## What changes were proposed in this pull request?
This is a followup pr of #20450.
We should've enabled `MutableColumnarRow.getMap()` as well.
## How was this patch tested?
Existing tests.
Author: Takuya UESHIN <ueshin@databricks.com>
Closes#20471 from ueshin/issues/SPARK-23280/fup2.
## What changes were proposed in this pull request?
This is a follow-up of #20450 which broke lint-java checks.
This pr fixes the lint-java issues.
```
[ERROR] src/main/java/org/apache/spark/sql/vectorized/ColumnVector.java:[20,8] (imports) UnusedImports: Unused import - org.apache.spark.sql.catalyst.util.MapData.
[ERROR] src/main/java/org/apache/spark/sql/vectorized/ColumnarArray.java:[21,8] (imports) UnusedImports: Unused import - org.apache.spark.sql.catalyst.util.MapData.
[ERROR] src/main/java/org/apache/spark/sql/vectorized/ColumnarRow.java:[22,8] (imports) UnusedImports: Unused import - org.apache.spark.sql.catalyst.util.MapData.
```
## How was this patch tested?
Checked manually in my local environment.
Author: Takuya UESHIN <ueshin@databricks.com>
Closes#20468 from ueshin/issues/SPARK-23280/fup1.
## What changes were proposed in this pull request?
This PR include the following changes:
- Make the capacity of `VectorizedParquetRecordReader` configurable;
- Make the capacity of `OrcColumnarBatchReader` configurable;
- Update the error message when required capacity in writable columnar vector cannot be fulfilled.
## How was this patch tested?
N/A
Author: Xingbo Jiang <xingbo.jiang@databricks.com>
Closes#20361 from jiangxb1987/vectorCapacity.
## What changes were proposed in this pull request?
1. create a new package for partitioning/distribution related classes.
As Spark will add new concrete implementations of `Distribution` in new releases, it is good to
have a new package for partitioning/distribution related classes.
2. move streaming related class to package `org.apache.spark.sql.sources.v2.reader/writer.streaming`, instead of `org.apache.spark.sql.sources.v2.streaming.reader/writer`.
So that the there won't be package reader/writer inside package streaming, which is quite confusing.
Before change:
```
v2
├── reader
├── streaming
│ ├── reader
│ └── writer
└── writer
```
After change:
```
v2
├── reader
│ └── streaming
└── writer
└── streaming
```
## How was this patch tested?
Unit test.
Author: Wang Gengliang <ltnwgl@gmail.com>
Closes#20435 from gengliangwang/new_pkg.
## What changes were proposed in this pull request?
Currently, we scan the execution plan of the data source, first the unsafe operation of each row of data, and then re traverse the data for the count of rows. In terms of performance, this is not necessary. this PR combines the two operations and makes statistics on the number of rows while performing the unsafe operation.
Before modified,
```
val unsafeRow = rdd.mapPartitionsWithIndexInternal { (index, iter) =>
val proj = UnsafeProjection.create(schema)
proj.initialize(index)
iter.map(proj)
}
val numOutputRows = longMetric("numOutputRows")
unsafeRow.map { r =>
numOutputRows += 1
r
}
```
After modified,
val numOutputRows = longMetric("numOutputRows")
rdd.mapPartitionsWithIndexInternal { (index, iter) =>
val proj = UnsafeProjection.create(schema)
proj.initialize(index)
iter.map( r => {
numOutputRows += 1
proj(r)
})
}
## How was this patch tested?
the existed test cases.
Author: caoxuewen <cao.xuewen@zte.com.cn>
Closes#20415 from heary-cao/DataSourceScanExec.
## What changes were proposed in this pull request?
Fill the last missing piece of `ColumnVector`: the map type support.
The idea is similar to the array type support. A map is basically 2 arrays: keys and values. We ask the implementations to provide a key array, a value array, and an offset and length to specify the range of this map in the key/value array.
In `WritableColumnVector`, we put the key array in first child vector, and value array in second child vector, and offsets and lengths in the current vector, which is very similar to how array type is implemented here.
## How was this patch tested?
a new test
Author: Wenchen Fan <wenchen@databricks.com>
Closes#20450 from cloud-fan/map.
## What changes were proposed in this pull request?
Here is the test snippet.
``` SQL
scala> Seq[(Integer, Integer)](
| (1, 1),
| (1, 3),
| (2, 3),
| (3, 3),
| (4, null),
| (5, null)
| ).toDF("key", "value").createOrReplaceTempView("src")
scala> sql(
| """
| |SELECT MAX(value) as value, key as col2
| |FROM src
| |GROUP BY key
| |ORDER BY value desc, key
| """.stripMargin).show
+-----+----+
|value|col2|
+-----+----+
| 3| 3|
| 3| 2|
| 3| 1|
| null| 5|
| null| 4|
+-----+----+
```SQL
Here is the explain output :
```SQL
== Parsed Logical Plan ==
'Sort ['value DESC NULLS LAST, 'key ASC NULLS FIRST], true
+- 'Aggregate ['key], ['MAX('value) AS value#9, 'key AS col2#10]
+- 'UnresolvedRelation `src`
== Analyzed Logical Plan ==
value: int, col2: int
Project [value#9, col2#10]
+- Sort [value#9 DESC NULLS LAST, col2#10 DESC NULLS LAST], true
+- Aggregate [key#5], [max(value#6) AS value#9, key#5 AS col2#10]
+- SubqueryAlias src
+- Project [_1#2 AS key#5, _2#3 AS value#6]
+- LocalRelation [_1#2, _2#3]
``` SQL
The sort direction is being wrongly changed from ASC to DSC while resolving ```Sort``` in
resolveAggregateFunctions.
The above testcase models TPCDS-Q71 and thus we have the same issue in Q71 as well.
## How was this patch tested?
A few tests are added in SQLQuerySuite.
Author: Dilip Biswal <dbiswal@us.ibm.com>
Closes#20453 from dilipbiswal/local_spark.
## What changes were proposed in this pull request?
Change DataSourceScanExec so that when grouping blocks together into partitions, also checks the end of the sorted list of splits to more efficiently fill out partitions.
## How was this patch tested?
Updated old test to reflect the new logic, which causes the # of partitions to drop from 4 -> 3
Also, a current test exists to test large non-splittable files at c575977a59/sql/core/src/test/scala/org/apache/spark/sql/execution/datasources/FileSourceStrategySuite.scala (L346)
## Rationale
The current bin-packing method of next-fit descending for blocks into partitions is sub-optimal in a lot of cases and will result in extra partitions, un-even distribution of block-counts across partitions, and un-even distribution of partition sizes.
As an example, 128 files ranging from 1MB, 2MB,...127MB,128MB. will result in 82 partitions with the current algorithm, but only 64 using this algorithm. Also in this example, the max # of blocks per partition in NFD is 13, while in this algorithm is is 2.
More generally, running a simulation of 1000 runs using 128MB blocksize, between 1-1000 normally distributed file sizes between 1-500Mb, you can see an improvement of approx 5% reduction of partition counts, and a large reduction in standard deviation of blocks per partition.
This algorithm also runs in O(n) time as NFD does, and in every case is strictly better results than NFD.
Overall, the more even distribution of blocks across partitions and therefore reduced partition counts should result in a small but significant performance increase across the board
Author: Glen Takahashi <gtakahashi@palantir.com>
Closes#20372 from glentakahashi/feature/improved-block-merging.
## What changes were proposed in this pull request?
In https://github.com/apache/spark/pull/19980 , we thought `anyNullsSet` can be simply implemented by `numNulls() > 0`. This is logically true, but may have performance problems.
`OrcColumnVector` is an example. It doesn't have the `numNulls` property, only has a `noNulls` property. We will lose a lot of performance if we use `numNulls() > 0` to check null.
This PR simply revert #19980, with a renaming to call it `hasNull`. Better name suggestions are welcome, e.g. `nullable`?
## How was this patch tested?
existing test
Author: Wenchen Fan <wenchen@databricks.com>
Closes#20452 from cloud-fan/null.
## What changes were proposed in this pull request?
`ColumnVector` is aimed to support all the data types, but `CalendarIntervalType` is missing. Actually we do support interval type for inner fields, e.g. `ColumnarRow`, `ColumnarArray` both support interval type. It's weird if we don't support interval type at the top level.
This PR adds the interval type support.
This PR also makes `ColumnVector.getChild` protect. We need it public because `MutableColumnaRow.getInterval` needs it. Now the interval implementation is in `ColumnVector.getInterval`.
## How was this patch tested?
a new test.
Author: Wenchen Fan <wenchen@databricks.com>
Closes#20438 from cloud-fan/interval.
## What changes were proposed in this pull request?
Console sink will redistribute collected local data and trigger a distributed job in each batch, this is not necessary, so here change to local job.
## How was this patch tested?
Existing UT and manual verification.
Author: jerryshao <sshao@hortonworks.com>
Closes#20447 from jerryshao/console-minor.
## What changes were proposed in this pull request?
This PR is to fix the `ReplaceExceptWithFilter` rule when the right's Filter contains the references that are not in the left output.
Before this PR, we got the error like
```
java.util.NoSuchElementException: key not found: a
at scala.collection.MapLike$class.default(MapLike.scala:228)
at scala.collection.AbstractMap.default(Map.scala:59)
at scala.collection.MapLike$class.apply(MapLike.scala:141)
at scala.collection.AbstractMap.apply(Map.scala:59)
```
After this PR, `ReplaceExceptWithFilter ` will not take an effect in this case.
## How was this patch tested?
Added tests
Author: gatorsmile <gatorsmile@gmail.com>
Closes#20444 from gatorsmile/fixReplaceExceptWithFilter.
## What changes were proposed in this pull request?
Still saw the performance regression introduced by `spark.sql.codegen.hugeMethodLimit` in our internal workloads. There are two major issues in the current solution.
- The size of the complied byte code is not identical to the bytecode size of the method. The detection is still not accurate.
- The bytecode size of a single operator (e.g., `SerializeFromObject`) could still exceed 8K limit. We saw the performance regression in such scenario.
Since it is close to the release of 2.3, we decide to increase it to 64K for avoiding the perf regression.
## How was this patch tested?
N/A
Author: gatorsmile <gatorsmile@gmail.com>
Closes#20434 from gatorsmile/revertConf.
## What changes were proposed in this pull request?
It is reported that the test `Cancelling stage in a query with Range` in `DataFrameRangeSuite` fails a few times in unrelated PRs. I personally also saw it too in my PR.
This test is not very flaky actually but only fails occasionally. Based on how the test works, I guess that is because `range` finishes before the listener calls `cancelStage`.
I increase the range number from `1000000000L` to `100000000000L` and count the range in one partition. I also reduce the `interval` of checking stage id. Hopefully it can make the test not flaky anymore.
## How was this patch tested?
The modified tests.
Author: Liang-Chi Hsieh <viirya@gmail.com>
Closes#20431 from viirya/SPARK-23222.
## What changes were proposed in this pull request?
Rename the public APIs and names of pandas udfs.
- `PANDAS SCALAR UDF` -> `SCALAR PANDAS UDF`
- `PANDAS GROUP MAP UDF` -> `GROUPED MAP PANDAS UDF`
- `PANDAS GROUP AGG UDF` -> `GROUPED AGG PANDAS UDF`
## How was this patch tested?
The existing tests
Author: gatorsmile <gatorsmile@gmail.com>
Closes#20428 from gatorsmile/renamePandasUDFs.
## What changes were proposed in this pull request?
All other classes in the reader/writer package doesn't have `V2` in their names, and the streaming reader/writer don't have `V2` either. It's more consistent to remove `V2` from `DataSourceV2Reader` and `DataSourceVWriter`.
Also rename `DataSourceV2Option` to remote the `V2`, we should only have `V2` in the root interface: `DataSourceV2`.
This PR also fixes some places that the mix-in interface doesn't extend the interface it aimed to mix in.
## How was this patch tested?
existing tests.
Author: Wenchen Fan <wenchen@databricks.com>
Closes#20427 from cloud-fan/ds-v2.
## What changes were proposed in this pull request?
It's not obvious from the comments that any added column must be a
function of the dataset that we are adding it to. Add a comment to
that effect to Scala, Python and R Data* methods.
Author: Henry Robinson <henry@cloudera.com>
Closes#20429 from henryr/SPARK-23157.
## What changes were proposed in this pull request?
In `ShuffleExchangeExec`, we don't need to insert extra local sort before round-robin partitioning, if the new partitioning has only 1 partition, because under that case all output rows go to the same partition.
## How was this patch tested?
The existing test cases.
Author: Xingbo Jiang <xingbo.jiang@databricks.com>
Closes#20426 from jiangxb1987/repartition1.
## What changes were proposed in this pull request?
This PR is to update the description of the join algorithm changes.
## How was this patch tested?
N/A
Author: gatorsmile <gatorsmile@gmail.com>
Closes#20420 from gatorsmile/followUp22916.
## What changes were proposed in this pull request?
It is a common pattern to apply multiple transforms to a `Dataset` (using `Dataset.withColumn` for example. This is currently quite expensive because we run `CheckAnalysis` on the full plan and create an encoder for each intermediate `Dataset`.
This PR extends the usage of the `AnalysisBarrier` to include `CheckAnalysis`. By doing this we hide the already analyzed plan from `CheckAnalysis` because barrier is a `LeafNode`. The `AnalysisBarrier` is in the `FinishAnalysis` phase of the optimizer.
We also make binding the `Dataset` encoder lazy. The bound encoder is only needed when we materialize the dataset.
## How was this patch tested?
Existing test should cover this.
Author: Herman van Hovell <hvanhovell@databricks.com>
Closes#20402 from hvanhovell/SPARK-23223.
## What changes were proposed in this pull request?
Correct some improper with view related method usage
Only change test cases
like:
```
test("list global temp views") {
try {
sql("CREATE GLOBAL TEMP VIEW v1 AS SELECT 3, 4")
sql("CREATE TEMP VIEW v2 AS SELECT 1, 2")
checkAnswer(sql(s"SHOW TABLES IN $globalTempDB"),
Row(globalTempDB, "v1", true) ::
Row("", "v2", true) :: Nil)
assert(spark.catalog.listTables(globalTempDB).collect().toSeq.map(_.name) == Seq("v1", "v2"))
} finally {
spark.catalog.dropTempView("v1")
spark.catalog.dropGlobalTempView("v2")
}
}
```
other change please review the code.
## How was this patch tested?
See test case.
Author: xubo245 <601450868@qq.com>
Closes#20250 from xubo245/DropTempViewError.
## What changes were proposed in this pull request?
Currently we have `ReadTask` in data source v2 reader, while in writer we have `DataWriterFactory`.
To make the naming consistent and better, renaming `ReadTask` to `DataReaderFactory`.
## How was this patch tested?
Unit test
Author: Wang Gengliang <ltnwgl@gmail.com>
Closes#20397 from gengliangwang/rename.
## What changes were proposed in this pull request?
Replace streaming V2 sinks with a unified StreamWriteSupport interface, with a shim to use it with microbatch execution.
Add a new SQL config to use for disabling V2 sinks, falling back to the V1 sink implementation.
## How was this patch tested?
Existing tests, which in the case of Kafka (the only existing continuous V2 sink) now use V2 for microbatch.
Author: Jose Torres <jose@databricks.com>
Closes#20369 from jose-torres/streaming-sink.
## What changes were proposed in this pull request?
Fix typo in ScalaDoc for DataFrameWriter - originally stated "This is applicable for all file-based data sources (e.g. Parquet, JSON) staring Spark 2.1.0", should be "starting with Spark 2.1.0".
## How was this patch tested?
Check of correct spelling in ScalaDoc
Please review http://spark.apache.org/contributing.html before opening a pull request.
Author: CCInCharge <charles.l.chen.clc@gmail.com>
Closes#20417 from CCInCharge/master.
## What changes were proposed in this pull request?
`lastExecution.executedPlan` is lazy val so accessing it in StreamTest may need to acquire the lock of `lastExecution`. It may be waiting forever when the streaming thread is holding it and running a continuous Spark job.
This PR changes to check if `s.lastExecution` is null to avoid accessing `lastExecution.executedPlan`.
## How was this patch tested?
Jenkins
Author: Jose Torres <jose@databricks.com>
Closes#20413 from zsxwing/SPARK-23245.
## What changes were proposed in this pull request?
This PR fixes Scala/Java doc examples in `Trigger.java`.
## How was this patch tested?
N/A.
Author: Dongjoon Hyun <dongjoon@apache.org>
Closes#20401 from dongjoon-hyun/SPARK-TRIGGER.
## What changes were proposed in this pull request?
This is a regression introduced by https://github.com/apache/spark/pull/19864
When we lookup cache, we should not carry the hint info, as this cache entry might be added by a plan having hint info, while the input plan for this lookup may not have hint info, or have different hint info.
## How was this patch tested?
a new test.
Author: Wenchen Fan <wenchen@databricks.com>
Closes#20394 from cloud-fan/cache.
## What changes were proposed in this pull request?
Currently shuffle repartition uses RoundRobinPartitioning, the generated result is nondeterministic since the sequence of input rows are not determined.
The bug can be triggered when there is a repartition call following a shuffle (which would lead to non-deterministic row ordering), as the pattern shows below:
upstream stage -> repartition stage -> result stage
(-> indicate a shuffle)
When one of the executors process goes down, some tasks on the repartition stage will be retried and generate inconsistent ordering, and some tasks of the result stage will be retried generating different data.
The following code returns 931532, instead of 1000000:
```
import scala.sys.process._
import org.apache.spark.TaskContext
val res = spark.range(0, 1000 * 1000, 1).repartition(200).map { x =>
x
}.repartition(200).map { x =>
if (TaskContext.get.attemptNumber == 0 && TaskContext.get.partitionId < 2) {
throw new Exception("pkill -f java".!!)
}
x
}
res.distinct().count()
```
In this PR, we propose a most straight-forward way to fix this problem by performing a local sort before partitioning, after we make the input row ordering deterministic, the function from rows to partitions is fully deterministic too.
The downside of the approach is that with extra local sort inserted, the performance of repartition() will go down, so we add a new config named `spark.sql.execution.sortBeforeRepartition` to control whether this patch is applied. The patch is default enabled to be safe-by-default, but user may choose to manually turn it off to avoid performance regression.
This patch also changes the output rows ordering of repartition(), that leads to a bunch of test cases failure because they are comparing the results directly.
## How was this patch tested?
Add unit test in ExchangeSuite.
With this patch(and `spark.sql.execution.sortBeforeRepartition` set to true), the following query returns 1000000:
```
import scala.sys.process._
import org.apache.spark.TaskContext
spark.conf.set("spark.sql.execution.sortBeforeRepartition", "true")
val res = spark.range(0, 1000 * 1000, 1).repartition(200).map { x =>
x
}.repartition(200).map { x =>
if (TaskContext.get.attemptNumber == 0 && TaskContext.get.partitionId < 2) {
throw new Exception("pkill -f java".!!)
}
x
}
res.distinct().count()
res7: Long = 1000000
```
Author: Xingbo Jiang <xingbo.jiang@databricks.com>
Closes#20393 from jiangxb1987/shuffle-repartition.
## What changes were proposed in this pull request?
`ColumnVector` is very flexible about how to implement array type. As a result `ColumnVector` has 3 abstract methods for array type: `arrayData`, `getArrayOffset`, `getArrayLength`. For example, in `WritableColumnVector` we use the first child vector as the array data vector, and store offsets and lengths in 2 arrays in the parent vector. `ArrowColumnVector` has a different implementation.
This PR simplifies `ColumnVector` by using only one abstract method for array type: `getArray`.
## How was this patch tested?
existing tests.
rerun `ColumnarBatchBenchmark`, there is no performance regression.
Author: Wenchen Fan <wenchen@databricks.com>
Closes#20395 from cloud-fan/vector.
## What changes were proposed in this pull request?
**Proposal**
Add a per-query ID to the codegen stages as represented by `WholeStageCodegenExec` operators. This ID will be used in
- the explain output of the physical plan, and in
- the generated class name.
Specifically, this ID will be stable within a query, counting up from 1 in depth-first post-order for all the `WholeStageCodegenExec` inserted into a plan.
The ID value 0 is reserved for "free-floating" `WholeStageCodegenExec` objects, which may have been created for one-off purposes, e.g. for fallback handling of codegen stages that failed to codegen the whole stage and wishes to codegen a subset of the children operators (as seen in `org.apache.spark.sql.execution.FileSourceScanExec#doExecute`).
Example: for the following query:
```scala
scala> spark.conf.set("spark.sql.autoBroadcastJoinThreshold", 1)
scala> val df1 = spark.range(10).select('id as 'x, 'id + 1 as 'y).orderBy('x).select('x + 1 as 'z, 'y)
df1: org.apache.spark.sql.DataFrame = [z: bigint, y: bigint]
scala> val df2 = spark.range(5)
df2: org.apache.spark.sql.Dataset[Long] = [id: bigint]
scala> val query = df1.join(df2, 'z === 'id)
query: org.apache.spark.sql.DataFrame = [z: bigint, y: bigint ... 1 more field]
```
The explain output before the change is:
```scala
scala> query.explain
== Physical Plan ==
*SortMergeJoin [z#9L], [id#13L], Inner
:- *Sort [z#9L ASC NULLS FIRST], false, 0
: +- Exchange hashpartitioning(z#9L, 200)
: +- *Project [(x#3L + 1) AS z#9L, y#4L]
: +- *Sort [x#3L ASC NULLS FIRST], true, 0
: +- Exchange rangepartitioning(x#3L ASC NULLS FIRST, 200)
: +- *Project [id#0L AS x#3L, (id#0L + 1) AS y#4L]
: +- *Range (0, 10, step=1, splits=8)
+- *Sort [id#13L ASC NULLS FIRST], false, 0
+- Exchange hashpartitioning(id#13L, 200)
+- *Range (0, 5, step=1, splits=8)
```
Note how codegen'd operators are annotated with a prefix `"*"`. See how the `SortMergeJoin` operator and its direct children `Sort` operators are adjacent and all annotated with the `"*"`, so it's hard to tell they're actually in separate codegen stages.
and after this change it'll be:
```scala
scala> query.explain
== Physical Plan ==
*(6) SortMergeJoin [z#9L], [id#13L], Inner
:- *(3) Sort [z#9L ASC NULLS FIRST], false, 0
: +- Exchange hashpartitioning(z#9L, 200)
: +- *(2) Project [(x#3L + 1) AS z#9L, y#4L]
: +- *(2) Sort [x#3L ASC NULLS FIRST], true, 0
: +- Exchange rangepartitioning(x#3L ASC NULLS FIRST, 200)
: +- *(1) Project [id#0L AS x#3L, (id#0L + 1) AS y#4L]
: +- *(1) Range (0, 10, step=1, splits=8)
+- *(5) Sort [id#13L ASC NULLS FIRST], false, 0
+- Exchange hashpartitioning(id#13L, 200)
+- *(4) Range (0, 5, step=1, splits=8)
```
Note that the annotated prefix becomes `"*(id) "`. See how the `SortMergeJoin` operator and its direct children `Sort` operators have different codegen stage IDs.
It'll also show up in the name of the generated class, as a suffix in the format of `GeneratedClass$GeneratedIterator$id`.
For example, note how `GeneratedClass$GeneratedIteratorForCodegenStage3` and `GeneratedClass$GeneratedIteratorForCodegenStage6` in the following stack trace corresponds to the IDs shown in the explain output above:
```
"Executor task launch worker for task 42412957" daemon prio=5 tid=0x58 nid=NA runnable
java.lang.Thread.State: RUNNABLE
at org.apache.spark.sql.execution.UnsafeExternalRowSorter.insertRow(UnsafeExternalRowSorter.java:109)
at org.apache.spark.sql.catalyst.expressions.GeneratedClass$GeneratedIteratorForCodegenStage3.sort_addToSorter$(generated.java:32)
at org.apache.spark.sql.catalyst.expressions.GeneratedClass$GeneratedIteratorForCodegenStage3.processNext(generated.java:41)
at org.apache.spark.sql.execution.BufferedRowIterator.hasNext(BufferedRowIterator.java:43)
at org.apache.spark.sql.execution.WholeStageCodegenExec$$anonfun$9$$anon$1.hasNext(WholeStageCodegenExec.scala:494)
at org.apache.spark.sql.catalyst.expressions.GeneratedClass$GeneratedIteratorForCodegenStage6.findNextInnerJoinRows$(generated.java:42)
at org.apache.spark.sql.catalyst.expressions.GeneratedClass$GeneratedIteratorForCodegenStage6.processNext(generated.java:101)
at org.apache.spark.sql.execution.BufferedRowIterator.hasNext(BufferedRowIterator.java:43)
at org.apache.spark.sql.execution.WholeStageCodegenExec$$anonfun$11$$anon$2.hasNext(WholeStageCodegenExec.scala:513)
at org.apache.spark.sql.execution.SparkPlan$$anonfun$2.apply(SparkPlan.scala:253)
at org.apache.spark.sql.execution.SparkPlan$$anonfun$2.apply(SparkPlan.scala:247)
at org.apache.spark.rdd.RDD$$anonfun$mapPartitionsInternal$1$$anonfun$apply$25.apply(RDD.scala:828)
at org.apache.spark.rdd.RDD$$anonfun$mapPartitionsInternal$1$$anonfun$apply$25.apply(RDD.scala:828)
at org.apache.spark.rdd.MapPartitionsRDD.compute(MapPartitionsRDD.scala:38)
at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:324)
at org.apache.spark.rdd.RDD.iterator(RDD.scala:288)
at org.apache.spark.scheduler.ResultTask.runTask(ResultTask.scala:87)
at org.apache.spark.scheduler.Task.run(Task.scala:109)
at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:345)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
at java.lang.Thread.run(Thread.java:748)
```
**Rationale**
Right now, the codegen from Spark SQL lacks the means to differentiate between a couple of things:
1. It's hard to tell which physical operators are in the same WholeStageCodegen stage. Note that this "stage" is a separate notion from Spark's RDD execution stages; this one is only to delineate codegen units.
There can be adjacent physical operators that are both codegen'd but are in separate codegen stages. Some of this is due to hacky implementation details, such as the case with `SortMergeJoin` and its `Sort` inputs -- they're hard coded to be split into separate stages although both are codegen'd.
When printing out the explain output of the physical plan, you'd only see the codegen'd physical operators annotated with a preceding star (`'*'`) but would have no way to figure out if they're in the same stage.
2. Performance/error diagnosis
The generated code has class/method names that are hard to differentiate between queries or even between codegen stages within the same query. If we use a Java-level profiler to collect profiles, or if we encounter a Java-level exception with a stack trace in it, it's really hard to tell which part of a query it's at.
By introducing a per-query codegen stage ID, we'd at least be able to know which codegen stage (and in turn, which group of physical operators) was a profile tick or an exception happened.
The reason why this proposal uses a per-query ID is because it's stable within a query, so that multiple runs of the same query will see the same resulting IDs. This both benefits understandability for users, and also it plays well with the codegen cache in Spark SQL which uses the generated source code as the key.
The downside to using per-query IDs as opposed to a per-session or globally incrementing ID is of course we can't tell apart different query runs with this ID alone. But for now I believe this is a good enough tradeoff.
## How was this patch tested?
Existing tests. This PR does not involve any runtime behavior changes other than some name changes.
The SQL query test suites that compares explain outputs have been updates to ignore the newly added `codegenStageId`.
Author: Kris Mok <kris.mok@databricks.com>
Closes#20224 from rednaxelafx/wsc-codegenstageid.
## What changes were proposed in this pull request?
Add colRegex API to PySpark
## How was this patch tested?
add a test in sql/tests.py
Author: Huaxin Gao <huaxing@us.ibm.com>
Closes#20390 from huaxingao/spark-23081.
## What changes were proposed in this pull request?
It has been observed in SPARK-21603 that whole-stage codegen suffers performance degradation, if the generated functions are too long to be optimized by JIT.
We basically produce a single function to incorporate generated codes from all physical operators in whole-stage. Thus, it is possibly to grow the size of generated function over a threshold that we can't have JIT optimization for it anymore.
This patch is trying to decouple the logic of consuming rows in physical operators to avoid a giant function processing rows.
## How was this patch tested?
Added tests.
Author: Liang-Chi Hsieh <viirya@gmail.com>
Closes#18931 from viirya/SPARK-21717.
…JSON / text
## What changes were proposed in this pull request?
Fix for JSON and CSV data sources when file names include characters
that would be changed by URL encoding.
## How was this patch tested?
New unit tests for JSON, CSV and text suites
Author: Henry Robinson <henry@cloudera.com>
Closes#20355 from henryr/spark-23148.
## What changes were proposed in this pull request?
We extract Python UDFs in logical aggregate which depends on aggregate expression or grouping key in ExtractPythonUDFFromAggregate rule. But Python UDFs which don't depend on above expressions should also be extracted to avoid the issue reported in the JIRA.
A small code snippet to reproduce that issue looks like:
```python
import pyspark.sql.functions as f
df = spark.createDataFrame([(1,2), (3,4)])
f_udf = f.udf(lambda: str("const_str"))
df2 = df.distinct().withColumn("a", f_udf())
df2.show()
```
Error exception is raised as:
```
: org.apache.spark.sql.catalyst.errors.package$TreeNodeException: Binding attribute, tree: pythonUDF0#50
at org.apache.spark.sql.catalyst.errors.package$.attachTree(package.scala:56)
at org.apache.spark.sql.catalyst.expressions.BindReferences$$anonfun$bindReference$1.applyOrElse(BoundAttribute.scala:91)
at org.apache.spark.sql.catalyst.expressions.BindReferences$$anonfun$bindReference$1.applyOrElse(BoundAttribute.scala:90)
at org.apache.spark.sql.catalyst.trees.TreeNode$$anonfun$2.apply(TreeNode.scala:267)
at org.apache.spark.sql.catalyst.trees.TreeNode$$anonfun$2.apply(TreeNode.scala:267)
at org.apache.spark.sql.catalyst.trees.CurrentOrigin$.withOrigin(TreeNode.scala:70)
at org.apache.spark.sql.catalyst.trees.TreeNode.transformDown(TreeNode.scala:266)
at org.apache.spark.sql.catalyst.trees.TreeNode$$anonfun$transformDown$1.apply(TreeNode.scala:272)
at org.apache.spark.sql.catalyst.trees.TreeNode$$anonfun$transformDown$1.apply(TreeNode.scala:272)
at org.apache.spark.sql.catalyst.trees.TreeNode$$anonfun$4.apply(TreeNode.scala:306)
at org.apache.spark.sql.catalyst.trees.TreeNode.mapProductIterator(TreeNode.scala:187)
at org.apache.spark.sql.catalyst.trees.TreeNode.mapChildren(TreeNode.scala:304)
at org.apache.spark.sql.catalyst.trees.TreeNode.transformDown(TreeNode.scala:272)
at org.apache.spark.sql.catalyst.trees.TreeNode.transform(TreeNode.scala:256)
at org.apache.spark.sql.catalyst.expressions.BindReferences$.bindReference(BoundAttribute.scala:90)
at org.apache.spark.sql.execution.aggregate.HashAggregateExec$$anonfun$38.apply(HashAggregateExec.scala:514)
at org.apache.spark.sql.execution.aggregate.HashAggregateExec$$anonfun$38.apply(HashAggregateExec.scala:513)
```
This exception raises because `HashAggregateExec` tries to bind the aliased Python UDF expression (e.g., `pythonUDF0#50 AS a#44`) to grouping key.
## How was this patch tested?
Added test.
Author: Liang-Chi Hsieh <viirya@gmail.com>
Closes#20360 from viirya/SPARK-23177.
## What changes were proposed in this pull request?
The broadcast hint of the cached plan is lost if we cache the plan. This PR is to correct it.
```Scala
val df1 = spark.createDataFrame(Seq((1, "4"), (2, "2"))).toDF("key", "value")
val df2 = spark.createDataFrame(Seq((1, "1"), (2, "2"))).toDF("key", "value")
broadcast(df2).cache()
df2.collect()
val df3 = df1.join(df2, Seq("key"), "inner")
```
## How was this patch tested?
Added a test.
Author: gatorsmile <gatorsmile@gmail.com>
Closes#20368 from gatorsmile/cachedBroadcastHint.
## What changes were proposed in this pull request?
The hint of the plan segment is lost, if the plan segment is replaced by the cached data.
```Scala
val df1 = spark.createDataFrame(Seq((1, "4"), (2, "2"))).toDF("key", "value")
val df2 = spark.createDataFrame(Seq((1, "1"), (2, "2"))).toDF("key", "value")
df2.cache()
val df3 = df1.join(broadcast(df2), Seq("key"), "inner")
```
This PR is to fix it.
## How was this patch tested?
Added a test
Author: gatorsmile <gatorsmile@gmail.com>
Closes#20365 from gatorsmile/fixBroadcastHintloss.
## What changes were proposed in this pull request?
We need to override the prettyName for bit_length and octet_length for getting the expected auto-generated alias name.
## How was this patch tested?
The existing tests
Author: gatorsmile <gatorsmile@gmail.com>
Closes#20358 from gatorsmile/test2.3More.
## What changes were proposed in this pull request?
Add support for using pandas UDFs with groupby().agg().
This PR introduces a new type of pandas UDF - group aggregate pandas UDF. This type of UDF defines a transformation of multiple pandas Series -> a scalar value. Group aggregate pandas UDFs can be used with groupby().agg(). Note group aggregate pandas UDF doesn't support partial aggregation, i.e., a full shuffle is required.
This PR doesn't support group aggregate pandas UDFs that return ArrayType, StructType or MapType. Support for these types is left for future PR.
## How was this patch tested?
GroupbyAggPandasUDFTests
Author: Li Jin <ice.xelloss@gmail.com>
Closes#19872 from icexelloss/SPARK-22274-groupby-agg.
## What changes were proposed in this pull request?
a new interface which allows data source to report partitioning and avoid shuffle at Spark side.
The design is pretty like the internal distribution/partitioing framework. Spark defines a `Distribution` interfaces and several concrete implementations, and ask the data source to report a `Partitioning`, the `Partitioning` should tell Spark if it can satisfy a `Distribution` or not.
## How was this patch tested?
new test
Author: Wenchen Fan <wenchen@databricks.com>
Closes#20201 from cloud-fan/partition-reporting.
## What changes were proposed in this pull request?
Typo fixes
## How was this patch tested?
Local build / Doc-only changes
Author: Jacek Laskowski <jacek@japila.pl>
Closes#20344 from jaceklaskowski/typo-fixes.
## What changes were proposed in this pull request?
Several improvements:
* provide a default implementation for the batch get methods
* rename `getChildColumn` to `getChild`, which is more concise
* remove `getStruct(int, int)`, it's only used to simplify the codegen, which is an internal thing, we should not add a public API for this purpose.
## How was this patch tested?
existing tests
Author: Wenchen Fan <wenchen@databricks.com>
Closes#20277 from cloud-fan/column-vector.
## What changes were proposed in this pull request?
Revert the unneeded test case changes we made in SPARK-23000
Also fixes the test suites that do not call `super.afterAll()` in the local `afterAll`. The `afterAll()` of `TestHiveSingleton` actually reset the environments.
## How was this patch tested?
N/A
Author: gatorsmile <gatorsmile@gmail.com>
Closes#20341 from gatorsmile/testRelated.
## What changes were proposed in this pull request?
This PR fixes the wrong comment on `org.apache.spark.sql.parquet.row.attributes`
which is useful for UDTs like Vector/Matrix. Please see [SPARK-22320](https://issues.apache.org/jira/browse/SPARK-22320) for the usage.
Originally, [SPARK-19411](bf493686eb (diff-ee26d4c4be21e92e92a02e9f16dbc285L314)) left this behind during removing optional column metadatas. In the same PR, the same comment was removed at line 310-311.
## How was this patch tested?
N/A (This is about comments).
Author: Dongjoon Hyun <dongjoon@apache.org>
Closes#20346 from dongjoon-hyun/minor_comment_parquet.
## What changes were proposed in this pull request?
CheckCartesianProduct raises an AnalysisException also when the join condition is always false/null. In this case, we shouldn't raise it, since the result will not be a cartesian product.
## How was this patch tested?
added UT
Author: Marco Gaido <marcogaido91@gmail.com>
Closes#20333 from mgaido91/SPARK-23087.
[SPARK-21786][SQL] The 'spark.sql.parquet.compression.codec' and 'spark.sql.orc.compression.codec' configuration doesn't take effect on hive table writing
What changes were proposed in this pull request?
Pass ‘spark.sql.parquet.compression.codec’ value to ‘parquet.compression’.
Pass ‘spark.sql.orc.compression.codec’ value to ‘orc.compress’.
How was this patch tested?
Add test.
Note:
This is the same issue mentioned in #19218 . That branch was deleted mistakenly, so make a new pr instead.
gatorsmile maropu dongjoon-hyun discipleforteen
Author: fjh100456 <fu.jinhua6@zte.com.cn>
Author: Takeshi Yamamuro <yamamuro@apache.org>
Author: Wenchen Fan <wenchen@databricks.com>
Author: gatorsmile <gatorsmile@gmail.com>
Author: Yinan Li <liyinan926@gmail.com>
Author: Marcelo Vanzin <vanzin@cloudera.com>
Author: Juliusz Sompolski <julek@databricks.com>
Author: Felix Cheung <felixcheung_m@hotmail.com>
Author: jerryshao <sshao@hortonworks.com>
Author: Li Jin <ice.xelloss@gmail.com>
Author: Gera Shegalov <gera@apache.org>
Author: chetkhatri <ckhatrimanjal@gmail.com>
Author: Joseph K. Bradley <joseph@databricks.com>
Author: Bago Amirbekian <bago@databricks.com>
Author: Xianjin YE <advancedxy@gmail.com>
Author: Bruce Robbins <bersprockets@gmail.com>
Author: zuotingbing <zuo.tingbing9@zte.com.cn>
Author: Kent Yao <yaooqinn@hotmail.com>
Author: hyukjinkwon <gurwls223@gmail.com>
Author: Adrian Ionescu <adrian@databricks.com>
Closes#20087 from fjh100456/HiveTableWriting.
## What changes were proposed in this pull request?
Narrow bound on approx quantile test to epsilon from 2*epsilon to match paper
## How was this patch tested?
Existing tests.
Author: Sean Owen <sowen@cloudera.com>
Closes#20324 from srowen/SPARK-23091.
## What changes were proposed in this pull request?
Several cleanups in `ColumnarBatch`
* remove `schema`. The `ColumnVector`s inside `ColumnarBatch` already have the data type information, we don't need this `schema`.
* remove `capacity`. `ColumnarBatch` is just a wrapper of `ColumnVector`s, not builders, it doesn't need a capacity property.
* remove `DEFAULT_BATCH_SIZE`. As a wrapper, `ColumnarBatch` can't decide the batch size, it should be decided by the reader, e.g. parquet reader, orc reader, cached table reader. The default batch size should also be defined by the reader.
## How was this patch tested?
existing tests.
Author: Wenchen Fan <wenchen@databricks.com>
Closes#20316 from cloud-fan/columnar-batch.
## What changes were proposed in this pull request?
After session cloning in `TestHive`, the conf of the singleton SparkContext for derby DB location is changed to a new directory. The new directory is created in `HiveUtils.newTemporaryConfiguration(useInMemoryDerby = false)`.
This PR is to keep the conf value of `ConfVars.METASTORECONNECTURLKEY.varname` unchanged during the session clone.
## How was this patch tested?
The issue can be reproduced by the command:
> build/sbt -Phive "hive/test-only org.apache.spark.sql.hive.HiveSessionStateSuite org.apache.spark.sql.hive.DataSourceWithHiveMetastoreCatalogSuite"
Also added a test case.
Author: gatorsmile <gatorsmile@gmail.com>
Closes#20328 from gatorsmile/fixTestFailure.
