues with new Bloom filter and
it's now > 1 year old (added in 6.6.0).
Updated check_format_compatible.sh and HISTORY.md
Pull Request resolved: https://github.com/facebook/rocksdb/pull/8017
Test Plan: tests updated (or prior bugs fixed)
Reviewed By: ajkr
Differential Revision: D26762197
Pulled By: pdillinger
fbshipit-source-id: 0e755c46b443087c1544da0fd545beb9c403d1c2
Summary:
For dictionary compression, we need to collect some representative samples of the data to be compressed, which we use to either generate or train (when `CompressionOptions::zstd_max_train_bytes > 0`) a dictionary. Previously, the strategy was to buffer all the data blocks during flush, and up to the target file size during compaction. That strategy allowed us to randomly pick samples from as wide a range as possible that'd be guaranteed to land in a single output file.
However, some users try to make huge files in memory-constrained environments, where this strategy can cause OOM. This PR introduces an option, `CompressionOptions::max_dict_buffer_bytes`, that limits how much data blocks are buffered before we switch to unbuffered mode (which means creating the per-SST dictionary, writing out the buffered data, and compressing/writing new blocks as soon as they are built). It is not strict as we currently buffer more than just data blocks -- also keys are buffered. But it does make a step towards giving users predictable memory usage.
Related changes include:
- Changed sampling for dictionary compression to select unique data blocks when there is limited availability of data blocks
- Made use of `BlockBuilder::SwapAndReset()` to save an allocation+memcpy when buffering data blocks for building a dictionary
- Changed `ParseBoolean()` to accept an input containing characters after the boolean. This is necessary since, with this PR, a value for `CompressionOptions::enabled` is no longer necessarily the final component in the `CompressionOptions` string.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7970
Test Plan:
- updated `CompressionOptions` unit tests to verify limit is respected (to the extent expected in the current implementation) in various scenarios of flush/compaction to bottommost/non-bottommost level
- looked at jemalloc heap profiles right before and after switching to unbuffered mode during flush/compaction. Verified memory usage in buffering is proportional to the limit set.
Reviewed By: pdillinger
Differential Revision: D26467994
Pulled By: ajkr
fbshipit-source-id: 3da4ef9fba59974e4ef40e40c01611002c861465
Summary:
this is a trivial PR for rocksdb java samples, I think it is a typo about write options. to do sync write, WAL should not be disabled
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7894
Reviewed By: jay-zhuang
Differential Revision: D26047128
Pulled By: mrambacher
fbshipit-source-id: a06ce54cb61af0d3f2578a709c34a0b1ccecb0b2
Summary:
This request is adding support for using DirectSlice in ReadOptions lower/upper bounds.
To be more efficient I have added setLength to DirectSlice so I can just update the length to be used by slice from direct buffer. It is also needed, because when one creates iterator it keep pointer to original slice so setting new slice in options does not help (it needs to reuse existing one). Using this approach one can modify the slice any time during operations with iterator.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7132
Reviewed By: zhichao-cao
Differential Revision: D25840092
Pulled By: jay-zhuang
fbshipit-source-id: 760167baf61568c9a35138145c4bf9b06824cb71
Summary:
Fix ColumnFamilyOptionsTest.cfPaths and OptionsTest.cfPaths in 6.15 branch (and probably other branches including master)
has_exception variable was not initialized which was causing test failures and incorrect behavior on s390 platform (and maybe others as variable content is undefined).
adamretter please take a look.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7853
Reviewed By: akankshamahajan15
Differential Revision: D25901639
Pulled By: jay-zhuang
fbshipit-source-id: 151b5db27b495fc6d8ed54c0eccbde2508215ac5
Summary:
Classes ColumnFamilyHandle and CapturingWriteBatchHandler.Event have
byte array fields as part of their identity, but they do not use the
arrays' content to compute the instance's hash, and instead rely on the
arrays' identity, causing instances to have different hashcodes
although they are equal.
The PR addresses it by using the arrays' content to compute the hash,
like the equals method does.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7860
Reviewed By: jay-zhuang
Differential Revision: D25901327
Pulled By: akankshamahajan15
fbshipit-source-id: 347e7b3d2ba7befe7faa956b033e6421b9d0c235
Summary:
* Fixes a Java test compilation issue on macOS
* Cleans up CircleCI RocksDBJava build config
* Adds CircleCI for RocksDBJava on MacOS
* Ensures backwards compatibility with older macOS via CircleCI
* Fixes RocksJava static builds ordering
* Adds missing RocksJava static builds to CircleCI for Mac and Linux
* Improves parallelism in RocksJava builds
* Reduces the size of the machines used for RocksJava CircleCI as they don't need to be so large (Saves credits)
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7769
Reviewed By: akankshamahajan15
Differential Revision: D25601293
Pulled By: pdillinger
fbshipit-source-id: 0a0bb9906f65438fe143487d78e37e1947364d08
Summary:
Consider the following sequence of events:
1. Db flushed an SST with file number N, appended to MANIFEST, and tried to sync the MANIFEST.
2. Syncing MANIFEST failed and db crashed.
3. Db tried to recover with this MANIFEST. In the meantime, no entry about the newly-flushed SST was found in the MANIFEST. Therefore, RocksDB replayed WAL and tried to flush to an SST file reusing the same file number N. This failed because file system does not support overwrite. Then Db deleted this file.
4. Db crashed again.
5. Db tried to recover. When db read the MANIFEST, there was an entry referencing N.sst. This could happen probably because the append in step 1 finally reached the MANIFEST and became visible. Since N.sst had been deleted in step 3, recovery failed.
It is possible that N.sst created in step 1 is valid. Although step 3 would still fail since the MANIFEST was not synced properly in step 1 and 2, deleting N.sst would make it impossible for the db to recover even if the remaining part of MANIFEST was appended and visible after step 5.
After this PR, in step 3, immediately after recovering from MANIFEST, a new MANIFEST is created, then we find that N.sst is not referenced in the MANIFEST, so we delete it, and we'll not reuse N as file number. Then in step 5, since the new MANIFEST does not contain N.sst, the recovery failure situation in step 5 won't happen.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7621
Test Plan:
1. some tests are updated, because these tests assume that new MANIFEST is created after WAL recovery.
2. a new unit test is added in db_basic_test to simulate step 3.
Reviewed By: riversand963
Differential Revision: D24668144
Pulled By: cheng-chang
fbshipit-source-id: 90d7487fbad2bc3714f5ede46ea949895b15ae3b
Summary:
The original test nests a lot of `try` blocks. This PR flattens these blocks into independent blocks, so that each `try` block closes the DB before opening the next DB instance.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7608
Test Plan: watch the existing java tests to pass
Reviewed By: zhichao-cao
Differential Revision: D24611621
Pulled By: cheng-chang
fbshipit-source-id: d486c5d37ac25d4b860d739ef2cdd58e6064d42d
Summary:
Fixes Issue https://github.com/facebook/rocksdb/issues/7497
When allow_data_in_errors db_options is set, log error key details in `ParseInternalKey()`
Have fixed most of the calls. Have few TODOs still pending - because have to make more deeper changes to pass in the allow_data_in_errors flag. Will do those in a separate PR later.
Tests:
- make check
- some of the existing tests that exercise the "internal key too small" condition are: dbformat_test, cuckoo_table_builder_test
- some of the existing tests that exercise the corrupted key path are: corruption_test, merge_helper_test, compaction_iterator_test
Example of new status returns:
- Key too small - `Corrupted Key: Internal Key too small. Size=5`
- Corrupt key with allow_data_in_errors option set to false: `Corrupted Key: '<redacted>' seq:3, type:3`
- Corrupt key with allow_data_in_errors option set to true: `Corrupted Key: '61' seq:3, type:3`
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7515
Reviewed By: ajkr
Differential Revision: D24240264
Pulled By: ramvadiv
fbshipit-source-id: bc48f5d4475ac19d7713e16df37505b31aac42e7
Summary:
- Takes the burden off developer to close ColumnFamilyHandle instances before closing RocksDB instance
- The change is backward-compatible
----
Previously the pattern for working with Column Families was:
```java
try (final ColumnFamilyOptions cfOpts = new ColumnFamilyOptions().optimizeUniversalStyleCompaction()) {
// list of column family descriptors, first entry must always be default column family
final List<ColumnFamilyDescriptor> cfDescriptors = Arrays.asList(
new ColumnFamilyDescriptor(RocksDB.DEFAULT_COLUMN_FAMILY, cfOpts),
new ColumnFamilyDescriptor("my-first-columnfamily".getBytes(), cfOpts)
);
// a list which will hold the handles for the column families once the db is opened
final List<ColumnFamilyHandle> columnFamilyHandleList =
new ArrayList<>();
try (final DBOptions options = new DBOptions()
.setCreateIfMissing(true)
.setCreateMissingColumnFamilies(true);
final RocksDB db = RocksDB.open(options,
"path/to/do", cfDescriptors,
columnFamilyHandleList)) {
try {
// do something
} finally {
// NOTE user must explicitly frees the column family handles before freeing the db
for (final ColumnFamilyHandle columnFamilyHandle :
columnFamilyHandleList) {
columnFamilyHandle.close();
}
} // frees the column family options
}
} // frees the db and the db options
```
With the changes in this PR, the Java user no longer has to worry about manually closing the Column Families, which allows them to write simpler symmetrical create/free oriented code like this:
```java
try (final ColumnFamilyOptions cfOpts = new ColumnFamilyOptions().optimizeUniversalStyleCompaction()) {
// list of column family descriptors, first entry must always be default column family
final List<ColumnFamilyDescriptor> cfDescriptors = Arrays.asList(
new ColumnFamilyDescriptor(RocksDB.DEFAULT_COLUMN_FAMILY, cfOpts),
new ColumnFamilyDescriptor("my-first-columnfamily".getBytes(), cfOpts)
);
// a list which will hold the handles for the column families once the db is opened
final List<ColumnFamilyHandle> columnFamilyHandleList =
new ArrayList<>();
try (final DBOptions options = new DBOptions()
.setCreateIfMissing(true)
.setCreateMissingColumnFamilies(true);
final RocksDB db = RocksDB.open(options,
"path/to/do", cfDescriptors,
columnFamilyHandleList)) {
// do something
} // frees the column family options, then frees the db and the db options
}
}
```
**NOTE**: The changes in this PR are backwards API compatible, which means existing code using the original approach will also continue to function correctly.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7428
Reviewed By: cheng-chang
Differential Revision: D24063348
Pulled By: jay-zhuang
fbshipit-source-id: 648d7526669923128c863ead94516bf4d50ac658
Summary:
Allows adding event listeners in RocksJava.
* Adds listeners getter and setter in `Options` and `DBOptions` classes.
* Adds `EventListener` Java interface and base class for implementing custom event listener callbacks - `AbstractEventListener`, which has an underlying native callback class implementing C++ `EventListener` class.
* `AbstractEventListener` class has mechanism for selectively enabling its callback methods in order to prevent invoking Java method if it is not implemented. This decreases performance cost in case only subset of event listener callback methods is needed - the JNI code for remaining "no-op" callbacks is not executed.
* The code is covered by unit tests in `EventListenerTest.java`, there are also tests added for setting/getting listeners field in `OptionsTest.java` and `DBOptionsTest.java`.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7425
Reviewed By: pdillinger
Differential Revision: D24063390
Pulled By: jay-zhuang
fbshipit-source-id: 508c359538983d6b765e70d9989c351794a944ee
Summary:
Add following stats for MultiGet in Histogram to get more insight on MultiGet.
1. Number of index and filter blocks read from file as part of MultiGet
request per level.
2. Number of data blocks read from file per level.
3. Number of SST files loaded from file system per level.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7366
Reviewed By: anand1976
Differential Revision: D24127040
Pulled By: akankshamahajan15
fbshipit-source-id: e63a003056b833729b277edc0639c08fb432756b
Summary:
as title
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7347
Test Plan: unit tests included
Reviewed By: jay-zhuang
Differential Revision: D23592552
Pulled By: pdillinger
fbshipit-source-id: 1c3571b6f42bfd0cfd723ff49d01fbc02a1be45b
Summary:
Previously RocksJava limited the format_version to 4. However, the C++ API is now at 5, and this will likely increase again in future. The Java API now allows any positive integer, and an exception is raised from JNI if the format_version is out-of-bounds.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7242
Reviewed By: cheng-chang
Differential Revision: D23077941
Pulled By: pdillinger
fbshipit-source-id: ee69f7203448acddc41c6d86b470ed987d3d366d
Summary:
The PR fixes a Java test for Merge operator `uint64add`.
The current implementation uses wrong byte order for long serialization, but fails to catch this error because the merge sum is lower than `256`.
The PR makes this test case more representative (i.e. it fails with wrong byte order) and changes the byte order to little endian.
Some background: RocksDB uses LittleEndian byte order for integer serialization across all platforms. `MergeTest` uses `ByteBuffer` that defaults to BigEndian byte order.
This test case might probably be used as a sample of `MergeOperator` usage in Java.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7243
Reviewed By: ajkr
Differential Revision: D23079593
Pulled By: pdillinger
fbshipit-source-id: 82e8e166901d66733e96a0116f88d0ec4761ddf1
Summary:
Improvements to the RocksJava release process:
* Generates the Maven artifact version number as part of the release step
* Also generates appropriate checksum files to speed the deploy and publish step
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7219
Reviewed By: ltamasi
Differential Revision: D22983481
Pulled By: pdillinger
fbshipit-source-id: 7b8ffaf46471cd3cda181eb830c962b317d2e688
Summary:
Adds compaction statistics (total bytes read and written) for compactions that occur for delete-triggered, periodic, and TTL compaction reasons.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/7165
Test Plan:
TTL and periodic can be checked by runnning db_bench with the options activated:
/db_bench --benchmarks="fillrandom,stats" --statistics --num=10000000 -base_background_compactions=16 -periodic_compaction_seconds=1
./db_bench --benchmarks="fillrandom,stats" --statistics --num=10000000 -base_background_compactions=16 -fifo_compaction_ttl=1
Setting the time to one second causes non-zero bytes read/written for those compaction reasons. Disabling them or setting them to times longer than the test run length causes the stats to return to zero as expected.
Delete-triggered compaction counting is tested in DBTablePropertiesTest.DeletionTriggeredCompactionMarking
Reviewed By: ajkr
Differential Revision: D22693050
Pulled By: akabcenell
fbshipit-source-id: d15cef4d94576f703015c8942d5f0d492f69401d
Summary:
SST Partitioner interface that allows to split SST files during compactions.
It basically instruct compaction to create a new file when needed. When one is using well defined prefixes and prefixed way of defining tables it is good to define also partitioning so that promotion of some SST file does not cover huge key space on next level (worst case complete space).
Pull Request resolved: https://github.com/facebook/rocksdb/pull/6957
Reviewed By: ajkr
Differential Revision: D22461239
fbshipit-source-id: 9ce07bba08b3ba89c2d45630520368f704d1316e
longlijian
