`numFeatures` is only used by multinomial logistic regression. Calling `.first()` for every GLM causes performance regression, especially in Python.
Author: Xiangrui Meng <meng@databricks.com>
Closes#4647 from mengxr/SPARK-5858 and squashes the following commits:
036dc7f [Xiangrui Meng] remove unnecessary first() call
12c5548 [Xiangrui Meng] check numFeatures only once
I've seen out of memory exceptions when trying
to run many parallel builds against the same Zinc
server during packaging. We should use the same
increased memory settings we use for Maven itself.
I tested this and confirmed that the Nailgun JVM
launched with higher memory.
Author: Patrick Wendell <patrick@databricks.com>
Closes#4643 from pwendell/zinc-memory and squashes the following commits:
717cfb0 [Patrick Wendell] SPARK-5856: Launch Zinc with larger memory options.
Author: jerryshao <saisai.shao@intel.com>
Closes#4612 from jerryshao/SPARK-5826 and squashes the following commits:
7ec71db [jerryshao] Remove transient for conf statement
88d84e6 [jerryshao] Fix Configuration not serializable problem
If we need to transform the input data, we should cache the output to avoid re-computing feature vectors every iteration. dbtsai
Author: Xiangrui Meng <meng@databricks.com>
Closes#4593 from mengxr/SPARK-5802 and squashes the following commits:
ae3be84 [Xiangrui Meng] cache transformed data in glm
Author: Reynold Xin <rxin@databricks.com>
Closes#4640 from rxin/SPARK-5853 and squashes the following commits:
9c6f569 [Reynold Xin] [SPARK-5853][SQL] Schema support in Row.
Author: Patrick Wendell <patrick@databricks.com>
Closes#4638 from pwendell/SPARK-5850 and squashes the following commits:
386126f [Patrick Wendell] SPARK-5850: Remove experimental label for Scala 2.11 and FlumePollingStream.
There is chance of dead lock that the Python process is waiting for ending mark from JVM, but which is eaten by corrupted stream.
This PR checks the ending mark from Python in non-block way, so it will not blocked by Python process.
There is a small chance that the ending mark is sent by Python process but not available right now, then Python process will not be used.
cc JoshRosen pwendell
Author: Davies Liu <davies@databricks.com>
Closes#4601 from davies/freeze and squashes the following commits:
e15a8c3 [Davies Liu] update logging
890329c [Davies Liu] Merge branch 'freeze' of github.com:davies/spark into freeze
2bd2228 [Davies Liu] add more logging
656d544 [Davies Liu] Update PythonRDD.scala
05e1085 [Davies Liu] check ending mark in non-block way
Added a bunch of tags.
Also changed parquetFile to take varargs rather than a string followed by varargs.
Author: Reynold Xin <rxin@databricks.com>
Closes#4636 from rxin/df-doc and squashes the following commits:
651f80c [Reynold Xin] Fixed parquetFile in PySpark.
8dc3024 [Reynold Xin] [SQL] Various DataFrame doc changes.
This patch improves SubmitRestProtocol's handling of invalid JSON requests in cases where those requests were parsable as JSON but not as JSON objects (e.g. they could be parsed as ararys or strings). I replaced an unchecked cast with pattern-matching and added a new test case.
Author: Josh Rosen <joshrosen@databricks.com>
Closes#4637 from JoshRosen/rest-protocol-cast and squashes the following commits:
b3f282b [Josh Rosen] [SPARK-5849] Handle more types of invalid JSON in SubmitRestProtocolMessage.parseAction
I created a patch that disables the environment variables.
Thereby scala or python shell log a warning message to notify user about the deprecation
with the following message:
scala: "ADD_JARS environment variable is deprecated, use --jar spark submit argument instead"
python: "Warning: ADD_FILES environment variable is deprecated, use --py-files argument instead"
Is it what is expected or the code associated with the variables should be just completely removed?
Should it be somewhere documented?
Author: azagrebin <azagrebin@gmail.com>
Closes#4616 from azagrebin/master and squashes the following commits:
bab1aa9 [azagrebin] [SPARK-3340] Deprecate ADD_JARS and ADD_FILES: minor readability issue
0643895 [azagrebin] [SPARK-3340] Deprecate ADD_JARS and ADD_FILES: add warning messages
42f0107 [azagrebin] [SPARK-3340] Deprecate ADD_JARS and ADD_FILES
The exception in Python writer thread will shutdown executor.
Author: Davies Liu <davies@databricks.com>
Closes#4577 from davies/exception and squashes the following commits:
eb0ceff [Davies Liu] Update PythonRDD.scala
139b0db [Davies Liu] capture the exception in python write thread
The timer is a GC root, and failing to terminate it leaks SparkContext
instances.
Author: Matt Whelan <mwhelan@perka.com>
Closes#4635 from MattWhelan/SPARK-5848 and squashes the following commits:
2a1e8a5 [Matt Whelan] SPARK-5848: teardown the ConsoleProgressBar timer
This PR adds a `ShowTablesCommand` to support `SHOW TABLES [IN databaseName]` SQL command. The result of `SHOW TABLE` has two columns, `tableName` and `isTemporary`. For temporary tables, the value of `isTemporary` column will be `false`.
JIRA: https://issues.apache.org/jira/browse/SPARK-4865
Author: Yin Huai <yhuai@databricks.com>
Closes#4618 from yhuai/showTablesCommand and squashes the following commits:
0c09791 [Yin Huai] Use ShowTablesCommand.
85ee76d [Yin Huai] Since SHOW TABLES is not a Hive native command any more and we will not see "OK" (originally generated by Hive's driver), use SHOW DATABASES in the test.
94bacac [Yin Huai] Add SHOW TABLES to the list of noExplainCommands.
d71ed09 [Yin Huai] Fix test.
a4a6ec3 [Yin Huai] Add SHOW TABLE command.
Existing implementation of arithmetic operators and BinaryComparison operators have redundant type checking codes, e.g.:
Expression.n2 is used by Add/Subtract/Multiply.
(1) n2 always checks left.dataType == right.dataType. However, this checking should be done once when we resolve expression types;
(2) n2 requires dataType is a NumericType. This can be done once.
This PR optimizes arithmetic and predicate operators by removing such redundant type-checking codes.
Some preliminary benchmarking on 10G TPC-H data over 5 r3.2xlarge EC2 machines shows that this PR can reduce the query time by 5.5% to 11%.
The benchmark queries follow the template below, where OP is plus/minus/times/divide/remainder/bitwise and/bitwise or/bitwise xor.
SELECT l_returnflag, l_linestatus, SUM(l_quantity OP cnt1), SUM(l_quantity OP cnt2), ...., SUM(l_quantity OP cnt700)
FROM (
SELECT l_returnflag, l_linestatus, l_quantity, 1 AS cnt1, 2 AS cnt2, ..., 700 AS cnt700
FROM lineitem
WHERE l_shipdate <= '1998-09-01'
)
GROUP BY l_returnflag, l_linestatus;
Author: kai <kaizeng@eecs.berkeley.edu>
Closes#4472 from kai-zeng/arithmetic-optimize and squashes the following commits:
fef0cf1 [kai] Merge branch 'master' of github.com:apache/spark into arithmetic-optimize
4b3a1bb [kai] chmod a-x
5a41e49 [kai] chmod a-x Expression.scala
cb37c94 [kai] rebase onto spark master
7f6e968 [kai] chmod 100755 -> 100644
6cddb46 [kai] format
7490dbc [kai] fix unresolved-expression exception for EqualTo
9c40bc0 [kai] fix bitwisenot
3cbd363 [kai] clean up test code
ca47801 [kai] override evalInternal for bitwise ops
8fa84a1 [kai] add bitwise or and xor
6892fc4 [kai] revert override evalInternal
f8eba24 [kai] override evalInternal
31ccdd4 [kai] rewrite all bitwise op and remove evalInternal
86297e2 [kai] generalized
cb92ae1 [kai] bitwise-and: override eval
97a7d6c [kai] bitwise-and: override evalInternal using and func
0906c39 [kai] add bitwise test
62abbbc [kai] clean up predicate and arithmetic
b34d58d [kai] add caching and benmark option
12c5b32 [kai] override eval
1cd7571 [kai] fix sqrt and maxof
03fd0c3 [kai] fix predicate
16fd84c [kai] optimize + - * / % -(unary) abs < > <= >=
fd95823 [kai] remove unnecessary type checking
24d062f [kai] test suite
JIRA: https://issues.apache.org/jira/browse/SPARK-5839
Author: Yin Huai <yhuai@databricks.com>
Closes#4626 from yhuai/SPARK-5839 and squashes the following commits:
f779d85 [Yin Huai] Use subqeury to wrap replaced ParquetRelation.
2695f13 [Yin Huai] Merge remote-tracking branch 'upstream/master' into SPARK-5839
f1ba6ca [Yin Huai] Address comment.
2c7fa08 [Yin Huai] Use Subqueries to wrap a data source table.
JIRA: https://issues.apache.org/jira/browse/SPARK-5746
liancheng marmbrus
Author: Yin Huai <yhuai@databricks.com>
Closes#4617 from yhuai/insertOverwrite and squashes the following commits:
8e3019d [Yin Huai] Fix compilation error.
499e8e7 [Yin Huai] Merge remote-tracking branch 'upstream/master' into insertOverwrite
e76e85a [Yin Huai] Address comments.
ac31b3c [Yin Huai] Merge remote-tracking branch 'upstream/master' into insertOverwrite
f30bdad [Yin Huai] Use toDF.
99da57e [Yin Huai] Merge remote-tracking branch 'upstream/master' into insertOverwrite
6b7545c [Yin Huai] Add a pre write check to the data source API.
a88c516 [Yin Huai] DDLParser will take a parsering function to take care CTAS statements.
This patch changes PySpark so that the GatewayServer's port is communicated back to the Python process that launches it over a local socket instead of a pipe. The old pipe-based approach was brittle and could fail if `spark-submit` printed unexpected to stdout.
To accomplish this, I wrote a custom `PythonGatewayServer.main()` function to use in place of Py4J's `GatewayServer.main()`.
Closes#3424.
Author: Josh Rosen <joshrosen@databricks.com>
Closes#4603 from JoshRosen/SPARK-2313 and squashes the following commits:
6a7740b [Josh Rosen] Remove EchoOutputThread since it's no longer needed
0db501f [Josh Rosen] Use select() so that we don't block if GatewayServer dies.
9bdb4b6 [Josh Rosen] Handle case where getListeningPort returns -1
3fb7ed1 [Josh Rosen] Remove stdout=PIPE
2458934 [Josh Rosen] Use underscore to mark env var. as private
d12c95d [Josh Rosen] Use Logging and Utils.tryOrExit()
e5f9730 [Josh Rosen] Wrap everything in a giant try-block
2f70689 [Josh Rosen] Use stdin PIPE to share fate with driver
8bf956e [Josh Rosen] Initial cut at passing Py4J gateway port back to driver via socket
Resolves https://issues.apache.org/jira/browse/SPARK-5357
In commons-codec 1.5, Base64 instances are not thread safe. That was only true from 1.4-1.6.
Author: Matt Whelan <mwhelan@perka.com>
Closes#4153 from MattWhelan/depsUpdate and squashes the following commits:
b4a91f4 [Matt Whelan] SPARK-5357: Update commons-codec version to 1.10 (current)
After a call to stop, the shutdown hook is redundant, and causes a
memory leak.
Author: Matt Whelan <mwhelan@perka.com>
Closes#4627 from MattWhelan/SPARK-5841 and squashes the following commits:
d5f5c7f [Matt Whelan] SPARK-5841: remove DiskBlockManager shutdown hook on stop
Lifts `HiveMetastoreCatalog.refreshTable` to `Catalog`. Adds `RefreshTable` command to refresh (possibly cached) metadata in external data sources tables.
<!-- Reviewable:start -->
[<img src="https://reviewable.io/review_button.png" height=40 alt="Review on Reviewable"/>](https://reviewable.io/reviews/apache/spark/4624)
<!-- Reviewable:end -->
Author: Cheng Lian <lian@databricks.com>
Closes#4624 from liancheng/refresh-table and squashes the following commits:
8d1aa4c [Cheng Lian] Adds REFRESH TABLE command
This PR adds the following filter types for data sources API:
- `IsNull`
- `IsNotNull`
- `Not`
- `And`
- `Or`
The code which converts Catalyst predicate expressions to data sources filters is very similar to filter conversion logics in `ParquetFilters` which converts Catalyst predicates to Parquet filter predicates. In this way we can support nested AND/OR/NOT predicates without changing current `BaseScan` type hierarchy.
<!-- Reviewable:start -->
[<img src="https://reviewable.io/review_button.png" height=40 alt="Review on Reviewable"/>](https://reviewable.io/reviews/apache/spark/4623)
<!-- Reviewable:end -->
Author: Cheng Lian <lian@databricks.com>
This patch had conflicts when merged, resolved by
Committer: Michael Armbrust <michael@databricks.com>
Closes#4623 from liancheng/more-fiters and squashes the following commits:
1b296f4 [Cheng Lian] Add more filter types for data sources API
before this change:
```scala
Time taken: 0.619 seconds
```
after this change :
```scala
Time taken: 0.619 seconds, Fetched: 4 row(s)
```
Author: OopsOutOfMemory <victorshengli@126.com>
Closes#4604 from OopsOutOfMemory/rowcount and squashes the following commits:
7252dea [OopsOutOfMemory] add fetched row count
Author: Michael Armbrust <michael@databricks.com>
Closes#4587 from marmbrus/position and squashes the following commits:
0810052 [Michael Armbrust] fix tests
395c019 [Michael Armbrust] Merge remote-tracking branch 'marmbrus/position' into position
e155dce [Michael Armbrust] more errors
f3efa51 [Michael Armbrust] Update AnalysisException.scala
d45ff60 [Michael Armbrust] [SQL] Initial support for reporting location of error in sql string
Author: Daoyuan Wang <daoyuan.wang@intel.com>
Closes#4609 from adrian-wang/ctas and squashes the following commits:
0a75d5a [Daoyuan Wang] reorder import
93d1863 [Daoyuan Wang] add null format in ctas and set default col comment to null
When profiling the Join / Aggregate queries via VisualVM, I noticed lots of `SpecificMutableRow` objects created, as well as the `MutableValue`, since the `SpecificMutableRow` are mostly used in data source implementation, but the `copy` method could be called multiple times in upper modules (e.g. in Join / aggregation etc.), duplicated instances created should be avoid.
Author: Cheng Hao <hao.cheng@intel.com>
Closes#4619 from chenghao-intel/specific_mutable_row and squashes the following commits:
9300d23 [Cheng Hao] update the SpecificMutableRow.copy
Revise JavaPairDStream API declaration on saveAs Hadoop methods, to allow it to be called directly as intended.
CC tdas for review
Author: Sean Owen <sowen@cloudera.com>
Closes#4608 from srowen/SPARK-5795 and squashes the following commits:
36f1ead [Sean Owen] Add code that shows compile problem and fix
036bd27 [Sean Owen] Revise JavaPairDStream API declaration on saveAs Hadoop methods, to allow it to be called directly as intended.
Now, deprecated runSVDPlusPlus and update run, for 1.4.0 / master only
Author: Sean Owen <sowen@cloudera.com>
Closes#4625 from srowen/SPARK-5815.2 and squashes the following commits:
6fd2ca5 [Sean Owen] Now, deprecated runSVDPlusPlus and update run, for 1.4.0 / master only
This PR migrates the Parquet data source to the new data source write support API. Now users can also overwriting and appending to existing tables. Notice that inserting into partitioned tables is not supported yet.
When Parquet data source is enabled, insertion to Hive Metastore Parquet tables is also fullfilled by the Parquet data source. This is done by the newly introduced `HiveMetastoreCatalog.ParquetConversions` rule, which is a "proper" implementation of the original hacky `HiveStrategies.ParquetConversion`. The latter is still preserved, and can be removed together with the old Parquet support in the future.
TODO:
- [x] Update outdated comments in `newParquet.scala`.
<!-- Reviewable:start -->
[<img src="https://reviewable.io/review_button.png" height=40 alt="Review on Reviewable"/>](https://reviewable.io/reviews/apache/spark/4563)
<!-- Reviewable:end -->
Author: Cheng Lian <lian@databricks.com>
Closes#4563 from liancheng/parquet-refining and squashes the following commits:
fa98d27 [Cheng Lian] Fixes test cases which should disable off Parquet data source
2476e82 [Cheng Lian] Fixes compilation error introduced during rebasing
a83d290 [Cheng Lian] Passes Hive Metastore partitioning information to ParquetRelation2
<!-- Reviewable:start -->
[<img src="https://reviewable.io/review_button.png" height=40 alt="Review on Reviewable"/>](https://reviewable.io/reviews/apache/spark/4613)
<!-- Reviewable:end -->
Author: Cheng Lian <lian@databricks.com>
Closes#4613 from liancheng/df-implicit-rename and squashes the following commits:
db8bdd3 [Cheng Lian] Renames stringRddToDataFrame to stringRddToDataFrameHolder for consistency
On a big dataset explicitly unpersist train and validation folds allows to load more data into memory in the next loop iteration. On my environment (single node 8Gb worker RAM, 2 GB dataset file, 3 folds for cross validation), saved more than 5 minutes.
Author: Peter Rudenko <petro.rudenko@gmail.com>
Closes#4595 from petro-rudenko/patch-2 and squashes the following commits:
66a7cfb [Peter Rudenko] Move validationDataset cache to declaration
c5f3265 [Peter Rudenko] [Ml] SPARK-5804 Explicitly manage cache in Crossvalidator k-fold loop
If it's a last estimator in Pipeline there's no need to transform data, since there's no next stage that would consume this data.
Author: Peter Rudenko <petro.rudenko@gmail.com>
Closes#4590 from petro-rudenko/patch-1 and squashes the following commits:
d13ec33 [Peter Rudenko] [Ml] SPARK-5796 Don't transform data on a last estimator in Pipeline
Deprecate SVDPlusPlus.run and introduce SVDPlusPlus.runSVDPlusPlus with return type that doesn't include DoubleMatrix
CC mengxr
Author: Sean Owen <sowen@cloudera.com>
Closes#4614 from srowen/SPARK-5815 and squashes the following commits:
288cb05 [Sean Owen] Clarify deprecation plans in scaladoc
497458e [Sean Owen] Deprecate SVDPlusPlus.run and introduce SVDPlusPlus.runSVDPlusPlus with return type that doesn't include DoubleMatrix
This PR allow Python users to set params in constructors and in setParams, where we use decorator `keyword_only` to force keyword arguments. The trade-off is discussed in the design doc of SPARK-4586.
Generated doc:
CC: davies rxin
Author: Xiangrui Meng <meng@databricks.com>
Closes#4564 from mengxr/py-pipeline-kw and squashes the following commits:
fedf720 [Xiangrui Meng] use toDF
d565f2c [Xiangrui Meng] Merge remote-tracking branch 'apache/master' into py-pipeline-kw
cbc15d3 [Xiangrui Meng] fix style
5032097 [Xiangrui Meng] update pipeline signature
950774e [Xiangrui Meng] simplify keyword_only and update constructor/setParams signatures
fdde5fc [Xiangrui Meng] fix style
c9384b8 [Xiangrui Meng] fix sphinx doc
8e59180 [Xiangrui Meng] add setParams and make constructors take params, where we force keyword args
Exclude libgfortran, libgcc bundled by JBLAS for Windows. This much is simple, and solves the essential license issue. But the more important question is whether MLlib works on Windows then.
Author: Sean Owen <sowen@cloudera.com>
Closes#4453 from srowen/SPARK-5669 and squashes the following commits:
734dd86 [Sean Owen] Exclude libgfortran, libgcc bundled by JBLAS, affecting Windows / OS X / Linux 32-bit (not Linux 64-bit)
User guide for isotonic regression added to docs/mllib-regression.md including code examples for Scala and Java.
Author: martinzapletal <zapletal-martin@email.cz>
Closes#4536 from zapletal-martin/SPARK-5502 and squashes the following commits:
67fe773 [martinzapletal] SPARK-5502 reworded model prediction rules to use more general language rather than the code/implementation specific terms
80bd4c3 [martinzapletal] SPARK-5502 created docs page for isotonic regression, added links to the page, updated data and examples
7d8136e [martinzapletal] SPARK-5502 Added documentation for Isotonic regression including examples for Scala and Java
504b5c3 [martinzapletal] SPARK-5502 Added documentation for Isotonic regression including examples for Scala and Java
If one tries an example by using copy&paste, throw an exception.
Author: Takeshi Yamamuro <linguin.m.s@gmail.com>
Closes#4615 from maropu/AddMissingImportInSqlContext and squashes the following commits:
ab21b66 [Takeshi Yamamuro] Add missing import in the example of SqlContext
When import the whole project into eclipse as maven project, found that the
src/main/scala & src/test/scala can not be set as source folder as default
behavior, so add a "add-source" goal in scala-maven-plugin to let this work.
Author: gli <gli@redhat.com>
Closes#4531 from ligangty/addsource and squashes the following commits:
4e4db4c [gli] [IDE] cannot import src/main/scala & src/test/scala into eclipse as source folder
Currently, Spark Streaming Programming Guide after updateStateByKey explanation links to file stateful_network_wordcount.py and note "For the complete Scala code ..." for any language tab selected. This is an incoherence.
I've changed the guide and link its pertinent example file. JavaStatefulNetworkWordCount.java example was not created so I added to the commit.
Author: gasparms <gmunoz@stratio.com>
Closes#4589 from gasparms/feature/streaming-guide and squashes the following commits:
7f37f89 [gasparms] More style changes
ec202b0 [gasparms] Follow spark style guide
f527328 [gasparms] Improve example to look like scala example
4d8785c [gasparms] Remove throw exception
e92e6b8 [gasparms] Fix incoherence
92db405 [gasparms] Fix Streaming Programming Guide. Change files according the selected language
This just unpersist()s each RDD in this code that was cache()ed.
Author: Sean Owen <sowen@cloudera.com>
Closes#4234 from srowen/SPARK-3290 and squashes the following commits:
66c1e11 [Sean Owen] unpersist() each RDD that was cache()ed
This patch fixes two difficult-to-reproduce Jenkins test failures in InputOutputMetricsSuite (SPARK-5227 and SPARK-5679). The problem was that WholeTextFileRecordReaderSuite modifies the `fs.local.block.size` Hadoop configuration and this change was affecting subsequent test suites due to Hadoop's caching of FileSystem instances (see HADOOP-8490 for more details).
The fix implemented here is to disable FileSystem caching in WholeTextFileRecordReaderSuite.
Author: Josh Rosen <joshrosen@databricks.com>
Closes#4599 from JoshRosen/inputoutputsuite-fix and squashes the following commits:
47dc447 [Josh Rosen] [SPARK-5227] [SPARK-5679] Disable FileSystem cache in WholeTextFileRecordReaderSuite
This PR adds three groups to the ScalaDoc: `param`, `setParam`, and `getParam`. Params will show up in the generated Scala API doc as the top group. Setters/getters will be at the bottom.
Preview:
Author: Xiangrui Meng <meng@databricks.com>
Closes#4600 from mengxr/SPARK-5730 and squashes the following commits:
febed9a [Xiangrui Meng] add doc groups to spark.ml components
because ArrayBuffer is not specialized.
Author: Xiangrui Meng <meng@databricks.com>
Closes#4594 from mengxr/SPARK-5803 and squashes the following commits:
1261bd5 [Xiangrui Meng] merge master
a4ea872 [Xiangrui Meng] use ArrayBuilder to build primitive arrays
