## What changes were proposed in this pull request?
This pr is to fix an `NullPointerException` issue caused by a following `limit + aggregate` query;
```
scala> val df = Seq(("a", 1), ("b", 2), ("c", 1), ("d", 5)).toDF("id", "value")
scala> df.limit(2).groupBy("id").count().show
WARN TaskSetManager: Lost task 0.0 in stage 9.0 (TID 8204, lvsp20hdn012.stubprod.com): java.lang.NullPointerException
at org.apache.spark.sql.catalyst.expressions.GeneratedClass$GeneratedIterator.agg_doAggregateWithKeys$(Unknown Source)
at org.apache.spark.sql.catalyst.expressions.GeneratedClass$GeneratedIterator.processNext(Unknown Source)
```
The root culprit is that [`$doAgg()`](https://github.com/apache/spark/blob/master/sql/core/src/main/scala/org/apache/spark/sql/execution/aggregate/HashAggregateExec.scala#L596) skips an initialization of [the buffer iterator](https://github.com/apache/spark/blob/master/sql/core/src/main/scala/org/apache/spark/sql/execution/aggregate/HashAggregateExec.scala#L603); `BaseLimitExec` sets `stopEarly=true` and `$doAgg()` exits in the middle without the initialization.
## How was this patch tested?
Added a test to check if no exception happens for limit + aggregates in `DataFrameAggregateSuite.scala`.
Author: Takeshi YAMAMURO <linguin.m.s@gmail.com>
Closes#15980 from maropu/SPARK-18528.
## What changes were proposed in this pull request?
Made update mode public. As part of that here are the changes.
- Update DatastreamWriter to accept "update"
- Changed package of InternalOutputModes from o.a.s.sql to o.a.s.sql.catalyst
- Added update mode state removing with watermark to StateStoreSaveExec
## How was this patch tested?
Added new tests in changed modules
Author: Tathagata Das <tathagata.das1565@gmail.com>
Closes#16360 from tdas/SPARK-18234.
## What changes were proposed in this pull request?
Currently, Spark writes a single file out per task, sometimes leading to very large files. It would be great to have an option to limit the max number of records written per file in a task, to avoid humongous files.
This patch introduces a new write config option `maxRecordsPerFile` (default to a session-wide setting `spark.sql.files.maxRecordsPerFile`) that limits the max number of records written to a single file. A non-positive value indicates there is no limit (same behavior as not having this flag).
## How was this patch tested?
Added test cases in PartitionedWriteSuite for both dynamic partition insert and non-dynamic partition insert.
Author: Reynold Xin <rxin@databricks.com>
Closes#16204 from rxin/SPARK-18775.
## What changes were proposed in this pull request?
Two changes
- Fix how delays specified in months and years are translated to milliseconds
- Following up on #16258, not show watermark when there is no watermarking in the query
## How was this patch tested?
Updated and new unit tests
Author: Tathagata Das <tathagata.das1565@gmail.com>
Closes#16304 from tdas/SPARK-18834-1.
## What changes were proposed in this pull request?
It's a huge waste to call `Catalog.listTables` in `SQLContext.tableNames`, which only need the table names, while `Catalog.listTables` will get the table metadata for each table name.
## How was this patch tested?
N/A
Author: Wenchen Fan <wenchen@databricks.com>
Closes#16352 from cloud-fan/minor.
### What changes were proposed in this pull request?
Currently, we only have a SQL interface for recovering all the partitions in the directory of a table and update the catalog. `MSCK REPAIR TABLE` or `ALTER TABLE table RECOVER PARTITIONS`. (Actually, very hard for me to remember `MSCK` and have no clue what it means)
After the new "Scalable Partition Handling", the table repair becomes much more important for making visible the data in the created data source partitioned table.
Thus, this PR is to add it into the Catalog interface. After this PR, users can repair the table by
```Scala
spark.catalog.recoverPartitions("testTable")
```
### How was this patch tested?
Modified the existing test cases.
Author: gatorsmile <gatorsmile@gmail.com>
Closes#16356 from gatorsmile/repairTable.
## What changes were proposed in this pull request?
Checkpoint Location can be defined for a StructuredStreaming on a per-query basis by the `DataStreamWriter` options, but it can also be provided through SparkSession configurations. It should be able to recover in both cases when the OutputMode is Complete for MemorySinks.
## How was this patch tested?
Unit tests
Author: Burak Yavuz <brkyvz@gmail.com>
Closes#16342 from brkyvz/chk-rec.
## What changes were proposed in this pull request?
When we append data to an existing table with `DataFrameWriter.saveAsTable`, we will do various checks to make sure the appended data is consistent with the existing data.
However, we get the information of the existing table by matching the table relation, instead of looking at the table metadata. This is error-prone, e.g. we only check the number of columns for `HadoopFsRelation`, we forget to check bucketing, etc.
This PR refactors the error checking by looking at the metadata of the existing table, and fix several bugs:
* SPARK-18899: We forget to check if the specified bucketing matched the existing table, which may lead to a problematic table that has different bucketing in different data files.
* SPARK-18912: We forget to check the number of columns for non-file-based data source table
* SPARK-18913: We don't support append data to a table with special column names.
## How was this patch tested?
new regression test.
Author: Wenchen Fan <wenchen@databricks.com>
Closes#16313 from cloud-fan/bug1.
## What changes were proposed in this pull request?
In order to respond to task cancellation, Spark tasks must periodically check `TaskContext.isInterrupted()`, but this check is missing on a few critical read paths used in Spark SQL, including `FileScanRDD`, `JDBCRDD`, and UnsafeSorter-based sorts. This can cause interrupted / cancelled tasks to continue running and become zombies (as also described in #16189).
This patch aims to fix this problem by adding `TaskContext.isInterrupted()` checks to these paths. Note that I could have used `InterruptibleIterator` to simply wrap a bunch of iterators but in some cases this would have an adverse performance penalty or might not be effective due to certain special uses of Iterators in Spark SQL. Instead, I inlined `InterruptibleIterator`-style logic into existing iterator subclasses.
## How was this patch tested?
Tested manually in `spark-shell` with two different reproductions of non-cancellable tasks, one involving scans of huge files and another involving sort-merge joins that spill to disk. Both causes of zombie tasks are fixed by the changes added here.
Author: Josh Rosen <joshrosen@databricks.com>
Closes#16340 from JoshRosen/sql-task-interruption.
## What changes were proposed in this pull request?
This PR proposes to fix lint-check failures and javadoc8 break.
Few errors were introduced as below:
**lint-check failures**
```
[ERROR] src/test/java/org/apache/spark/network/TransportClientFactorySuite.java:[45,1] (imports) RedundantImport: Duplicate import to line 43 - org.apache.spark.network.util.MapConfigProvider.
[ERROR] src/main/java/org/apache/spark/unsafe/types/CalendarInterval.java:[255,10] (modifier) RedundantModifier: Redundant 'final' modifier.
```
**javadoc8**
```
[error] .../spark/sql/core/target/java/org/apache/spark/sql/streaming/StreamingQueryProgress.java:19: error: bad use of '>'
[error] * "max" -> "2016-12-05T20:54:20.827Z" // maximum event time seen in this trigger
[error] ^
[error] .../spark/sql/core/target/java/org/apache/spark/sql/streaming/StreamingQueryProgress.java:20: error: bad use of '>'
[error] * "min" -> "2016-12-05T20:54:20.827Z" // minimum event time seen in this trigger
[error] ^
[error] .../spark/sql/core/target/java/org/apache/spark/sql/streaming/StreamingQueryProgress.java:21: error: bad use of '>'
[error] * "avg" -> "2016-12-05T20:54:20.827Z" // average event time seen in this trigger
[error] ^
[error] .../spark/sql/core/target/java/org/apache/spark/sql/streaming/StreamingQueryProgress.java:22: error: bad use of '>'
[error] * "watermark" -> "2016-12-05T20:54:20.827Z" // watermark used in this trigger
[error]
```
## How was this patch tested?
Manually checked as below:
**lint-check failures**
```
./dev/lint-java
Checkstyle checks passed.
```
**javadoc8**
This seems hidden in the API doc but I manually checked after removing access modifier as below:
It looks not rendering properly (scaladoc).
![2016-12-16 3 40 34](https://cloud.githubusercontent.com/assets/6477701/21255175/8df1fe6e-c3ad-11e6-8cda-ce7f76c6677a.png)
After this PR, it renders as below:
- scaladoc
![2016-12-16 3 40 23](https://cloud.githubusercontent.com/assets/6477701/21255135/4a11dab6-c3ad-11e6-8ab2-b091c4f45029.png)
- javadoc
![2016-12-16 3 41 10](https://cloud.githubusercontent.com/assets/6477701/21255137/4bba1d9c-c3ad-11e6-9b88-62f1f697b56a.png)
Author: hyukjinkwon <gurwls223@gmail.com>
Closes#16307 from HyukjinKwon/lint-javadoc8.
## What changes were proposed in this pull request?
A vectorized parquet reader fails to read column data if data schema and partition schema overlap with each other and inferred types in the partition schema differ from ones in the data schema. An example code to reproduce this bug is as follows;
```
scala> case class A(a: Long, b: Int)
scala> val as = Seq(A(1, 2))
scala> spark.createDataFrame(as).write.parquet("/data/a=1/")
scala> val df = spark.read.parquet("/data/")
scala> df.printSchema
root
|-- a: long (nullable = true)
|-- b: integer (nullable = true)
scala> df.collect
java.lang.NullPointerException
at org.apache.spark.sql.execution.vectorized.OnHeapColumnVector.getLong(OnHeapColumnVector.java:283)
at org.apache.spark.sql.execution.vectorized.ColumnarBatch$Row.getLong(ColumnarBatch.java:191)
at org.apache.spark.sql.catalyst.expressions.GeneratedClass$SpecificUnsafeProjection.apply(Unknown Source)
at org.apache.spark.sql.catalyst.expressions.GeneratedClass$SpecificUnsafeProjection.apply(Unknown Source)
at scala.collection.Iterator$$anon$11.next(Iterator.scala:409)
at scala.collection.Iterator$$anon$11.next(Iterator.scala:409)
```
The root cause is that a logical layer (`HadoopFsRelation`) and a physical layer (`VectorizedParquetRecordReader`) have a different assumption on partition schema; the logical layer trusts the data schema to infer the type the overlapped partition columns, and, on the other hand, the physical layer trusts partition schema which is inferred from path string. To fix this bug, this pr simply updates `HadoopFsRelation.schema` to respect the partition columns position in data schema and respect the partition columns type in partition schema.
## How was this patch tested?
Add tests in `ParquetPartitionDiscoverySuite`
Author: Takeshi YAMAMURO <linguin.m.s@gmail.com>
Closes#16030 from maropu/SPARK-18108.
## What changes were proposed in this pull request?
This PR adds StreamingQueryWrapper to make StreamExecution and progress classes serializable because it is too easy for it to get captured with normal usage. If StreamingQueryWrapper gets captured in a closure but no place calls its methods, it should not fail the Spark tasks. However if its methods are called, then this PR will throw a better message.
## How was this patch tested?
`test("StreamingQuery should be Serializable but cannot be used in executors")`
`test("progress classes should be Serializable")`
Author: Shixiong Zhu <shixiong@databricks.com>
Closes#16272 from zsxwing/SPARK-18850.
## What changes were proposed in this pull request?
When starting a stream with a lot of backfill and maxFilesPerTrigger, the user could often want to start with most recent files first. This would let you keep low latency for recent data and slowly backfill historical data.
This PR adds a new option `latestFirst` to control this behavior. When it's true, `FileStreamSource` will sort the files by the modified time from latest to oldest, and take the first `maxFilesPerTrigger` files as a new batch.
## How was this patch tested?
The added test.
Author: Shixiong Zhu <shixiong@databricks.com>
Closes#16251 from zsxwing/newest-first.
## What changes were proposed in this pull request?
Right now, once a user set the comment of a column with create table command, he/she cannot update the comment. It will be useful to provide a public interface (e.g. SQL) to do that.
This PR implements the following SQL statement:
```
ALTER TABLE table [PARTITION partition_spec]
CHANGE [COLUMN] column_old_name column_new_name column_dataType
[COMMENT column_comment]
[FIRST | AFTER column_name];
```
For further expansion, we could support alter `name`/`dataType`/`index` of a column too.
## How was this patch tested?
Add new test cases in `ExternalCatalogSuite` and `SessionCatalogSuite`.
Add sql file test for `ALTER TABLE CHANGE COLUMN` statement.
Author: jiangxingbo <jiangxb1987@gmail.com>
Closes#15717 from jiangxb1987/change-column.
## What changes were proposed in this pull request?
In `DataSource`, if the table is not analyzed, we will use 0 as the default value for table size. This is dangerous, we may broadcast a large table and cause OOM. We should use `defaultSizeInBytes` instead.
## How was this patch tested?
new regression test
Author: Wenchen Fan <wenchen@databricks.com>
Closes#16280 from cloud-fan/bug.
## What changes were proposed in this pull request?
This is a bug introduced by subquery handling. numberedTreeString (which uses generateTreeString under the hood) numbers trees including innerChildren (used to print subqueries), but apply (which uses getNodeNumbered) ignores innerChildren. As a result, apply(i) would return the wrong plan node if there are subqueries.
This patch fixes the bug.
## How was this patch tested?
Added a test case in SubquerySuite.scala to test both the depth-first traversal of numbering as well as making sure the two methods are consistent.
Author: Reynold Xin <rxin@databricks.com>
Closes#16277 from rxin/SPARK-18854.
## What changes were proposed in this pull request?
Right now `StreamingQuery.lastProgress` throws NoSuchElementException and it's hard to be used in Python since Python user will just see Py4jError.
This PR just makes it return null instead.
## How was this patch tested?
`test("lastProgress should be null when recentProgress is empty")`
Author: Shixiong Zhu <shixiong@databricks.com>
Closes#16273 from zsxwing/SPARK-18852.
## What changes were proposed in this pull request?
Currently, `FileSourceStrategy` does not handle the case when the pushed-down filter is `Literal(null)` and removes it at the post-filter in Spark-side.
For example, the codes below:
```scala
val df = Seq(Tuple1(Some(true)), Tuple1(None), Tuple1(Some(false))).toDF()
df.filter($"_1" === "true").explain(true)
```
shows it keeps `null` properly.
```
== Parsed Logical Plan ==
'Filter ('_1 = true)
+- LocalRelation [_1#17]
== Analyzed Logical Plan ==
_1: boolean
Filter (cast(_1#17 as double) = cast(true as double))
+- LocalRelation [_1#17]
== Optimized Logical Plan ==
Filter (isnotnull(_1#17) && null)
+- LocalRelation [_1#17]
== Physical Plan ==
*Filter (isnotnull(_1#17) && null) << Here `null` is there
+- LocalTableScan [_1#17]
```
However, when we read it back from Parquet,
```scala
val path = "/tmp/testfile"
df.write.parquet(path)
spark.read.parquet(path).filter($"_1" === "true").explain(true)
```
`null` is removed at the post-filter.
```
== Parsed Logical Plan ==
'Filter ('_1 = true)
+- Relation[_1#11] parquet
== Analyzed Logical Plan ==
_1: boolean
Filter (cast(_1#11 as double) = cast(true as double))
+- Relation[_1#11] parquet
== Optimized Logical Plan ==
Filter (isnotnull(_1#11) && null)
+- Relation[_1#11] parquet
== Physical Plan ==
*Project [_1#11]
+- *Filter isnotnull(_1#11) << Here `null` is missing
+- *FileScan parquet [_1#11] Batched: true, Format: ParquetFormat, Location: InMemoryFileIndex[file:/tmp/testfile], PartitionFilters: [null], PushedFilters: [IsNotNull(_1)], ReadSchema: struct<_1:boolean>
```
This PR fixes it to keep it properly. In more details,
```scala
val partitionKeyFilters =
ExpressionSet(normalizedFilters.filter(_.references.subsetOf(partitionSet)))
```
This keeps this `null` in `partitionKeyFilters` as `Literal` always don't have `children` and `references` is being empty which is always the subset of `partitionSet`.
And then in
```scala
val afterScanFilters = filterSet -- partitionKeyFilters
```
`null` is always removed from the post filter. So, if the referenced fields are empty, it should be applied into data columns too.
After this PR, it becomes as below:
```
== Parsed Logical Plan ==
'Filter ('_1 = true)
+- Relation[_1#276] parquet
== Analyzed Logical Plan ==
_1: boolean
Filter (cast(_1#276 as double) = cast(true as double))
+- Relation[_1#276] parquet
== Optimized Logical Plan ==
Filter (isnotnull(_1#276) && null)
+- Relation[_1#276] parquet
== Physical Plan ==
*Project [_1#276]
+- *Filter (isnotnull(_1#276) && null)
+- *FileScan parquet [_1#276] Batched: true, Format: ParquetFormat, Location: InMemoryFileIndex[file:/private/var/folders/9j/gf_c342d7d150mwrxvkqnc180000gn/T/spark-a5d59bdb-5b..., PartitionFilters: [null], PushedFilters: [IsNotNull(_1)], ReadSchema: struct<_1:boolean>
```
## How was this patch tested?
Unit test in `FileSourceStrategySuite`
Author: hyukjinkwon <gurwls223@gmail.com>
Closes#16184 from HyukjinKwon/SPARK-18753.
## What changes were proposed in this pull request?
`OverwriteOptions` was introduced in https://github.com/apache/spark/pull/15705, to carry the information of static partitions. However, after further refactor, this information becomes duplicated and we can remove `OverwriteOptions`.
## How was this patch tested?
N/A
Author: Wenchen Fan <wenchen@databricks.com>
Closes#15995 from cloud-fan/overwrite.
## What changes were proposed in this pull request?
Add implicit encoders for BigDecimal, timestamp and date.
## How was this patch tested?
Add an unit test. Pass build, unit tests, and some tests below .
Before:
```
scala> spark.createDataset(Seq(new java.math.BigDecimal(10)))
<console>:24: error: Unable to find encoder for type stored in a Dataset. Primitive types (Int, String, etc) and Product types (case classes) are supported by importing spark.implicits._ Support for serializing other types will be added in future releases.
spark.createDataset(Seq(new java.math.BigDecimal(10)))
^
scala>
```
After:
```
scala> spark.createDataset(Seq(new java.math.BigDecimal(10)))
res0: org.apache.spark.sql.Dataset[java.math.BigDecimal] = [value: decimal(38,18)]
```
Author: Weiqing Yang <yangweiqing001@gmail.com>
Closes#16176 from weiqingy/SPARK-18746.
## What changes were proposed in this pull request?
- Changed `StreamingQueryProgress.watermark` to `StreamingQueryProgress.queryTimestamps` which is a `Map[String, String]` containing the following keys: "eventTime.max", "eventTime.min", "eventTime.avg", "processingTime", "watermark". All of them UTC formatted strings.
- Renamed `StreamingQuery.timestamp` to `StreamingQueryProgress.triggerTimestamp` to differentiate from `queryTimestamps`. It has the timestamp of when the trigger was started.
## How was this patch tested?
Updated tests
Author: Tathagata Das <tathagata.das1565@gmail.com>
Closes#16258 from tdas/SPARK-18834.
## What changes were proposed in this pull request?
Change the statement `SHOW TABLES [EXTENDED] [(IN|FROM) database_name] [[LIKE] 'identifier_with_wildcards'] [PARTITION(partition_spec)]` to the following statements:
- SHOW TABLES [(IN|FROM) database_name] [[LIKE] 'identifier_with_wildcards']
- SHOW TABLE EXTENDED [(IN|FROM) database_name] LIKE 'identifier_with_wildcards' [PARTITION(partition_spec)]
After this change, the statements `SHOW TABLE/SHOW TABLES` have the same syntax with that HIVE has.
## How was this patch tested?
Modified the test sql file `show-tables.sql`;
Modified the test suite `DDLSuite`.
Author: jiangxingbo <jiangxb1987@gmail.com>
Closes#16262 from jiangxb1987/show-table-extended.
## What changes were proposed in this pull request?
Some places in SQL may call `RpcEndpointRef.askWithRetry` (e.g., ParquetFileFormat.buildReader -> SparkContext.broadcast -> ... -> BlockManagerMaster.updateBlockInfo -> RpcEndpointRef.askWithRetry), which will finally call `Await.result`. It may cause `java.lang.IllegalArgumentException: spark.sql.execution.id is already set` when running in Scala ForkJoinPool.
This PR includes the following changes to fix this issue:
- Remove `ThreadUtils.awaitResult`
- Rename `ThreadUtils. awaitResultInForkJoinSafely` to `ThreadUtils.awaitResult`
- Replace `Await.result` in RpcTimeout with `ThreadUtils.awaitResult`.
## How was this patch tested?
Jenkins
Author: Shixiong Zhu <shixiong@databricks.com>
Closes#16230 from zsxwing/fix-SPARK-13747.
## What changes were proposed in this pull request?
I **believe** that I _only_ removed duplicated code (that adds nothing but noise). I'm gonna remove the comment after Jenkins has built the changes with no issues and Spark devs has agreed to include the changes.
Remove explicit `RDD` and `Partition` overrides (that turn out code duplication)
## How was this patch tested?
Local build. Awaiting Jenkins.
…cation)
Author: Jacek Laskowski <jacek@japila.pl>
Closes#16145 from jaceklaskowski/rdd-overrides-removed.
## What changes were proposed in this pull request?
Major change in this PR:
- Add `pendingQueryNames` and `pendingQueryIds` to track that are going to start but not yet put into `activeQueries` so that we don't need to hold a lock when starting a query.
Minor changes:
- Fix a potential NPE when the user sets `checkpointLocation` using SQLConf but doesn't specify a query name.
- Add missing docs in `StreamingQueryListener`
## How was this patch tested?
Jenkins
Author: Shixiong Zhu <shixiong@databricks.com>
Closes#16220 from zsxwing/SPARK-18796.
The value of the "isSrcLocal" parameter passed to Hive's loadTable and
loadPartition methods needs to be set according to the user query (e.g.
"LOAD DATA LOCAL"), and not the current code that tries to guess what
it should be.
For existing versions of Hive the current behavior is probably ok, but
some recent changes in the Hive code changed the semantics slightly,
making code that sets "isSrcLocal" to "true" incorrectly to do the
wrong thing. It would end up moving the parent directory of the files
into the final location, instead of the file themselves, resulting
in a table that cannot be read.
I modified HiveCommandSuite so that existing "LOAD DATA" tests are run
both in local and non-local mode, since the semantics are slightly different.
The tests include a few new checks to make sure the semantics follow
what Hive describes in its documentation.
Tested with existing unit tests and also ran some Hive integration tests
with a version of Hive containing the changes that surfaced the problem.
Author: Marcelo Vanzin <vanzin@cloudera.com>
Closes#16179 from vanzin/SPARK-18752.
The problem is if it is run with no fix throws an exception and causes the following error:
"Cannot specify a column width on data type bit."
The problem stems from the fact that the "java.sql.types.BIT" type is mapped as BIT[n] that really must be mapped as BIT.
This concerns the type Boolean.
As for the type String with maximum length of characters it must be mapped as VARCHAR (MAX) instead of TEXT which is a type deprecated in SQLServer.
Here is the list of mappings for SQL Server:
https://msdn.microsoft.com/en-us/library/ms378878(v=sql.110).aspxCloses#13944 from meknio/master.
## What changes were proposed in this pull request?
Instead of only keeping the minimum number of offsets around, we should keep enough information to allow us to roll back n batches and reexecute the stream starting from a given point. In particular, we should create a config in SQLConf, spark.sql.streaming.retainedBatches that defaults to 100 and ensure that we keep enough log files in the following places to roll back the specified number of batches:
the offsets that are present in each batch
versions of the state store
the files lists stored for the FileStreamSource
the metadata log stored by the FileStreamSink
marmbrus zsxwing
## How was this patch tested?
The following tests were added.
### StreamExecution offset metadata
Test added to StreamingQuerySuite that ensures offset metadata is garbage collected according to minBatchesRetain
### CompactibleFileStreamLog
Tests added in CompactibleFileStreamLogSuite to ensure that logs are purged starting before the first compaction file that proceeds the current batch id - minBatchesToRetain.
Please review http://spark.apache.org/contributing.html before opening a pull request.
Author: Tyson Condie <tcondie@gmail.com>
Closes#16219 from tcondie/offset_hist.
### What changes were proposed in this pull request?
Currently, when users use Python UDF in Filter, BatchEvalPython is always generated below FilterExec. However, not all the predicates need to be evaluated after Python UDF execution. Thus, this PR is to push down the determinisitc predicates through `BatchEvalPython`.
```Python
>>> df = spark.createDataFrame([(1, "1"), (2, "2"), (1, "2"), (1, "2")], ["key", "value"])
>>> from pyspark.sql.functions import udf, col
>>> from pyspark.sql.types import BooleanType
>>> my_filter = udf(lambda a: a < 2, BooleanType())
>>> sel = df.select(col("key"), col("value")).filter((my_filter(col("key"))) & (df.value < "2"))
>>> sel.explain(True)
```
Before the fix, the plan looks like
```
== Optimized Logical Plan ==
Filter ((isnotnull(value#1) && <lambda>(key#0L)) && (value#1 < 2))
+- LogicalRDD [key#0L, value#1]
== Physical Plan ==
*Project [key#0L, value#1]
+- *Filter ((isnotnull(value#1) && pythonUDF0#9) && (value#1 < 2))
+- BatchEvalPython [<lambda>(key#0L)], [key#0L, value#1, pythonUDF0#9]
+- Scan ExistingRDD[key#0L,value#1]
```
After the fix, the plan looks like
```
== Optimized Logical Plan ==
Filter ((isnotnull(value#1) && <lambda>(key#0L)) && (value#1 < 2))
+- LogicalRDD [key#0L, value#1]
== Physical Plan ==
*Project [key#0L, value#1]
+- *Filter pythonUDF0#9: boolean
+- BatchEvalPython [<lambda>(key#0L)], [key#0L, value#1, pythonUDF0#9]
+- *Filter (isnotnull(value#1) && (value#1 < 2))
+- Scan ExistingRDD[key#0L,value#1]
```
### How was this patch tested?
Added both unit test cases for `BatchEvalPythonExec` and also add an end-to-end test case in Python test suite.
Author: gatorsmile <gatorsmile@gmail.com>
Closes#16193 from gatorsmile/pythonUDFPredicatePushDown.
## What changes were proposed in this pull request?
1. In SparkStrategies.canBroadcast, I will add the check plan.statistics.sizeInBytes >= 0
2. In LocalRelations.statistics, when calculate the statistics, I will change the size to BigInt so it won't overflow.
## How was this patch tested?
I will add a test case to make sure the statistics.sizeInBytes won't overflow.
Author: Huaxin Gao <huaxing@us.ibm.com>
Closes#16175 from huaxingao/spark-17460.
## What changes were proposed in this pull request?
When you start a stream, if we are trying to resolve the source of the stream, for example if we need to resolve partition columns, this could take a long time. This long execution time should not block the main thread where `query.start()` was called on. It should happen in the stream execution thread possibly before starting any triggers.
## How was this patch tested?
Unit test added. Made sure test fails with no code changes.
Author: Burak Yavuz <brkyvz@gmail.com>
Closes#16238 from brkyvz/SPARK-18811.
## What changes were proposed in this pull request?
This PR avoids that a result of a cast `toInt` is negative due to signed integer overflow (e.g. 0x0000_0000_1???????L.toInt < 0 ). This PR performs casts after we can ensure the value is within range of signed integer (the result of `max(array.length, ???)` is always integer).
## How was this patch tested?
Manually executed query68 of TPC-DS with 100TB
Author: Kazuaki Ishizaki <ishizaki@jp.ibm.com>
Closes#16235 from kiszk/SPARK-18745.
## What changes were proposed in this pull request?
* This PR changes `JVMObjectTracker` from `object` to `class` and let its instance associated with each RBackend. So we can manage the lifecycle of JVM objects when there are multiple `RBackend` sessions. `RBackend.close` will clear the object tracker explicitly.
* I assume that `SQLUtils` and `RRunner` do not need to track JVM instances, which could be wrong.
* Small refactor of `SerDe.sqlSerDe` to increase readability.
## How was this patch tested?
* Added unit tests for `JVMObjectTracker`.
* Wait for Jenkins to run full tests.
Author: Xiangrui Meng <meng@databricks.com>
Closes#16154 from mengxr/SPARK-17822.
## What changes were proposed in this pull request?
- Changed FileStreamSource to use new FileStreamSourceOffset rather than LongOffset. The field is named as `logOffset` to make it more clear that this is a offset in the file stream log.
- Fixed bug in FileStreamSourceLog, the field endId in the FileStreamSourceLog.get(startId, endId) was not being used at all. No test caught it earlier. Only my updated tests caught it.
Other minor changes
- Dont use batchId in the FileStreamSource, as calling it batch id is extremely miss leading. With multiple sources, it may happen that a new batch has no new data from a file source. So offset of FileStreamSource != batchId after that batch.
## How was this patch tested?
Updated unit test.
Author: Tathagata Das <tathagata.das1565@gmail.com>
Closes#16205 from tdas/SPARK-18776.
## What changes were proposed in this pull request?
This patch fixes the format specification in explain for file sources (Parquet and Text formats are the only two that are different from the rest):
Before:
```
scala> spark.read.text("test.text").explain()
== Physical Plan ==
*FileScan text [value#15] Batched: false, Format: org.apache.spark.sql.execution.datasources.text.TextFileFormatxyz, Location: InMemoryFileIndex[file:/scratch/rxin/spark/test.text], PartitionFilters: [], PushedFilters: [], ReadSchema: struct<value:string>
```
After:
```
scala> spark.read.text("test.text").explain()
== Physical Plan ==
*FileScan text [value#15] Batched: false, Format: Text, Location: InMemoryFileIndex[file:/scratch/rxin/spark/test.text], PartitionFilters: [], PushedFilters: [], ReadSchema: struct<value:string>
```
Also closes#14680.
## How was this patch tested?
Verified in spark-shell.
Author: Reynold Xin <rxin@databricks.com>
Closes#16187 from rxin/SPARK-18760.
## What changes were proposed in this pull request?
`input_file_name` doesn't return filename when working with UDF in PySpark. An example shows the problem:
from pyspark.sql.functions import *
from pyspark.sql.types import *
def filename(path):
return path
sourceFile = udf(filename, StringType())
spark.read.json("tmp.json").select(sourceFile(input_file_name())).show()
+---------------------------+
|filename(input_file_name())|
+---------------------------+
| |
+---------------------------+
The cause of this issue is, we group rows in `BatchEvalPythonExec` for batching processing of PythonUDF. Currently we group rows first and then evaluate expressions on the rows. If the data is less than the required number of rows for a group, the iterator will be consumed to the end before the evaluation. However, once the iterator reaches the end, we will unset input filename. So the input_file_name expression can't return correct filename.
This patch fixes the approach to group the batch of rows. We evaluate the expression first and then group evaluated results to batch.
## How was this patch tested?
Added unit test to PySpark.
Please review http://spark.apache.org/contributing.html before opening a pull request.
Author: Liang-Chi Hsieh <viirya@gmail.com>
Closes#16115 from viirya/fix-py-udf-input-filename.
## What changes were proposed in this pull request?
When `ignoreCorruptFiles` is enabled, it's better to also ignore non-existing files.
## How was this patch tested?
Jenkins
Author: Shixiong Zhu <shixiong@databricks.com>
Closes#16203 from zsxwing/ignore-file-not-found.
## What changes were proposed in this pull request?
Listeners added with `sparkSession.streams.addListener(l)` are added to a SparkSession. So events only from queries in the same session as a listener should be posted to the listener. Currently, all the events gets rerouted through the Spark's main listener bus, that is,
- StreamingQuery posts event to StreamingQueryListenerBus. Only the queries associated with the same session as the bus posts events to it.
- StreamingQueryListenerBus posts event to Spark's main LiveListenerBus as a SparkEvent.
- StreamingQueryListenerBus also subscribes to LiveListenerBus events thus getting back the posted event in a different thread.
- The received is posted to the registered listeners.
The problem is that *all StreamingQueryListenerBuses in all sessions* gets the events and posts them to their listeners. This is wrong.
In this PR, I solve it by making StreamingQueryListenerBus track active queries (by their runIds) when a query posts the QueryStarted event to the bus. This allows the rerouted events to be filtered using the tracked queries.
Note that this list needs to be maintained separately
from the `StreamingQueryManager.activeQueries` because a terminated query is cleared from
`StreamingQueryManager.activeQueries` as soon as it is stopped, but the this ListenerBus must
clear a query only after the termination event of that query has been posted lazily, much after the query has been terminated.
Credit goes to zsxwing for coming up with the initial idea.
## How was this patch tested?
Updated test harness code to use the correct session, and added new unit test.
Author: Tathagata Das <tathagata.das1565@gmail.com>
Closes#16186 from tdas/SPARK-18758.
Based on an informal survey, users find this option easier to understand / remember.
Author: Michael Armbrust <michael@databricks.com>
Closes#16182 from marmbrus/renameRecentProgress.
## What changes were proposed in this pull request?
It's better to add a warning log when skipping a corrupted file. It will be helpful when we want to finish the job first, then find them in the log and fix these files.
## How was this patch tested?
Jenkins
Author: Shixiong Zhu <shixiong@databricks.com>
Closes#16192 from zsxwing/SPARK-18764.
## What changes were proposed in this pull request?
Easier to read while debugging as a formatted string (in ISO8601 format) than in millis
## How was this patch tested?
Updated unit tests
Author: Tathagata Das <tathagata.das1565@gmail.com>
Closes#16166 from tdas/SPARK-18734.
## What changes were proposed in this pull request?
Many Spark developers often want to test the runtime of some function in interactive debugging and testing. This patch adds a simple time function to SparkSession:
```
scala> spark.time { spark.range(1000).count() }
Time taken: 77 ms
res1: Long = 1000
```
## How was this patch tested?
I tested this interactively in spark-shell.
Author: Reynold Xin <rxin@databricks.com>
Closes#16140 from rxin/SPARK-18714.
## What changes were proposed in this pull request?
Right now ForeachSink creates a new physical plan, so StreamExecution cannot retrieval metrics and watermark.
This PR changes ForeachSink to manually convert InternalRows to objects without creating a new plan.
## How was this patch tested?
`test("foreach with watermark: append")`.
Author: Shixiong Zhu <shixiong@databricks.com>
Closes#16160 from zsxwing/SPARK-18721.
(Link to Jira issue: https://issues.apache.org/jira/browse/SPARK-18572)
## What changes were proposed in this pull request?
Currently Spark answers the `SHOW PARTITIONS` command by fetching all of the table's partition metadata from the external catalog and constructing partition names therefrom. The Hive client has a `getPartitionNames` method which is many times faster for this purpose, with the performance improvement scaling with the number of partitions in a table.
To test the performance impact of this PR, I ran the `SHOW PARTITIONS` command on two Hive tables with large numbers of partitions. One table has ~17,800 partitions, and the other has ~95,000 partitions. For the purposes of this PR, I'll call the former table `table1` and the latter table `table2`. I ran 5 trials for each table with before-and-after versions of this PR. The results are as follows:
Spark at bdc8153, `SHOW PARTITIONS table1`, times in seconds:
7.901
3.983
4.018
4.331
4.261
Spark at bdc8153, `SHOW PARTITIONS table2`
(Timed out after 10 minutes with a `SocketTimeoutException`.)
Spark at this PR, `SHOW PARTITIONS table1`, times in seconds:
3.801
0.449
0.395
0.348
0.336
Spark at this PR, `SHOW PARTITIONS table2`, times in seconds:
5.184
1.63
1.474
1.519
1.41
Taking the best times from each trial, we get a 12x performance improvement for a table with ~17,800 partitions and at least a 426x improvement for a table with ~95,000 partitions. More significantly, the latter command doesn't even complete with the current code in master.
This is actually a patch we've been using in-house at VideoAmp since Spark 1.1. It's made all the difference in the practical usability of our largest tables. Even with tables with about 1,000 partitions there's a performance improvement of about 2-3x.
## How was this patch tested?
I added a unit test to `VersionsSuite` which tests that the Hive client's `getPartitionNames` method returns the correct number of partitions.
Author: Michael Allman <michael@videoamp.com>
Closes#15998 from mallman/spark-18572-list_partition_names.
## What changes were proposed in this pull request?
Move no data rate limit from StreamExecution to ProgressReporter to make `recentProgresses` and listener events consistent.
## How was this patch tested?
Jenkins
Author: Shixiong Zhu <shixiong@databricks.com>
Closes#16155 from zsxwing/SPARK-18722.
## What changes were proposed in this pull request?
DataSet.na.fill(0) used on a DataSet which has a long value column, it will change the original long value.
The reason is that the type of the function fill's param is Double, and the numeric columns are always cast to double(`fillCol[Double](f, value)`) .
```
def fill(value: Double, cols: Seq[String]): DataFrame = {
val columnEquals = df.sparkSession.sessionState.analyzer.resolver
val projections = df.schema.fields.map { f =>
// Only fill if the column is part of the cols list.
if (f.dataType.isInstanceOf[NumericType] && cols.exists(col => columnEquals(f.name, col))) {
fillCol[Double](f, value)
} else {
df.col(f.name)
}
}
df.select(projections : _*)
}
```
For example:
```
scala> val df = Seq[(Long, Long)]((1, 2), (-1, -2), (9123146099426677101L, 9123146560113991650L)).toDF("a", "b")
df: org.apache.spark.sql.DataFrame = [a: bigint, b: bigint]
scala> df.show
+-------------------+-------------------+
| a| b|
+-------------------+-------------------+
| 1| 2|
| -1| -2|
|9123146099426677101|9123146560113991650|
+-------------------+-------------------+
scala> df.na.fill(0).show
+-------------------+-------------------+
| a| b|
+-------------------+-------------------+
| 1| 2|
| -1| -2|
|9123146099426676736|9123146560113991680|
+-------------------+-------------------+
```
the original values changed [which is not we expected result]:
```
9123146099426677101 -> 9123146099426676736
9123146560113991650 -> 9123146560113991680
```
## How was this patch tested?
unit test added.
Author: root <root@iZbp1gsnrlfzjxh82cz80vZ.(none)>
Closes#15994 from windpiger/nafillMissupOriginalValue.
### What changes were proposed in this pull request?
Our existing withColumn for adding metadata can simply use the existing public withColumn API.
### How was this patch tested?
The existing test cases cover it.
Author: gatorsmile <gatorsmile@gmail.com>
Closes#16152 from gatorsmile/withColumnRefactoring.
## What changes were proposed in this pull request?
Here are the major changes in this PR.
- Added the ability to recover `StreamingQuery.id` from checkpoint location, by writing the id to `checkpointLoc/metadata`.
- Added `StreamingQuery.runId` which is unique for every query started and does not persist across restarts. This is to identify each restart of a query separately (same as earlier behavior of `id`).
- Removed auto-generation of `StreamingQuery.name`. The purpose of name was to have the ability to define an identifier across restarts, but since id is precisely that, there is no need for a auto-generated name. This means name becomes purely cosmetic, and is null by default.
- Added `runId` to `StreamingQueryListener` events and `StreamingQueryProgress`.
Implementation details
- Renamed existing `StreamExecutionMetadata` to `OffsetSeqMetadata`, and moved it to the file `OffsetSeq.scala`, because that is what this metadata is tied to. Also did some refactoring to make the code cleaner (got rid of a lot of `.json` and `.getOrElse("{}")`).
- Added the `id` as the new `StreamMetadata`.
- When a StreamingQuery is created it gets or writes the `StreamMetadata` from `checkpointLoc/metadata`.
- All internal logging in `StreamExecution` uses `(name, id, runId)` instead of just `name`
TODO
- [x] Test handling of name=null in json generation of StreamingQueryProgress
- [x] Test handling of name=null in json generation of StreamingQueryListener events
- [x] Test python API of runId
## How was this patch tested?
Updated unit tests and new unit tests
Author: Tathagata Das <tathagata.das1565@gmail.com>
Closes#16113 from tdas/SPARK-18657.
## What changes were proposed in this pull request?
Move DataFrame.collect out of synchronized block so that we can query content in MemorySink when `DataFrame.collect` is running.
## How was this patch tested?
Jenkins
Author: Shixiong Zhu <shixiong@databricks.com>
Closes#16162 from zsxwing/SPARK-18729.
## What changes were proposed in this pull request?
As reported in the Jira, there are some weird issues with exploding Python UDFs in SparkSQL.
The following test code can reproduce it. Notice: the following test code is reported to return wrong results in the Jira. However, as I tested on master branch, it causes exception and so can't return any result.
>>> from pyspark.sql.functions import *
>>> from pyspark.sql.types import *
>>>
>>> df = spark.range(10)
>>>
>>> def return_range(value):
... return [(i, str(i)) for i in range(value - 1, value + 1)]
...
>>> range_udf = udf(return_range, ArrayType(StructType([StructField("integer_val", IntegerType()),
... StructField("string_val", StringType())])))
>>>
>>> df.select("id", explode(range_udf(df.id))).show()
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/spark/python/pyspark/sql/dataframe.py", line 318, in show
print(self._jdf.showString(n, 20))
File "/spark/python/lib/py4j-0.10.4-src.zip/py4j/java_gateway.py", line 1133, in __call__
File "/spark/python/pyspark/sql/utils.py", line 63, in deco
return f(*a, **kw)
File "/spark/python/lib/py4j-0.10.4-src.zip/py4j/protocol.py", line 319, in get_return_value py4j.protocol.Py4JJavaError: An error occurred while calling o126.showString.: java.lang.AssertionError: assertion failed
at scala.Predef$.assert(Predef.scala:156)
at org.apache.spark.sql.execution.CodegenSupport$class.consume(WholeStageCodegenExec.scala:120)
at org.apache.spark.sql.execution.GenerateExec.consume(GenerateExec.scala:57)
The cause of this issue is, in `ExtractPythonUDFs` we insert `BatchEvalPythonExec` to run PythonUDFs in batch. `BatchEvalPythonExec` will add extra outputs (e.g., `pythonUDF0`) to original plan. In above case, the original `Range` only has one output `id`. After `ExtractPythonUDFs`, the added `BatchEvalPythonExec` has two outputs `id` and `pythonUDF0`.
Because the output of `GenerateExec` is given after analysis phase, in above case, it is the combination of `id`, i.e., the output of `Range`, and `col`. But in planning phase, we change `GenerateExec`'s child plan to `BatchEvalPythonExec` with additional output attributes.
It will cause no problem in non wholestage codegen. Because when evaluating the additional attributes are projected out the final output of `GenerateExec`.
However, as `GenerateExec` now supports wholestage codegen, the framework will input all the outputs of the child plan to `GenerateExec`. Then when consuming `GenerateExec`'s output data (i.e., calling `consume`), the number of output attributes is different to the output variables in wholestage codegen.
To solve this issue, this patch only gives the generator's output to `GenerateExec` after analysis phase. `GenerateExec`'s output is the combination of its child plan's output and the generator's output. So when we change `GenerateExec`'s child, its output is still correct.
## How was this patch tested?
Added test cases to PySpark.
Please review http://spark.apache.org/contributing.html before opening a pull request.
Author: Liang-Chi Hsieh <viirya@gmail.com>
Closes#16120 from viirya/fix-py-udf-with-generator.
## What changes were proposed in this pull request?
- Add StreamingQuery.explain and exception to Python.
- Fix StreamingQueryException to not expose `OffsetSeq`.
## How was this patch tested?
Jenkins
Author: Shixiong Zhu <shixiong@databricks.com>
Closes#16125 from zsxwing/py-streaming-explain.
## What changes were proposed in this pull request?
We currently have function input_file_name to get the path of the input file, but don't have functions to get the block start offset and length. This patch introduces two functions:
1. input_file_block_start: returns the file block start offset, or -1 if not available.
2. input_file_block_length: returns the file block length, or -1 if not available.
## How was this patch tested?
Updated existing test cases in ColumnExpressionSuite that covered input_file_name to also cover the two new functions.
Author: Reynold Xin <rxin@databricks.com>
Closes#16133 from rxin/SPARK-18702.
## What changes were proposed in this pull request?
Even though in 2.1 creating a partitioned datasource table will not populate the partition data by default (until the user issues MSCK REPAIR TABLE), it seems we still scan the filesystem for no good reason.
We should avoid doing this when the user specifies a schema.
## How was this patch tested?
Perf stat tests.
Author: Eric Liang <ekl@databricks.com>
Closes#16090 from ericl/spark-18661.
## What changes were proposed in this pull request?
This patch significantly improves the IO / file listing performance of schema inference in Spark's built-in CSV data source.
Previously, this data source used the legacy `SparkContext.hadoopFile` and `SparkContext.hadoopRDD` methods to read files during its schema inference step, causing huge file-listing bottlenecks on the driver.
This patch refactors this logic to use Spark SQL's `text` data source to read files during this step. The text data source still performs some unnecessary file listing (since in theory we already have resolved the table prior to schema inference and therefore should be able to scan without performing _any_ extra listing), but that listing is much faster and takes place in parallel. In one production workload operating over tens of thousands of files, this change managed to reduce schema inference time from 7 minutes to 2 minutes.
A similar problem also affects the JSON file format and this patch originally fixed that as well, but I've decided to split that change into a separate patch so as not to conflict with changes in another JSON PR.
## How was this patch tested?
Existing unit tests, plus manual benchmarking on a production workload.
Author: Josh Rosen <joshrosen@databricks.com>
Closes#15813 from JoshRosen/use-text-data-source-in-csv-and-json.
## What changes were proposed in this pull request?
This PR adds a sql conf `spark.sql.streaming.noDataReportInterval` to control how long to wait before outputing the next StreamProgressEvent when there is no data.
## How was this patch tested?
The added unit test.
Author: Shixiong Zhu <shixiong@databricks.com>
Closes#16108 from zsxwing/SPARK-18670.
## What changes were proposed in this pull request?
Two bugs are addressed here
1. INSERT OVERWRITE TABLE sometime crashed when catalog partition management was enabled. This was because when dropping partitions after an overwrite operation, the Hive client will attempt to delete the partition files. If the entire partition directory was dropped, this would fail. The PR fixes this by adding a flag to control whether the Hive client should attempt to delete files.
2. The static partition spec for OVERWRITE TABLE was not correctly resolved to the case-sensitive original partition names. This resulted in the entire table being overwritten if you did not correctly capitalize your partition names.
cc yhuai cloud-fan
## How was this patch tested?
Unit tests. Surprisingly, the existing overwrite table tests did not catch these edge cases.
Author: Eric Liang <ekl@databricks.com>
Closes#16088 from ericl/spark-18659.
## What changes were proposed in this pull request?
Currently, `JDBCRelation.insert` removes Spark options too early by mistakenly using `asConnectionProperties`. Spark options like `numPartitions` should be passed into `DataFrameWriter.jdbc` correctly. This bug have been **hidden** because `JDBCOptions.asConnectionProperties` fails to filter out the mixed-case options. This PR aims to fix both.
**JDBCRelation.insert**
```scala
override def insert(data: DataFrame, overwrite: Boolean): Unit = {
val url = jdbcOptions.url
val table = jdbcOptions.table
- val properties = jdbcOptions.asConnectionProperties
+ val properties = jdbcOptions.asProperties
data.write
.mode(if (overwrite) SaveMode.Overwrite else SaveMode.Append)
.jdbc(url, table, properties)
```
**JDBCOptions.asConnectionProperties**
```scala
scala> import org.apache.spark.sql.execution.datasources.jdbc.JDBCOptions
scala> import org.apache.spark.sql.catalyst.util.CaseInsensitiveMap
scala> new JDBCOptions(Map("url" -> "jdbc:mysql://localhost:3306/temp", "dbtable" -> "t1", "numPartitions" -> "10")).asConnectionProperties
res0: java.util.Properties = {numpartitions=10}
scala> new JDBCOptions(new CaseInsensitiveMap(Map("url" -> "jdbc:mysql://localhost:3306/temp", "dbtable" -> "t1", "numPartitions" -> "10"))).asConnectionProperties
res1: java.util.Properties = {numpartitions=10}
```
## How was this patch tested?
Pass the Jenkins with a new testcase.
Author: Dongjoon Hyun <dongjoon@apache.org>
Closes#15863 from dongjoon-hyun/SPARK-18419.
## What changes were proposed in this pull request?
In Spark 2.1 ListingFileCatalog was significantly refactored (and renamed to InMemoryFileIndex). This introduced a regression where parallelism could only be introduced at the very top of the tree. However, in many cases (e.g. `spark.read.parquet(topLevelDir)`), the top of the tree is only a single directory.
This PR simplifies and fixes the parallel recursive listing code to allow parallelism to be introduced at any level during recursive descent (though note that once we decide to list a sub-tree in parallel, the sub-tree is listed in serial on executors).
cc mallman cloud-fan
## How was this patch tested?
Checked metrics in unit tests.
Author: Eric Liang <ekl@databricks.com>
Closes#16112 from ericl/spark-18679.
This PR targets to both master and branch-2.1.
## What changes were proposed in this pull request?
Due to PARQUET-686, Parquet doesn't do string comparison correctly while doing filter push-down for string columns. This PR disables filter push-down for both string and binary columns to work around this issue. Binary columns are also affected because some Parquet data models (like Hive) may store string columns as a plain Parquet `binary` instead of a `binary (UTF8)`.
## How was this patch tested?
New test case added in `ParquetFilterSuite`.
Author: Cheng Lian <lian@databricks.com>
Closes#16106 from liancheng/spark-17213-bad-string-ppd.
## What changes were proposed in this pull request?
This replaces uses of `TextOutputFormat` with an `OutputStream`, which will either write directly to the filesystem or indirectly via a compressor (if so configured). This avoids intermediate buffering.
The inverse of this (reading directly from a stream) is necessary for streaming large JSON records (when `wholeFile` is enabled) so I wanted to keep the read and write paths symmetric.
## How was this patch tested?
Existing unit tests.
Author: Nathan Howell <nhowell@godaddy.com>
Closes#16089 from NathanHowell/SPARK-18658.
## What changes were proposed in this pull request?
SQL query generated for the JDBC data source is not quoting columns in the predicate clause. When the source table has quoted column names, spark jdbc read fails with column not found error incorrectly.
Error:
org.h2.jdbc.JdbcSQLException: Column "ID" not found;
Source SQL statement:
SELECT "Name","Id" FROM TEST."mixedCaseCols" WHERE (Id < 1)
This PR fixes by quoting column names in the generated SQL for predicate clause when filters are pushed down to the data source.
Source SQL statement after the fix:
SELECT "Name","Id" FROM TEST."mixedCaseCols" WHERE ("Id" < 1)
## How was this patch tested?
Added new test case to the JdbcSuite
Author: sureshthalamati <suresh.thalamati@gmail.com>
Closes#15662 from sureshthalamati/filter_quoted_cols-SPARK-18141.
## What changes were proposed in this pull request?
The current error message of USING join is quite confusing, for example:
```
scala> val df1 = List(1,2,3).toDS.withColumnRenamed("value", "c1")
df1: org.apache.spark.sql.DataFrame = [c1: int]
scala> val df2 = List(1,2,3).toDS.withColumnRenamed("value", "c2")
df2: org.apache.spark.sql.DataFrame = [c2: int]
scala> df1.join(df2, usingColumn = "c1")
org.apache.spark.sql.AnalysisException: using columns ['c1] can not be resolved given input columns: [c1, c2] ;;
'Join UsingJoin(Inner,List('c1))
:- Project [value#1 AS c1#3]
: +- LocalRelation [value#1]
+- Project [value#7 AS c2#9]
+- LocalRelation [value#7]
```
after this PR, it becomes:
```
scala> val df1 = List(1,2,3).toDS.withColumnRenamed("value", "c1")
df1: org.apache.spark.sql.DataFrame = [c1: int]
scala> val df2 = List(1,2,3).toDS.withColumnRenamed("value", "c2")
df2: org.apache.spark.sql.DataFrame = [c2: int]
scala> df1.join(df2, usingColumn = "c1")
org.apache.spark.sql.AnalysisException: USING column `c1` can not be resolved with the right join side, the right output is: [c2];
```
## How was this patch tested?
updated tests
Author: Wenchen Fan <wenchen@databricks.com>
Closes#16100 from cloud-fan/natural.
### What changes were proposed in this pull request?
The following two `DataFrameReader` JDBC APIs ignore the user-specified parameters of parallelism degree.
```Scala
def jdbc(
url: String,
table: String,
columnName: String,
lowerBound: Long,
upperBound: Long,
numPartitions: Int,
connectionProperties: Properties): DataFrame
```
```Scala
def jdbc(
url: String,
table: String,
predicates: Array[String],
connectionProperties: Properties): DataFrame
```
This PR is to fix the issues. To verify the behavior correctness, we improve the plan output of `EXPLAIN` command by adding `numPartitions` in the `JDBCRelation` node.
Before the fix,
```
== Physical Plan ==
*Scan JDBCRelation(TEST.PEOPLE) [NAME#1896,THEID#1897] ReadSchema: struct<NAME:string,THEID:int>
```
After the fix,
```
== Physical Plan ==
*Scan JDBCRelation(TEST.PEOPLE) [numPartitions=3] [NAME#1896,THEID#1897] ReadSchema: struct<NAME:string,THEID:int>
```
### How was this patch tested?
Added the verification logics on all the test cases for JDBC concurrent fetching.
Author: gatorsmile <gatorsmile@gmail.com>
Closes#15975 from gatorsmile/jdbc.
## What changes were proposed in this pull request?
Currently we haven't implemented `SHOW TABLE EXTENDED` in Spark 2.0. This PR is to implement the statement.
Goals:
1. Support `SHOW TABLES EXTENDED LIKE 'identifier_with_wildcards'`;
2. Explicitly output an unsupported error message for `SHOW TABLES [EXTENDED] ... PARTITION` statement;
3. Improve test cases for `SHOW TABLES` statement.
## How was this patch tested?
1. Add new test cases in file `show-tables.sql`.
2. Modify tests for `SHOW TABLES` in `DDLSuite`.
Author: jiangxingbo <jiangxb1987@gmail.com>
Closes#15958 from jiangxb1987/show-table-extended.
## What changes were proposed in this pull request?
- Add StreamingQueryStatus.json
- Make it not case class (to avoid unnecessarily exposing implicit object StreamingQueryStatus, consistent with StreamingQueryProgress)
- Add StreamingQuery.status to Python
- Fix post-termination status
## How was this patch tested?
New unit tests
Author: Tathagata Das <tathagata.das1565@gmail.com>
Closes#16075 from tdas/SPARK-18516-1.
## What changes were proposed in this pull request?
`AggregateFunction` currently implements `ImplicitCastInputTypes` (which enables implicit input type casting). There are actually quite a few situations in which we don't need this, or require more control over our input. A recent example is the aggregate for `CountMinSketch` which should only take string, binary or integral types inputs.
This PR removes `ImplicitCastInputTypes` from the `AggregateFunction` and makes a case-by-case decision on what kind of input validation we should use.
## How was this patch tested?
Refactoring only. Existing tests.
Author: Herman van Hovell <hvanhovell@databricks.com>
Closes#16066 from hvanhovell/SPARK-18632.
This PR separates the status of a `StreamingQuery` into two separate APIs:
- `status` - describes the status of a `StreamingQuery` at this moment, including what phase of processing is currently happening and if data is available.
- `recentProgress` - an array of statistics about the most recent microbatches that have executed.
A recent progress contains the following information:
```
{
"id" : "2be8670a-fce1-4859-a530-748f29553bb6",
"name" : "query-29",
"timestamp" : 1479705392724,
"inputRowsPerSecond" : 230.76923076923077,
"processedRowsPerSecond" : 10.869565217391303,
"durationMs" : {
"triggerExecution" : 276,
"queryPlanning" : 3,
"getBatch" : 5,
"getOffset" : 3,
"addBatch" : 234,
"walCommit" : 30
},
"currentWatermark" : 0,
"stateOperators" : [ ],
"sources" : [ {
"description" : "KafkaSource[Subscribe[topic-14]]",
"startOffset" : {
"topic-14" : {
"2" : 0,
"4" : 1,
"1" : 0,
"3" : 0,
"0" : 0
}
},
"endOffset" : {
"topic-14" : {
"2" : 1,
"4" : 2,
"1" : 0,
"3" : 0,
"0" : 1
}
},
"numRecords" : 3,
"inputRowsPerSecond" : 230.76923076923077,
"processedRowsPerSecond" : 10.869565217391303
} ]
}
```
Additionally, in order to make it possible to correlate progress updates across restarts, we change the `id` field from an integer that is unique with in the JVM to a `UUID` that is globally unique.
Author: Tathagata Das <tathagata.das1565@gmail.com>
Author: Michael Armbrust <michael@databricks.com>
Closes#15954 from marmbrus/queryProgress.
## What changes were proposed in this pull request?
Re-partitioning logic in ExchangeCoordinator changed so that adding another pre-shuffle partition to the post-shuffle partition will not be done if doing so would cause the size of the post-shuffle partition to exceed the target partition size.
## How was this patch tested?
Existing tests updated to reflect new expectations.
Author: Mark Hamstra <markhamstra@gmail.com>
Closes#16065 from markhamstra/SPARK-17064.
Revise HDFSMetadataLog API such that metadata object serialization and final batch file write are separated. This will allow serialization checks without worrying about batch file name formats. marmbrus zsxwing
Existing tests already ensure this API faithfully support core functionality i.e., creation of batch files.
Author: Tyson Condie <tcondie@gmail.com>
Closes#15924 from tcondie/SPARK-18498.
Signed-off-by: Michael Armbrust <michael@databricks.com>
## What changes were proposed in this pull request?
This PR make `sbt unidoc` complete with Java 8.
This PR roughly includes several fixes as below:
- Fix unrecognisable class and method links in javadoc by changing it from `[[..]]` to `` `...` ``
```diff
- * A column that will be computed based on the data in a [[DataFrame]].
+ * A column that will be computed based on the data in a `DataFrame`.
```
- Fix throws annotations so that they are recognisable in javadoc
- Fix URL links to `<a href="http..."></a>`.
```diff
- * [[http://en.wikipedia.org/wiki/Decision_tree_learning Decision tree]] model for regression.
+ * <a href="http://en.wikipedia.org/wiki/Decision_tree_learning">
+ * Decision tree (Wikipedia)</a> model for regression.
```
```diff
- * see http://en.wikipedia.org/wiki/Receiver_operating_characteristic
+ * see <a href="http://en.wikipedia.org/wiki/Receiver_operating_characteristic">
+ * Receiver operating characteristic (Wikipedia)</a>
```
- Fix < to > to
- `greater than`/`greater than or equal to` or `less than`/`less than or equal to` where applicable.
- Wrap it with `{{{...}}}` to print them in javadoc or use `{code ...}` or `{literal ..}`. Please refer https://github.com/apache/spark/pull/16013#discussion_r89665558
- Fix `</p>` complaint
## How was this patch tested?
Manually tested by `jekyll build` with Java 7 and 8
```
java version "1.7.0_80"
Java(TM) SE Runtime Environment (build 1.7.0_80-b15)
Java HotSpot(TM) 64-Bit Server VM (build 24.80-b11, mixed mode)
```
```
java version "1.8.0_45"
Java(TM) SE Runtime Environment (build 1.8.0_45-b14)
Java HotSpot(TM) 64-Bit Server VM (build 25.45-b02, mixed mode)
```
Author: hyukjinkwon <gurwls223@gmail.com>
Closes#16013 from HyukjinKwon/SPARK-3359-errors-more.
## What changes were proposed in this pull request?
For the following workflow:
1. I have a column called time which is at minute level precision in a Streaming DataFrame
2. I want to perform groupBy time, count
3. Then I want my MemorySink to only have the last 30 minutes of counts and I perform this by
.where('time >= current_timestamp().cast("long") - 30 * 60)
what happens is that the `filter` gets pushed down before the aggregation, and the filter happens on the source data for the aggregation instead of the result of the aggregation (where I actually want to filter).
I guess the main issue here is that `current_timestamp` is non-deterministic in the streaming context and shouldn't be pushed down the filter.
Does this require us to store the `current_timestamp` for each trigger of the streaming job, that is something to discuss.
Furthermore, we want to persist current batch timestamp and watermark timestamp to the offset log so that these values are consistent across multiple executions of the same batch.
brkyvz zsxwing tdas
## How was this patch tested?
A test was added to StreamingAggregationSuite ensuring the above use case is handled. The test injects a stream of time values (in seconds) to a query that runs in complete mode and only outputs the (count) aggregation results for the past 10 seconds.
Author: Tyson Condie <tcondie@gmail.com>
Closes#15949 from tcondie/SPARK-18339.
## What changes were proposed in this pull request?
We failed to properly propagate table metadata for existing tables for the saveAsTable command. This caused a downstream component to think the table was MANAGED, writing data to the wrong location.
## How was this patch tested?
Unit test that fails before the patch.
Author: Eric Liang <ekl@databricks.com>
Closes#15983 from ericl/spark-18544.
## What changes were proposed in this pull request?
This is absolutely minor. PR https://github.com/apache/spark/pull/15595 uses `dt1.asNullable == dt2.asNullable` expressions in a few places. It is however more efficient to call `dt1.sameType(dt2)`. I have replaced every instance of the first pattern with the second pattern (3/5 were introduced by #15595).
## How was this patch tested?
Existing tests.
Author: Herman van Hovell <hvanhovell@databricks.com>
Closes#16041 from hvanhovell/SPARK-18058.
## What changes were proposed in this pull request?
This PR fixes a random OOM issue occurred while running `ObjectHashAggregateSuite`.
This issue can be steadily reproduced under the following conditions:
1. The aggregation must be evaluated using `ObjectHashAggregateExec`;
2. There must be an input column whose data type involves `ArrayType` (an input column of `MapType` may even cause SIGSEGV);
3. Sort-based aggregation fallback must be triggered during evaluation.
The root cause is that while falling back to sort-based aggregation, we must sort and feed already evaluated partial aggregation buffers living in the hash map to the sort-based aggregator using an external sorter. However, the underlying mutable byte buffer of `UnsafeRow`s produced by the iterator of the external sorter is reused and may get overwritten when the iterator steps forward. After the last entry is consumed, the byte buffer points to a block of uninitialized memory filled by `5a`. Therefore, while reading an `UnsafeArrayData` out of the `UnsafeRow`, `5a5a5a5a` is treated as array size and triggers a memory allocation for a ridiculously large array and immediately blows up the JVM with an OOM.
To fix this issue, we only need to add `.copy()` accordingly.
## How was this patch tested?
New regression test case added in `ObjectHashAggregateSuite`.
Author: Cheng Lian <lian@databricks.com>
Closes#15976 from liancheng/investigate-oom.
## What changes were proposed in this pull request?
`CatalogTable` has a parameter named `tracksPartitionsInCatalog`, and in `CatalogTable.toString` we use `"Partition Provider: Catalog"` to represent it. This PR fixes `DESC TABLE` to make it consistent with `CatalogTable.toString`.
## How was this patch tested?
N/A
Author: Wenchen Fan <wenchen@databricks.com>
Closes#16035 from cloud-fan/minor.
## What changes were proposed in this pull request?
https://github.com/apache/spark/pull/15704 will fail if we use int literal in `DROP PARTITION`, and we have reverted it in branch-2.1.
This PR reverts it in master branch, and add a regression test for it, to make sure the master branch is healthy.
## How was this patch tested?
new regression test
Author: Wenchen Fan <wenchen@databricks.com>
Closes#16036 from cloud-fan/revert.
### What changes were proposed in this pull request?
Currently, the name validation checks are limited to table creation. It is enfored by Analyzer rule: `PreWriteCheck`.
However, table renaming and database creation have the same issues. It makes more sense to do the checks in `SessionCatalog`. This PR is to add it into `SessionCatalog`.
### How was this patch tested?
Added test cases
Author: gatorsmile <gatorsmile@gmail.com>
Closes#16018 from gatorsmile/nameValidate.
## What changes were proposed in this pull request?
This PR is to fix incorrect `code` tag in `sql-programming-guide.md`
## How was this patch tested?
Manually.
Author: Weiqing Yang <yangweiqing001@gmail.com>
Closes#15941 from weiqingy/fixtag.
## What changes were proposed in this pull request?
The expression `in(empty seq)` is invalid in some data source. Since `in(empty seq)` is always false, we should generate `in(empty seq)` to false literal in optimizer.
The sql `SELECT * FROM t WHERE a IN ()` throws a `ParseException` which is consistent with Hive, don't need to change that behavior.
## How was this patch tested?
Add new test case in `OptimizeInSuite`.
Author: jiangxingbo <jiangxb1987@gmail.com>
Closes#15977 from jiangxb1987/isin-empty.
## What changes were proposed in this pull request?
This is a follow-up PR of #15868 to merge `maxConnections` option into `numPartitions` options.
## How was this patch tested?
Pass the existing tests.
Author: Dongjoon Hyun <dongjoon@apache.org>
Closes#15966 from dongjoon-hyun/SPARK-18413-2.
## What changes were proposed in this pull request?
This PR only tries to fix things that looks pretty straightforward and were fixed in other previous PRs before.
This PR roughly fixes several things as below:
- Fix unrecognisable class and method links in javadoc by changing it from `[[..]]` to `` `...` ``
```
[error] .../spark/sql/core/target/java/org/apache/spark/sql/streaming/DataStreamReader.java:226: error: reference not found
[error] * Loads text files and returns a {link DataFrame} whose schema starts with a string column named
```
- Fix an exception annotation and remove code backticks in `throws` annotation
Currently, sbt unidoc with Java 8 complains as below:
```
[error] .../java/org/apache/spark/sql/streaming/StreamingQuery.java:72: error: unexpected text
[error] * throws StreamingQueryException, if <code>this</code> query has terminated with an exception.
```
`throws` should specify the correct class name from `StreamingQueryException,` to `StreamingQueryException` without backticks. (see [JDK-8007644](https://bugs.openjdk.java.net/browse/JDK-8007644)).
- Fix `[[http..]]` to `<a href="http..."></a>`.
```diff
- * [[https://blogs.oracle.com/java-platform-group/entry/diagnosing_tls_ssl_and_https Oracle
- * blog page]].
+ * <a href="https://blogs.oracle.com/java-platform-group/entry/diagnosing_tls_ssl_and_https">
+ * Oracle blog page</a>.
```
`[[http...]]` link markdown in scaladoc is unrecognisable in javadoc.
- It seems class can't have `return` annotation. So, two cases of this were removed.
```
[error] .../java/org/apache/spark/mllib/regression/IsotonicRegression.java:27: error: invalid use of return
[error] * return New instance of IsotonicRegression.
```
- Fix < to `<` and > to `>` according to HTML rules.
- Fix `</p>` complaint
- Exclude unrecognisable in javadoc, `constructor`, `todo` and `groupname`.
## How was this patch tested?
Manually tested by `jekyll build` with Java 7 and 8
```
java version "1.7.0_80"
Java(TM) SE Runtime Environment (build 1.7.0_80-b15)
Java HotSpot(TM) 64-Bit Server VM (build 24.80-b11, mixed mode)
```
```
java version "1.8.0_45"
Java(TM) SE Runtime Environment (build 1.8.0_45-b14)
Java HotSpot(TM) 64-Bit Server VM (build 25.45-b02, mixed mode)
```
Note: this does not yet make sbt unidoc suceed with Java 8 yet but it reduces the number of errors with Java 8.
Author: hyukjinkwon <gurwls223@gmail.com>
Closes#15999 from HyukjinKwon/SPARK-3359-errors.
## What changes were proposed in this pull request?
This PR addressed the rest comments in #15951.
## How was this patch tested?
Jenkins
Author: Shixiong Zhu <shixiong@databricks.com>
Closes#15997 from zsxwing/SPARK-18510-follow-up.
## What changes were proposed in this pull request?
### The Issue
If I specify my schema when doing
```scala
spark.read
.schema(someSchemaWherePartitionColumnsAreStrings)
```
but if the partition inference can infer it as IntegerType or I assume LongType or DoubleType (basically fixed size types), then once UnsafeRows are generated, your data will be corrupted.
### Proposed solution
The partition handling code path is kind of a mess. In my fix I'm probably adding to the mess, but at least trying to standardize the code path.
The real issue is that a user that uses the `spark.read` code path can never clearly specify what the partition columns are. If you try to specify the fields in `schema`, we practically ignore what the user provides, and fall back to our inferred data types. What happens in the end is data corruption.
My solution tries to fix this by always trying to infer partition columns the first time you specify the table. Once we find what the partition columns are, we try to find them in the user specified schema and use the dataType provided there, or fall back to the smallest common data type.
We will ALWAYS append partition columns to the user's schema, even if they didn't ask for it. We will only use the data type they provided if they specified it. While this is confusing, this has been the behavior since Spark 1.6, and I didn't want to change this behavior in the QA period of Spark 2.1. We may revisit this decision later.
A side effect of this PR is that we won't need https://github.com/apache/spark/pull/15942 if this PR goes in.
## How was this patch tested?
Regression tests
Author: Burak Yavuz <brkyvz@gmail.com>
Closes#15951 from brkyvz/partition-corruption.
## What changes were proposed in this pull request?
When we try to create the default database, we ask hive to do nothing if it already exists. However, Hive will log an error message instead of doing nothing, and the error message is quite annoying and confusing.
In this PR, we only create default database if it doesn't exist.
## How was this patch tested?
N/A
Author: Wenchen Fan <wenchen@databricks.com>
Closes#15993 from cloud-fan/default-db.
## What changes were proposed in this pull request?
The current implementation of column stats uses the base64 encoding of the internal UnsafeRow format to persist statistics (in table properties in Hive metastore). This is an internal format that is not stable across different versions of Spark and should NOT be used for persistence. In addition, it would be better if statistics stored in the catalog is human readable.
This pull request introduces the following changes:
1. Created a single ColumnStat class to for all data types. All data types track the same set of statistics.
2. Updated the implementation for stats collection to get rid of the dependency on internal data structures (e.g. InternalRow, or storing DateType as an int32). For example, previously dates were stored as a single integer, but are now stored as java.sql.Date. When we implement the next steps of CBO, we can add code to convert those back into internal types again.
3. Documented clearly what JVM data types are being used to store what data.
4. Defined a simple Map[String, String] interface for serializing and deserializing column stats into/from the catalog.
5. Rearranged the method/function structure so it is more clear what the supported data types are, and also moved how stats are generated into ColumnStat class so they are easy to find.
## How was this patch tested?
Removed most of the original test cases created for column statistics, and added three very simple ones to cover all the cases. The three test cases validate:
1. Roundtrip serialization works.
2. Behavior when analyzing non-existent column or unsupported data type column.
3. Result for stats collection for all valid data types.
Also moved parser related tests into a parser test suite and added an explicit serialization test for the Hive external catalog.
Author: Reynold Xin <rxin@databricks.com>
Closes#15959 from rxin/SPARK-18522.
## What changes were proposed in this pull request?
Updates links to the wiki to links to the new location of content on spark.apache.org.
## How was this patch tested?
Doc builds
Author: Sean Owen <sowen@cloudera.com>
Closes#15967 from srowen/SPARK-18073.1.
## What changes were proposed in this pull request?
Fixes the inconsistency of error raised between data source and hive serde
tables when schema is specified in CTAS scenario. In the process the grammar for
create table (datasource) is simplified.
**before:**
``` SQL
spark-sql> create table t2 (c1 int, c2 int) using parquet as select * from t1;
Error in query:
mismatched input 'as' expecting {<EOF>, '.', 'OPTIONS', 'CLUSTERED', 'PARTITIONED'}(line 1, pos 64)
== SQL ==
create table t2 (c1 int, c2 int) using parquet as select * from t1
----------------------------------------------------------------^^^
```
**After:**
```SQL
spark-sql> create table t2 (c1 int, c2 int) using parquet as select * from t1
> ;
Error in query:
Operation not allowed: Schema may not be specified in a Create Table As Select (CTAS) statement(line 1, pos 0)
== SQL ==
create table t2 (c1 int, c2 int) using parquet as select * from t1
^^^
```
## How was this patch tested?
Added a new test in CreateTableAsSelectSuite
Author: Dilip Biswal <dbiswal@us.ibm.com>
Closes#15968 from dilipbiswal/ctas.
### What changes were proposed in this pull request?
In Spark 2.0, `SaveAsTable` does not work when the target table is a Hive serde table, but Spark 1.6 works.
**Spark 1.6**
``` Scala
scala> sql("create table sample.sample stored as SEQUENCEFILE as select 1 as key, 'abc' as value")
res2: org.apache.spark.sql.DataFrame = []
scala> val df = sql("select key, value as value from sample.sample")
df: org.apache.spark.sql.DataFrame = [key: int, value: string]
scala> df.write.mode("append").saveAsTable("sample.sample")
scala> sql("select * from sample.sample").show()
+---+-----+
|key|value|
+---+-----+
| 1| abc|
| 1| abc|
+---+-----+
```
**Spark 2.0**
``` Scala
scala> df.write.mode("append").saveAsTable("sample.sample")
org.apache.spark.sql.AnalysisException: Saving data in MetastoreRelation sample, sample
is not supported.;
```
So far, we do not plan to support it in Spark 2.1 due to the risk. Spark 1.6 works because it internally uses insertInto. But, if we change it back it will break the semantic of saveAsTable (this method uses by-name resolution instead of using by-position resolution used by insertInto). More extra changes are needed to support `hive` as a `format` in DataFrameWriter.
Instead, users should use insertInto API. This PR corrects the error messages. Users can understand how to bypass it before we support it in a separate PR.
### How was this patch tested?
Test cases are added
Author: gatorsmile <gatorsmile@gmail.com>
Closes#15926 from gatorsmile/saveAsTableFix5.
## What changes were proposed in this pull request?
While this behavior is debatable, consider the following use case:
```sql
UNCACHE TABLE foo;
CACHE TABLE foo AS
SELECT * FROM bar
```
The command above fails the first time you run it. But I want to run the command above over and over again, and I don't want to change my code just for the first run of it.
The issue is that subsequent `CACHE TABLE` commands do not overwrite the existing table.
Now we can do:
```sql
UNCACHE TABLE IF EXISTS foo;
CACHE TABLE foo AS
SELECT * FROM bar
```
## How was this patch tested?
Unit tests
Author: Burak Yavuz <brkyvz@gmail.com>
Closes#15896 from brkyvz/uncache.
## What changes were proposed in this pull request?
This PR adds two of the newly added methods of `Dataset`s to Python:
`withWatermark` and `checkpoint`
## How was this patch tested?
Doc tests
Author: Burak Yavuz <brkyvz@gmail.com>
Closes#15921 from brkyvz/py-watermark.
## What changes were proposed in this pull request?
Currently, `DROP TABLE IF EXISTS` shows warning for non-existing tables. However, it had better be quiet for this case by definition of the command.
**BEFORE**
```scala
scala> sql("DROP TABLE IF EXISTS nonexist")
16/11/20 20:48:26 WARN DropTableCommand: org.apache.spark.sql.catalyst.analysis.NoSuchTableException: Table or view 'nonexist' not found in database 'default';
```
**AFTER**
```scala
scala> sql("DROP TABLE IF EXISTS nonexist")
res0: org.apache.spark.sql.DataFrame = []
```
## How was this patch tested?
Manual because this is related to the warning messages instead of exceptions.
Author: Dongjoon Hyun <dongjoon@apache.org>
Closes#15953 from dongjoon-hyun/SPARK-18517.
## What changes were proposed in this pull request?
This PR adds a new JDBCOption `maxConnections` which means the maximum number of simultaneous JDBC connections allowed. This option applies only to writing with coalesce operation if needed. It defaults to the number of partitions of RDD. Previously, SQL users cannot cannot control this while Scala/Java/Python users can use `coalesce` (or `repartition`) API.
**Reported Scenario**
For the following cases, the number of connections becomes 200 and database cannot handle all of them.
```sql
CREATE OR REPLACE TEMPORARY VIEW resultview
USING org.apache.spark.sql.jdbc
OPTIONS (
url "jdbc:oracle:thin:10.129.10.111:1521:BKDB",
dbtable "result",
user "HIVE",
password "HIVE"
);
-- set spark.sql.shuffle.partitions=200
INSERT OVERWRITE TABLE resultview SELECT g, count(1) AS COUNT FROM tnet.DT_LIVE_INFO GROUP BY g
```
## How was this patch tested?
Manual. Do the followings and see Spark UI.
**Step 1 (MySQL)**
```
CREATE TABLE t1 (a INT);
CREATE TABLE data (a INT);
INSERT INTO data VALUES (1);
INSERT INTO data VALUES (2);
INSERT INTO data VALUES (3);
```
**Step 2 (Spark)**
```scala
SPARK_HOME=$PWD bin/spark-shell --driver-memory 4G --driver-class-path mysql-connector-java-5.1.40-bin.jar
scala> sql("SET spark.sql.shuffle.partitions=3")
scala> sql("CREATE OR REPLACE TEMPORARY VIEW data USING org.apache.spark.sql.jdbc OPTIONS (url 'jdbc:mysql://localhost:3306/t', dbtable 'data', user 'root', password '')")
scala> sql("CREATE OR REPLACE TEMPORARY VIEW t1 USING org.apache.spark.sql.jdbc OPTIONS (url 'jdbc:mysql://localhost:3306/t', dbtable 't1', user 'root', password '', maxConnections '1')")
scala> sql("INSERT OVERWRITE TABLE t1 SELECT a FROM data GROUP BY a")
scala> sql("CREATE OR REPLACE TEMPORARY VIEW t1 USING org.apache.spark.sql.jdbc OPTIONS (url 'jdbc:mysql://localhost:3306/t', dbtable 't1', user 'root', password '', maxConnections '2')")
scala> sql("INSERT OVERWRITE TABLE t1 SELECT a FROM data GROUP BY a")
scala> sql("CREATE OR REPLACE TEMPORARY VIEW t1 USING org.apache.spark.sql.jdbc OPTIONS (url 'jdbc:mysql://localhost:3306/t', dbtable 't1', user 'root', password '', maxConnections '3')")
scala> sql("INSERT OVERWRITE TABLE t1 SELECT a FROM data GROUP BY a")
scala> sql("CREATE OR REPLACE TEMPORARY VIEW t1 USING org.apache.spark.sql.jdbc OPTIONS (url 'jdbc:mysql://localhost:3306/t', dbtable 't1', user 'root', password '', maxConnections '4')")
scala> sql("INSERT OVERWRITE TABLE t1 SELECT a FROM data GROUP BY a")
```
![maxconnections](https://cloud.githubusercontent.com/assets/9700541/20287987/ed8409c2-aa84-11e6-8aab-ae28e63fe54d.png)
Author: Dongjoon Hyun <dongjoon@apache.org>
Closes#15868 from dongjoon-hyun/SPARK-18413.
## What changes were proposed in this pull request?
This PR adds code generation to `Generate`. It supports two code paths:
- General `TraversableOnce` based iteration. This used for regular `Generator` (code generation supporting) expressions. This code path expects the expression to return a `TraversableOnce[InternalRow]` and it will iterate over the returned collection. This PR adds code generation for the `stack` generator.
- Specialized `ArrayData/MapData` based iteration. This is used for the `explode`, `posexplode` & `inline` functions and operates directly on the `ArrayData`/`MapData` result that the child of the generator returns.
### Benchmarks
I have added some benchmarks and it seems we can create a nice speedup for explode:
#### Environment
```
Java HotSpot(TM) 64-Bit Server VM 1.8.0_92-b14 on Mac OS X 10.11.6
Intel(R) Core(TM) i7-4980HQ CPU 2.80GHz
```
#### Explode Array
##### Before
```
generate explode array: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
generate explode array wholestage off 7377 / 7607 2.3 439.7 1.0X
generate explode array wholestage on 6055 / 6086 2.8 360.9 1.2X
```
##### After
```
generate explode array: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
generate explode array wholestage off 7432 / 7696 2.3 443.0 1.0X
generate explode array wholestage on 631 / 646 26.6 37.6 11.8X
```
#### Explode Map
##### Before
```
generate explode map: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
generate explode map wholestage off 12792 / 12848 1.3 762.5 1.0X
generate explode map wholestage on 11181 / 11237 1.5 666.5 1.1X
```
##### After
```
generate explode map: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
generate explode map wholestage off 10949 / 10972 1.5 652.6 1.0X
generate explode map wholestage on 870 / 913 19.3 51.9 12.6X
```
#### Posexplode
##### Before
```
generate posexplode array: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
generate posexplode array wholestage off 7547 / 7580 2.2 449.8 1.0X
generate posexplode array wholestage on 5786 / 5838 2.9 344.9 1.3X
```
##### After
```
generate posexplode array: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
generate posexplode array wholestage off 7535 / 7548 2.2 449.1 1.0X
generate posexplode array wholestage on 620 / 624 27.1 37.0 12.1X
```
#### Inline
##### Before
```
generate inline array: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
generate inline array wholestage off 6935 / 6978 2.4 413.3 1.0X
generate inline array wholestage on 6360 / 6400 2.6 379.1 1.1X
```
##### After
```
generate inline array: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
generate inline array wholestage off 6940 / 6966 2.4 413.6 1.0X
generate inline array wholestage on 1002 / 1012 16.7 59.7 6.9X
```
#### Stack
##### Before
```
generate stack: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
generate stack wholestage off 12980 / 13104 1.3 773.7 1.0X
generate stack wholestage on 11566 / 11580 1.5 689.4 1.1X
```
##### After
```
generate stack: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
generate stack wholestage off 12875 / 12949 1.3 767.4 1.0X
generate stack wholestage on 840 / 845 20.0 50.0 15.3X
```
## How was this patch tested?
Existing tests.
Author: Herman van Hovell <hvanhovell@databricks.com>
Author: Herman van Hovell <hvanhovell@questtec.nl>
Closes#13065 from hvanhovell/SPARK-15214.
## What changes were proposed in this pull request?
Fix since 2.1.0 on new SparkSession.close() method. I goofed in https://github.com/apache/spark/pull/15932 because it was back-ported to 2.1 instead of just master as originally planned.
Author: Sean Owen <sowen@cloudera.com>
Closes#15938 from srowen/SPARK-18448.2.
## What changes were proposed in this pull request?
Just adds `close()` + `Closeable` as a synonym for `stop()`. This makes it usable in Java in try-with-resources, as suggested by ash211 (`Closeable` extends `AutoCloseable` BTW)
## How was this patch tested?
Existing tests
Author: Sean Owen <sowen@cloudera.com>
Closes#15932 from srowen/SPARK-18448.
## What changes were proposed in this pull request?
The issue in ForeachSink is the new created DataSet still uses the old QueryExecution. When `foreachPartition` is called, `QueryExecution.toString` will be called and then fail because it doesn't know how to plan EventTimeWatermark.
This PR just replaces the QueryExecution with IncrementalExecution to fix the issue.
## How was this patch tested?
`test("foreach with watermark")`.
Author: Shixiong Zhu <shixiong@databricks.com>
Closes#15934 from zsxwing/SPARK-18497.
## What changes were proposed in this pull request?
I'm spending more time at the design & code level for cost-based optimizer now, and have found a number of issues related to maintainability and compatibility that I will like to address.
This is a small pull request to clean up AnalyzeColumnCommand:
1. Removed warning on duplicated columns. Warnings in log messages are useless since most users that run SQL don't see them.
2. Removed the nested updateStats function, by just inlining the function.
3. Renamed a few functions to better reflect what they do.
4. Removed the factory apply method for ColumnStatStruct. It is a bad pattern to use a apply method that returns an instantiation of a class that is not of the same type (ColumnStatStruct.apply used to return CreateNamedStruct).
5. Renamed ColumnStatStruct to just AnalyzeColumnCommand.
6. Added more documentation explaining some of the non-obvious return types and code blocks.
In follow-up pull requests, I'd like to address the following:
1. Get rid of the Map[String, ColumnStat] map, since internally we should be using Attribute to reference columns, rather than strings.
2. Decouple the fields exposed by ColumnStat and internals of Spark SQL's execution path. Currently the two are coupled because ColumnStat takes in an InternalRow.
3. Correctness: Remove code path that stores statistics in the catalog using the base64 encoding of the UnsafeRow format, which is not stable across Spark versions.
4. Clearly document the data representation stored in the catalog for statistics.
## How was this patch tested?
Affected test cases have been updated.
Author: Reynold Xin <rxin@databricks.com>
Closes#15933 from rxin/SPARK-18505.
## What changes were proposed in this pull request?
HDFS `write` may just hang until timeout if some network error happens. It's better to enable interrupts to allow stopping the query fast on HDFS.
This PR just changes the logic to only disable interrupts for local file system, as HADOOP-10622 only happens for local file system.
## How was this patch tested?
Jenkins
Author: Shixiong Zhu <shixiong@databricks.com>
Closes#15911 from zsxwing/interrupt-on-dfs.
## What changes were proposed in this pull request?
CompactibleFileStreamLog relys on "compactInterval" to detect a compaction batch. If the "compactInterval" is reset by user, CompactibleFileStreamLog will return wrong answer, resulting data loss. This PR procides a way to check the validity of 'compactInterval', and calculate an appropriate value.
## How was this patch tested?
When restart a stream, we change the 'spark.sql.streaming.fileSource.log.compactInterval' different with the former one.
The primary solution to this issue was given by uncleGen
Added extensions include an additional metadata field in OffsetSeq and CompactibleFileStreamLog APIs. zsxwing
Author: Tyson Condie <tcondie@gmail.com>
Author: genmao.ygm <genmao.ygm@genmaoygmdeMacBook-Air.local>
Closes#15852 from tcondie/spark-18187.
## What changes were proposed in this pull request?
This patch fixes a `ClassCastException: java.lang.Integer cannot be cast to java.lang.Long` error which could occur in the HistoryServer while trying to process a deserialized `SparkListenerDriverAccumUpdates` event.
The problem stems from how `jackson-module-scala` handles primitive type parameters (see https://github.com/FasterXML/jackson-module-scala/wiki/FAQ#deserializing-optionint-and-other-primitive-challenges for more details). This was causing a problem where our code expected a field to be deserialized as a `(Long, Long)` tuple but we got an `(Int, Int)` tuple instead.
This patch hacks around this issue by registering a custom `Converter` with Jackson in order to deserialize the tuples as `(Object, Object)` and perform the appropriate casting.
## How was this patch tested?
New regression tests in `SQLListenerSuite`.
Author: Josh Rosen <joshrosen@databricks.com>
Closes#15922 from JoshRosen/SPARK-18462.
## What changes were proposed in this pull request?
In ShuffleExchange, the nodename's extraInfo are the same when exchangeCoordinator.isEstimated
is true or false.
Merge the two situation in the PR.
Author: root <root@iZbp1gsnrlfzjxh82cz80vZ.(none)>
Closes#15920 from windpiger/DupNodeNameShuffleExchange.
## What changes were proposed in this pull request?
I found the documentation for the sample method to be confusing, this adds more clarification across all languages.
- [x] Scala
- [x] Python
- [x] R
- [x] RDD Scala
- [ ] RDD Python with SEED
- [X] RDD Java
- [x] RDD Java with SEED
- [x] RDD Python
## How was this patch tested?
NA
Please review https://cwiki.apache.org/confluence/display/SPARK/Contributing+to+Spark before opening a pull request.
Author: anabranch <wac.chambers@gmail.com>
Author: Bill Chambers <bill@databricks.com>
Closes#15815 from anabranch/SPARK-18365.
## What changes were proposed in this pull request?
Before Spark 2.1, users can create an external data source table without schema, and we will infer the table schema at runtime. In Spark 2.1, we decided to infer the schema when the table was created, so that we don't need to infer it again and again at runtime.
This is a good improvement, but we should still respect and support old tables which doesn't store table schema in metastore.
## How was this patch tested?
regression test.
Author: Wenchen Fan <wenchen@databricks.com>
Closes#15900 from cloud-fan/hive-catalog.
## What changes were proposed in this pull request?
SPARK-18459: triggerId seems like a number that should be increasing with each trigger, whether or not there is data in it. However, actually, triggerId increases only where there is a batch of data in a trigger. So its better to rename it to batchId.
SPARK-18460: triggerDetails was missing from json representation. Fixed it.
## How was this patch tested?
Updated existing unit tests.
Author: Tathagata Das <tathagata.das1565@gmail.com>
Closes#15895 from tdas/SPARK-18459.
### What changes were proposed in this pull request?
Currently, when CTE is used in RunnableCommand, the Analyzer does not replace the logical node `With`. The child plan of RunnableCommand is not resolved. Thus, the output of the `With` plan node looks very confusing.
For example,
```
sql(
"""
|CREATE VIEW cte_view AS
|WITH w AS (SELECT 1 AS n), cte1 (select 2), cte2 as (select 3)
|SELECT n FROM w
""".stripMargin).explain()
```
The output is like
```
ExecutedCommand
+- CreateViewCommand `cte_view`, WITH w AS (SELECT 1 AS n), cte1 (select 2), cte2 as (select 3)
SELECT n FROM w, false, false, PersistedView
+- 'With [(w,SubqueryAlias w
+- Project [1 AS n#16]
+- OneRowRelation$
), (cte1,'SubqueryAlias cte1
+- 'Project [unresolvedalias(2, None)]
+- OneRowRelation$
), (cte2,'SubqueryAlias cte2
+- 'Project [unresolvedalias(3, None)]
+- OneRowRelation$
)]
+- 'Project ['n]
+- 'UnresolvedRelation `w`
```
After the fix, the output is as shown below.
```
ExecutedCommand
+- CreateViewCommand `cte_view`, WITH w AS (SELECT 1 AS n), cte1 (select 2), cte2 as (select 3)
SELECT n FROM w, false, false, PersistedView
+- CTE [w, cte1, cte2]
: :- SubqueryAlias w
: : +- Project [1 AS n#16]
: : +- OneRowRelation$
: :- 'SubqueryAlias cte1
: : +- 'Project [unresolvedalias(2, None)]
: : +- OneRowRelation$
: +- 'SubqueryAlias cte2
: +- 'Project [unresolvedalias(3, None)]
: +- OneRowRelation$
+- 'Project ['n]
+- 'UnresolvedRelation `w`
```
BTW, this PR also fixes the output of the view type.
### How was this patch tested?
Manual
Author: gatorsmile <gatorsmile@gmail.com>
Closes#15854 from gatorsmile/cteName.
## What changes were proposed in this pull request?
This PR aims to improve DataSource option keys to be more case-insensitive
DataSource partially use CaseInsensitiveMap in code-path. For example, the following fails to find url.
```scala
val df = spark.createDataFrame(sparkContext.parallelize(arr2x2), schema2)
df.write.format("jdbc")
.option("UrL", url1)
.option("dbtable", "TEST.SAVETEST")
.options(properties.asScala)
.save()
```
This PR makes DataSource options to use CaseInsensitiveMap internally and also makes DataSource to use CaseInsensitiveMap generally except `InMemoryFileIndex` and `InsertIntoHadoopFsRelationCommand`. We can not pass them CaseInsensitiveMap because they creates new case-sensitive HadoopConfs by calling newHadoopConfWithOptions(options) inside.
## How was this patch tested?
Pass the Jenkins test with newly added test cases.
Author: Dongjoon Hyun <dongjoon@apache.org>
Closes#15884 from dongjoon-hyun/SPARK-18433.
## What changes were proposed in this pull request?
it's weird that every session can set its own warehouse path at runtime, we should forbid it and make it a static conf.
## How was this patch tested?
existing tests.
Author: Wenchen Fan <wenchen@databricks.com>
Closes#15825 from cloud-fan/warehouse.
## What changes were proposed in this pull request?
SPARK-18012 refactored the file write path in FileStreamSink using FileFormatWriter which always uses the default non-streaming QueryExecution to perform the writes. This is wrong for FileStreamSink, because the streaming QueryExecution (i.e. IncrementalExecution) should be used for correctly incrementalizing aggregation. The addition of watermarks in SPARK-18124, file stream sink should logically supports aggregation + watermark + append mode. But actually it fails with
```
16:23:07.389 ERROR org.apache.spark.sql.execution.streaming.StreamExecution: Query query-0 terminated with error
java.lang.AssertionError: assertion failed: No plan for EventTimeWatermark timestamp#7: timestamp, interval 10 seconds
+- LocalRelation [timestamp#7]
at scala.Predef$.assert(Predef.scala:170)
at org.apache.spark.sql.catalyst.planning.QueryPlanner.plan(QueryPlanner.scala:92)
at org.apache.spark.sql.catalyst.planning.QueryPlanner$$anonfun$2$$anonfun$apply$2.apply(QueryPlanner.scala:77)
at org.apache.spark.sql.catalyst.planning.QueryPlanner$$anonfun$2$$anonfun$apply$2.apply(QueryPlanner.scala:74)
at scala.collection.TraversableOnce$$anonfun$foldLeft$1.apply(TraversableOnce.scala:157)
at scala.collection.TraversableOnce$$anonfun$foldLeft$1.apply(TraversableOnce.scala:157)
at scala.collection.Iterator$class.foreach(Iterator.scala:893)
at scala.collection.AbstractIterator.foreach(Iterator.scala:1336)
at scala.collection.TraversableOnce$class.foldLeft(TraversableOnce.scala:157)
at scala.collection.AbstractIterator.foldLeft(Iterator.scala:1336)
at org.apache.spark.sql.catalyst.planning.QueryPlanner$$anonfun$2.apply(QueryPlanner.scala:74)
at org.apache.spark.sql.catalyst.planning.QueryPlanner$$anonfun$2.apply(QueryPlanner.scala:66)
at scala.collection.Iterator$$anon$12.nextCur(Iterator.scala:434)
at scala.collection.Iterator$$anon$12.hasNext(Iterator.scala:440)
at org.apache.spark.sql.catalyst.planning.QueryPlanner.plan(QueryPlanner.scala:92)
at org.apache.spark.sql.catalyst.planning.QueryPlanner$$anonfun$2$$anonfun$apply$2.apply(QueryPlanner.scala:77)
at org.apache.spark.sql.catalyst.planning.QueryPlanner$$anonfun$2$$anonfun$apply$2.apply(QueryPlanner.scala:74)
```
This PR fixes it by passing the correct query execution.
## How was this patch tested?
New unit test
Author: Tathagata Das <tathagata.das1565@gmail.com>
Closes#15885 from tdas/SPARK-18440.
## What changes were proposed in this pull request?
It would be nice if memory sinks can also recover from checkpoints. For correctness reasons, the only time we should support it is in `Complete` OutputMode. We can support this in CompleteMode, because the output of the StateStore is already persisted in the checkpoint directory.
## How was this patch tested?
Unit test
Author: Burak Yavuz <brkyvz@gmail.com>
Closes#15801 from brkyvz/mem-stream.
## What changes were proposed in this pull request?
The largest parallelism in PartitioningAwareFileIndex #listLeafFilesInParallel() is 10000 in hard code. We may need to make this number configurable. And in PR, I reduce it to 100.
## How was this patch tested?
Existing ut.
Author: genmao.ygm <genmao.ygm@genmaoygmdeMacBook-Air.local>
Author: dylon <hustyugm@gmail.com>
Closes#15829 from uncleGen/SPARK-18379.
This PR adds a new method `withWatermark` to the `Dataset` API, which can be used specify an _event time watermark_. An event time watermark allows the streaming engine to reason about the point in time after which we no longer expect to see late data. This PR also has augmented `StreamExecution` to use this watermark for several purposes:
- To know when a given time window aggregation is finalized and thus results can be emitted when using output modes that do not allow updates (e.g. `Append` mode).
- To minimize the amount of state that we need to keep for on-going aggregations, by evicting state for groups that are no longer expected to change. Although, we do still maintain all state if the query requires (i.e. if the event time is not present in the `groupBy` or when running in `Complete` mode).
An example that emits windowed counts of records, waiting up to 5 minutes for late data to arrive.
```scala
df.withWatermark("eventTime", "5 minutes")
.groupBy(window($"eventTime", "1 minute") as 'window)
.count()
.writeStream
.format("console")
.mode("append") // In append mode, we only output finalized aggregations.
.start()
```
### Calculating the watermark.
The current event time is computed by looking at the `MAX(eventTime)` seen this epoch across all of the partitions in the query minus some user defined _delayThreshold_. An additional constraint is that the watermark must increase monotonically.
Note that since we must coordinate this value across partitions occasionally, the actual watermark used is only guaranteed to be at least `delay` behind the actual event time. In some cases we may still process records that arrive more than delay late.
This mechanism was chosen for the initial implementation over processing time for two reasons:
- it is robust to downtime that could affect processing delay
- it does not require syncing of time or timezones between the producer and the processing engine.
### Other notable implementation details
- A new trigger metric `eventTimeWatermark` outputs the current value of the watermark.
- We mark the event time column in the `Attribute` metadata using the key `spark.watermarkDelay`. This allows downstream operations to know which column holds the event time. Operations like `window` propagate this metadata.
- `explain()` marks the watermark with a suffix of `-T${delayMs}` to ease debugging of how this information is propagated.
- Currently, we don't filter out late records, but instead rely on the state store to avoid emitting records that are both added and filtered in the same epoch.
### Remaining in this PR
- [ ] The test for recovery is currently failing as we don't record the watermark used in the offset log. We will need to do so to ensure determinism, but this is deferred until #15626 is merged.
### Other follow-ups
There are some natural additional features that we should consider for future work:
- Ability to write records that arrive too late to some external store in case any out-of-band remediation is required.
- `Update` mode so you can get partial results before a group is evicted.
- Other mechanisms for calculating the watermark. In particular a watermark based on quantiles would be more robust to outliers.
Author: Michael Armbrust <michael@databricks.com>
Closes#15702 from marmbrus/watermarks.
## What changes were proposed in this pull request?
StateStore.get() causes temporary files to be created immediately, even if the store is not used to make updates for new version. The temp file is not closed as store.commit() is not called in those cases, thus keeping the output stream to temp file open forever.
This PR fixes it by opening the temp file only when there are updates being made.
## How was this patch tested?
New unit test
Author: Tathagata Das <tathagata.das1565@gmail.com>
Closes#15859 from tdas/SPARK-18416.
## What changes were proposed in this pull request?
Currently, `SQLBuilder` handles `LIMIT` by always adding `LIMIT` at the end of the generated subSQL. It makes `RuntimeException`s like the following. This PR adds a parenthesis always except `SubqueryAlias` is used together with `LIMIT`.
**Before**
``` scala
scala> sql("CREATE TABLE tbl(id INT)")
scala> sql("CREATE VIEW v1(id2) AS SELECT id FROM tbl LIMIT 2")
java.lang.RuntimeException: Failed to analyze the canonicalized SQL: ...
```
**After**
``` scala
scala> sql("CREATE TABLE tbl(id INT)")
scala> sql("CREATE VIEW v1(id2) AS SELECT id FROM tbl LIMIT 2")
scala> sql("SELECT id2 FROM v1")
res4: org.apache.spark.sql.DataFrame = [id2: int]
```
**Fixed cases in this PR**
The following two cases are the detail query plans having problematic SQL generations.
1. `SELECT * FROM (SELECT id FROM tbl LIMIT 2)`
Please note that **FROM SELECT** part of the generated SQL in the below. When we don't use '()' for limit, this fails.
```scala
# Original logical plan:
Project [id#1]
+- GlobalLimit 2
+- LocalLimit 2
+- Project [id#1]
+- MetastoreRelation default, tbl
# Canonicalized logical plan:
Project [gen_attr_0#1 AS id#4]
+- SubqueryAlias tbl
+- Project [gen_attr_0#1]
+- GlobalLimit 2
+- LocalLimit 2
+- Project [gen_attr_0#1]
+- SubqueryAlias gen_subquery_0
+- Project [id#1 AS gen_attr_0#1]
+- SQLTable default, tbl, [id#1]
# Generated SQL:
SELECT `gen_attr_0` AS `id` FROM (SELECT `gen_attr_0` FROM SELECT `gen_attr_0` FROM (SELECT `id` AS `gen_attr_0` FROM `default`.`tbl`) AS gen_subquery_0 LIMIT 2) AS tbl
```
2. `SELECT * FROM (SELECT id FROM tbl TABLESAMPLE (2 ROWS))`
Please note that **((~~~) AS gen_subquery_0 LIMIT 2)** in the below. When we use '()' for limit on `SubqueryAlias`, this fails.
```scala
# Original logical plan:
Project [id#1]
+- Project [id#1]
+- GlobalLimit 2
+- LocalLimit 2
+- MetastoreRelation default, tbl
# Canonicalized logical plan:
Project [gen_attr_0#1 AS id#4]
+- SubqueryAlias tbl
+- Project [gen_attr_0#1]
+- GlobalLimit 2
+- LocalLimit 2
+- SubqueryAlias gen_subquery_0
+- Project [id#1 AS gen_attr_0#1]
+- SQLTable default, tbl, [id#1]
# Generated SQL:
SELECT `gen_attr_0` AS `id` FROM (SELECT `gen_attr_0` FROM ((SELECT `id` AS `gen_attr_0` FROM `default`.`tbl`) AS gen_subquery_0 LIMIT 2)) AS tbl
```
## How was this patch tested?
Pass the Jenkins test with a newly added test case.
Author: Dongjoon Hyun <dongjoon@apache.org>
Closes#15546 from dongjoon-hyun/SPARK-17982.
## What changes were proposed in this pull request?
As of current 2.1, INSERT OVERWRITE with dynamic partitions against a Datasource table will overwrite the entire table instead of only the partitions matching the static keys, as in Hive. It also doesn't respect custom partition locations.
This PR adds support for all these operations to Datasource tables managed by the Hive metastore. It is implemented as follows
- During planning time, the full set of partitions affected by an INSERT or OVERWRITE command is read from the Hive metastore.
- The planner identifies any partitions with custom locations and includes this in the write task metadata.
- FileFormatWriter tasks refer to this custom locations map when determining where to write for dynamic partition output.
- When the write job finishes, the set of written partitions is compared against the initial set of matched partitions, and the Hive metastore is updated to reflect the newly added / removed partitions.
It was necessary to introduce a method for staging files with absolute output paths to `FileCommitProtocol`. These files are not handled by the Hadoop output committer but are moved to their final locations when the job commits.
The overwrite behavior of legacy Datasource tables is also changed: no longer will the entire table be overwritten if a partial partition spec is present.
cc cloud-fan yhuai
## How was this patch tested?
Unit tests, existing tests.
Author: Eric Liang <ekl@databricks.com>
Author: Wenchen Fan <wenchen@databricks.com>
Closes#15814 from ericl/sc-5027.
## What changes were proposed in this pull request?
This PR corrects several partition related behaviors of `ExternalCatalog`:
1. default partition location should not always lower case the partition column names in path string(fix `HiveExternalCatalog`)
2. rename partition should not always lower case the partition column names in updated partition path string(fix `HiveExternalCatalog`)
3. rename partition should update the partition location only for managed table(fix `InMemoryCatalog`)
4. create partition with existing directory should be fine(fix `InMemoryCatalog`)
5. create partition with non-existing directory should create that directory(fix `InMemoryCatalog`)
6. drop partition from external table should not delete the directory(fix `InMemoryCatalog`)
## How was this patch tested?
new tests in `ExternalCatalogSuite`
Author: Wenchen Fan <wenchen@databricks.com>
Closes#15797 from cloud-fan/partition.
(Link to Jira issue: https://issues.apache.org/jira/browse/SPARK-17993)
## What changes were proposed in this pull request?
PR #14690 broke parquet log output redirection for converted partitioned Hive tables. For example, when querying parquet files written by Parquet-mr 1.6.0 Spark prints a torrent of (harmless) warning messages from the Parquet reader:
```
Oct 18, 2016 7:42:18 PM WARNING: org.apache.parquet.CorruptStatistics: Ignoring statistics because created_by could not be parsed (see PARQUET-251): parquet-mr version 1.6.0
org.apache.parquet.VersionParser$VersionParseException: Could not parse created_by: parquet-mr version 1.6.0 using format: (.+) version ((.*) )?\(build ?(.*)\)
at org.apache.parquet.VersionParser.parse(VersionParser.java:112)
at org.apache.parquet.CorruptStatistics.shouldIgnoreStatistics(CorruptStatistics.java:60)
at org.apache.parquet.format.converter.ParquetMetadataConverter.fromParquetStatistics(ParquetMetadataConverter.java:263)
at org.apache.parquet.hadoop.ParquetFileReader$Chunk.readAllPages(ParquetFileReader.java:583)
at org.apache.parquet.hadoop.ParquetFileReader.readNextRowGroup(ParquetFileReader.java:513)
at org.apache.spark.sql.execution.datasources.parquet.VectorizedParquetRecordReader.checkEndOfRowGroup(VectorizedParquetRecordReader.java:270)
at org.apache.spark.sql.execution.datasources.parquet.VectorizedParquetRecordReader.nextBatch(VectorizedParquetRecordReader.java:225)
at org.apache.spark.sql.execution.datasources.parquet.VectorizedParquetRecordReader.nextKeyValue(VectorizedParquetRecordReader.java:137)
at org.apache.spark.sql.execution.datasources.RecordReaderIterator.hasNext(RecordReaderIterator.scala:39)
at org.apache.spark.sql.execution.datasources.FileScanRDD$$anon$1.hasNext(FileScanRDD.scala:102)
at org.apache.spark.sql.execution.datasources.FileScanRDD$$anon$1.nextIterator(FileScanRDD.scala:162)
at org.apache.spark.sql.execution.datasources.FileScanRDD$$anon$1.hasNext(FileScanRDD.scala:102)
at org.apache.spark.sql.catalyst.expressions.GeneratedClass$GeneratedIterator.scan_nextBatch$(Unknown Source)
at org.apache.spark.sql.catalyst.expressions.GeneratedClass$GeneratedIterator.processNext(Unknown Source)
at org.apache.spark.sql.execution.BufferedRowIterator.hasNext(BufferedRowIterator.java:43)
at org.apache.spark.sql.execution.WholeStageCodegenExec$$anonfun$8$$anon$1.hasNext(WholeStageCodegenExec.scala:372)
at org.apache.spark.sql.execution.SparkPlan$$anonfun$2.apply(SparkPlan.scala:231)
at org.apache.spark.sql.execution.SparkPlan$$anonfun$2.apply(SparkPlan.scala:225)
at org.apache.spark.rdd.RDD$$anonfun$mapPartitionsInternal$1$$anonfun$apply$24.apply(RDD.scala:803)
at org.apache.spark.rdd.RDD$$anonfun$mapPartitionsInternal$1$$anonfun$apply$24.apply(RDD.scala:803)
at org.apache.spark.rdd.MapPartitionsRDD.compute(MapPartitionsRDD.scala:38)
at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:319)
at org.apache.spark.rdd.RDD.iterator(RDD.scala:283)
at org.apache.spark.scheduler.ResultTask.runTask(ResultTask.scala:87)
at org.apache.spark.scheduler.Task.run(Task.scala:99)
at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:282)
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:745)
```
This only happens during execution, not planning, and it doesn't matter what log level the `SparkContext` is set to. That's because Parquet (versions < 1.9) doesn't use slf4j for logging. Note, you can tell that log redirection is not working here because the log message format does not conform to the default Spark log message format.
This is a regression I noted as something we needed to fix as a follow up.
It appears that the problem arose because we removed the call to `inferSchema` during Hive table conversion. That call is what triggered the output redirection.
## How was this patch tested?
I tested this manually in four ways:
1. Executing `spark.sqlContext.range(10).selectExpr("id as a").write.mode("overwrite").parquet("test")`.
2. Executing `spark.read.format("parquet").load(legacyParquetFile).show` for a Parquet file `legacyParquetFile` written using Parquet-mr 1.6.0.
3. Executing `select * from legacy_parquet_table limit 1` for some unpartitioned Parquet-based Hive table written using Parquet-mr 1.6.0.
4. Executing `select * from legacy_partitioned_parquet_table where partcol=x limit 1` for some partitioned Parquet-based Hive table written using Parquet-mr 1.6.0.
I ran each test with a new instance of `spark-shell` or `spark-sql`.
Incidentally, I found that test case 3 was not a regression—redirection was not occurring in the master codebase prior to #14690.
I spent some time working on a unit test, but based on my experience working on this ticket I feel that automated testing here is far from feasible.
cc ericl dongjoon-hyun
Author: Michael Allman <michael@videoamp.com>
Closes#15538 from mallman/spark-17993-fix_parquet_log_redirection.
## What changes were proposed in this pull request?
Currently we use java serialization for the WAL that stores the offsets contained in each batch. This has two main issues:
It can break across spark releases (though this is not the only thing preventing us from upgrading a running query)
It is unnecessarily opaque to the user.
I'd propose we require offsets to provide a user readable serialization and use that instead. JSON is probably a good option.
## How was this patch tested?
Tests were added for KafkaSourceOffset in [KafkaSourceOffsetSuite](external/kafka-0-10-sql/src/test/scala/org/apache/spark/sql/kafka010/KafkaSourceOffsetSuite.scala) and for LongOffset in [OffsetSuite](sql/core/src/test/scala/org/apache/spark/sql/streaming/OffsetSuite.scala)
Please review https://cwiki.apache.org/confluence/display/SPARK/Contributing+to+Spark before opening a pull request.
zsxwing marmbrus
Author: Tyson Condie <tcondie@gmail.com>
Author: Tyson Condie <tcondie@clash.local>
Closes#15626 from tcondie/spark-8360.
## What changes were proposed in this pull request?
`InsertIntoHadoopFsRelationCommand` does not keep track if it inserts into a table and what table it inserts to. This can make debugging these statements problematic. This PR adds table information the `InsertIntoHadoopFsRelationCommand`. Explaining this SQL command `insert into prq select * from range(0, 100000)` now yields the following executed plan:
```
== Physical Plan ==
ExecutedCommand
+- InsertIntoHadoopFsRelationCommand file:/dev/assembly/spark-warehouse/prq, ParquetFormat, <function1>, Map(serialization.format -> 1, path -> file:/dev/assembly/spark-warehouse/prq), Append, CatalogTable(
Table: `default`.`prq`
Owner: hvanhovell
Created: Wed Nov 09 17:42:30 CET 2016
Last Access: Thu Jan 01 01:00:00 CET 1970
Type: MANAGED
Schema: [StructField(id,LongType,true)]
Provider: parquet
Properties: [transient_lastDdlTime=1478709750]
Storage(Location: file:/dev/assembly/spark-warehouse/prq, InputFormat: org.apache.hadoop.hive.ql.io.parquet.MapredParquetInputFormat, OutputFormat: org.apache.hadoop.hive.ql.io.parquet.MapredParquetOutputFormat, Serde: org.apache.hadoop.hive.ql.io.parquet.serde.ParquetHiveSerDe, Properties: [serialization.format=1]))
+- Project [id#7L]
+- Range (0, 100000, step=1, splits=None)
```
## How was this patch tested?
Added extra checks to the `ParquetMetastoreSuite`
Author: Herman van Hovell <hvanhovell@databricks.com>
Closes#15832 from hvanhovell/SPARK-18370.
### What changes were proposed in this pull request?
`Partitioned View` is not supported by SPARK SQL. For Hive partitioned view, SHOW CREATE TABLE is unable to generate the right DDL. Thus, SHOW CREATE TABLE should not support it like the other Hive-only features. This PR is to issue an exception when detecting the view is a partitioned view.
### How was this patch tested?
Added a test case
Author: gatorsmile <gatorsmile@gmail.com>
Closes#15233 from gatorsmile/partitionedView.
## What changes were proposed in this pull request?
These are no longer needed after https://issues.apache.org/jira/browse/SPARK-17183
cc cloud-fan
## How was this patch tested?
Existing parquet and orc tests.
Author: Eric Liang <ekl@databricks.com>
Closes#15799 from ericl/sc-4929.
## What changes were proposed in this pull request?
If the rename operation in the state store fails (`fs.rename` returns `false`), the StateStore should throw an exception and have the task retry. Currently if renames fail, nothing happens during execution immediately. However, you will observe that snapshot operations will fail, and then any attempt at recovery (executor failure / checkpoint recovery) also fails.
## How was this patch tested?
Unit test
Author: Burak Yavuz <brkyvz@gmail.com>
Closes#15804 from brkyvz/rename-state.
## What changes were proposed in this pull request?
This PR port RDD API to use commit protocol, the changes made here:
1. Add new internal helper class that saves an RDD using a Hadoop OutputFormat named `SparkNewHadoopWriter`, it's similar with `SparkHadoopWriter` but uses commit protocol. This class supports the newer `mapreduce` API, instead of the old `mapred` API which is supported by `SparkHadoopWriter`;
2. Rewrite `PairRDDFunctions.saveAsNewAPIHadoopDataset` function, so it uses commit protocol now.
## How was this patch tested?
Exsiting test cases.
Author: jiangxingbo <jiangxb1987@gmail.com>
Closes#15769 from jiangxb1987/rdd-commit.
## What changes were proposed in this pull request?
a follow up of https://github.com/apache/spark/pull/15688
## How was this patch tested?
updated test in `DDLSuite`
Author: Wenchen Fan <wenchen@databricks.com>
Closes#15805 from cloud-fan/truncate.
### What changes were proposed in this pull request?
Based on the discussion in [SPARK-18209](https://issues.apache.org/jira/browse/SPARK-18209). It doesn't really make sense to create permanent views based on temporary views or temporary UDFs.
To disallow the supports and issue the exceptions, this PR needs to detect whether a temporary view/UDF is being used when defining a permanent view. Basically, this PR can be split to two sub-tasks:
**Task 1:** detecting a temporary view from the query plan of view definition.
When finding an unresolved temporary view, Analyzer replaces it by a `SubqueryAlias` with the corresponding logical plan, which is stored in an in-memory HashMap. After replacement, it is impossible to detect whether the `SubqueryAlias` is added/generated from a temporary view. Thus, to detect the usage of a temporary view in view definition, this PR traverses the unresolved logical plan and uses the name of an `UnresolvedRelation` to detect whether it is a (global) temporary view.
**Task 2:** detecting a temporary UDF from the query plan of view definition.
Detecting usage of a temporary UDF in view definition is not straightfoward.
First, in the analyzed plan, we are having different forms to represent the functions. More importantly, some classes (e.g., `HiveGenericUDF`) are not accessible from `CreateViewCommand`, which is part of `sql/core`. Thus, we used the unanalyzed plan `child` of `CreateViewCommand` to detect the usage of a temporary UDF. Because the plan has already been successfully analyzed, we can assume the functions have been defined/registered.
Second, in Spark, the functions have four forms: Spark built-in functions, built-in hash functions, permanent UDFs and temporary UDFs. We do not have any direct way to determine whether a function is temporary or not. Thus, we introduced a function `isTemporaryFunction` in `SessionCatalog`. This function contains the detailed logics to determine whether a function is temporary or not.
### How was this patch tested?
Added test cases.
Author: gatorsmile <gatorsmile@gmail.com>
Closes#15764 from gatorsmile/blockTempFromPermViewCreation.
## What changes were proposed in this pull request?
Right now, there is no way to join the output of a memory sink with any table:
> UnsupportedOperationException: LeafNode MemoryPlan must implement statistics
This patch adds statistics to MemorySink, making joining snapshots of memory streams with tables possible.
## How was this patch tested?
Added a test case.
Author: Liwei Lin <lwlin7@gmail.com>
Closes#15786 from lw-lin/memory-sink-stat.
## What changes were proposed in this pull request?
This adds support for Hive variables:
* Makes values set via `spark-sql --hivevar name=value` accessible
* Adds `getHiveVar` and `setHiveVar` to the `HiveClient` interface
* Adds a SessionVariables trait for sessions like Hive that support variables (including Hive vars)
* Adds SessionVariables support to variable substitution
* Adds SessionVariables support to the SET command
## How was this patch tested?
* Adds a test to all supported Hive versions for accessing Hive variables
* Adds HiveVariableSubstitutionSuite
Author: Ryan Blue <blue@apache.org>
Closes#15738 from rdblue/SPARK-18086-add-hivevar-support.
## What changes were proposed in this pull request?
Previously `TRUNCATE TABLE ... PARTITION` will always truncate the whole table for data source tables, this PR fixes it and improve `InMemoryCatalog` to make this command work with it.
## How was this patch tested?
existing tests
Author: Wenchen Fan <wenchen@databricks.com>
Closes#15688 from cloud-fan/truncate.
## What changes were proposed in this pull request?
Currently, there are the three cases when reading CSV by datasource when it is `PERMISSIVE` parse mode.
- schema == parsed tokens (from each line)
No problem to cast the value in the tokens to the field in the schema as they are equal.
- schema < parsed tokens (from each line)
It slices the tokens into the number of fields in schema.
- schema > parsed tokens (from each line)
It appends `null` into parsed tokens so that safely values can be casted with the schema.
However, when `null` is appended in the third case, we should take `null` into account when casting the values.
In case of `StringType`, it is fine as `UTF8String.fromString(datum)` produces `null` when the input is `null`. Therefore, this case will happen only when schema is explicitly given and schema includes data types that are not `StringType`.
The codes below:
```scala
val path = "/tmp/a"
Seq("1").toDF().write.text(path.getAbsolutePath)
val schema = StructType(
StructField("a", IntegerType, true) ::
StructField("b", IntegerType, true) :: Nil)
spark.read.schema(schema).option("header", "false").csv(path).show()
```
prints
**Before**
```
java.lang.NumberFormatException: null
at java.lang.Integer.parseInt(Integer.java:542)
at java.lang.Integer.parseInt(Integer.java:615)
at scala.collection.immutable.StringLike$class.toInt(StringLike.scala:272)
at scala.collection.immutable.StringOps.toInt(StringOps.scala:29)
at org.apache.spark.sql.execution.datasources.csv.CSVTypeCast$.castTo(CSVInferSchema.scala:24)
```
**After**
```
+---+----+
| a| b|
+---+----+
| 1|null|
+---+----+
```
## How was this patch tested?
Unit test in `CSVSuite.scala` and `CSVTypeCastSuite.scala`
Author: hyukjinkwon <gurwls223@gmail.com>
Closes#15767 from HyukjinKwon/SPARK-18269.
## What changes were proposed in this pull request?
For data source tables, we will put its table schema, partition columns, etc. to table properties, to work around some hive metastore issues, e.g. not case-preserving, bad decimal type support, etc.
We should also do this for hive serde tables, to reduce the difference between hive serde tables and data source tables, e.g. column names should be case preserving.
## How was this patch tested?
existing tests, and a new test in `HiveExternalCatalog`
Author: Wenchen Fan <wenchen@databricks.com>
Closes#14750 from cloud-fan/minor1.
## What changes were proposed in this pull request?
`QueryExecution.toString` currently captures `java.lang.Throwable`s; this is far from a best practice and can lead to confusing situation or invalid application states. This PR fixes this by only capturing `AnalysisException`s.
## How was this patch tested?
Added a `QueryExecutionSuite`.
Author: Herman van Hovell <hvanhovell@databricks.com>
Closes#15760 from hvanhovell/SPARK-18259.
## What changes were proposed in this pull request?
This patch renames partitionProviderIsHive to tracksPartitionsInCatalog, as the old name was too Hive specific.
## How was this patch tested?
Should be covered by existing tests.
Author: Reynold Xin <rxin@databricks.com>
Closes#15750 from rxin/SPARK-18244.
## What changes were proposed in this pull request?
This PR adds a new hash-based aggregate operator named `ObjectHashAggregateExec` that supports `TypedImperativeAggregate`, which may use arbitrary Java objects as aggregation states. Please refer to the [design doc](https://issues.apache.org/jira/secure/attachment/12834260/%5BDesign%20Doc%5D%20Support%20for%20Arbitrary%20Aggregation%20States.pdf) attached in [SPARK-17949](https://issues.apache.org/jira/browse/SPARK-17949) for more details about it.
The major benefit of this operator is better performance when evaluating `TypedImperativeAggregate` functions, especially when there are relatively few distinct groups. Functions like Hive UDAFs, `collect_list`, and `collect_set` may also benefit from this after being migrated to `TypedImperativeAggregate`.
The following feature flag is introduced to enable or disable the new aggregate operator:
- Name: `spark.sql.execution.useObjectHashAggregateExec`
- Default value: `true`
We can also configure the fallback threshold using the following SQL operation:
- Name: `spark.sql.objectHashAggregate.sortBased.fallbackThreshold`
- Default value: 128
Fallback to sort-based aggregation when more than 128 distinct groups are accumulated in the aggregation hash map. This number is intentionally made small to avoid GC problems since aggregation buffers of this operator may contain arbitrary Java objects.
This may be improved by implementing size tracking for this operator, but that can be done in a separate PR.
Code generation and size tracking are planned to be implemented in follow-up PRs.
## Benchmark results
### `ObjectHashAggregateExec` vs `SortAggregateExec`
The first benchmark compares `ObjectHashAggregateExec` and `SortAggregateExec` by evaluating `typed_count`, a testing `TypedImperativeAggregate` version of the SQL `count` function.
```
Java HotSpot(TM) 64-Bit Server VM 1.8.0_92-b14 on Mac OS X 10.10.5
Intel(R) Core(TM) i7-4960HQ CPU 2.60GHz
object agg v.s. sort agg: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
sort agg w/ group by 31251 / 31908 3.4 298.0 1.0X
object agg w/ group by w/o fallback 6903 / 7141 15.2 65.8 4.5X
object agg w/ group by w/ fallback 20945 / 21613 5.0 199.7 1.5X
sort agg w/o group by 4734 / 5463 22.1 45.2 6.6X
object agg w/o group by w/o fallback 4310 / 4529 24.3 41.1 7.3X
```
The next benchmark compares `ObjectHashAggregateExec` and `SortAggregateExec` by evaluating the Spark native version of `percentile_approx`.
Note that `percentile_approx` is so heavy an aggregate function that the bottleneck of the benchmark is evaluating the aggregate function itself rather than the aggregate operator since I couldn't run a large scale benchmark on my laptop. That's why the results are so close and looks counter-intuitive (aggregation with grouping is even faster than that aggregation without grouping).
```
Java HotSpot(TM) 64-Bit Server VM 1.8.0_92-b14 on Mac OS X 10.10.5
Intel(R) Core(TM) i7-4960HQ CPU 2.60GHz
object agg v.s. sort agg: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
sort agg w/ group by 3418 / 3530 0.6 1630.0 1.0X
object agg w/ group by w/o fallback 3210 / 3314 0.7 1530.7 1.1X
object agg w/ group by w/ fallback 3419 / 3511 0.6 1630.1 1.0X
sort agg w/o group by 4336 / 4499 0.5 2067.3 0.8X
object agg w/o group by w/o fallback 4271 / 4372 0.5 2036.7 0.8X
```
### Hive UDAF vs Spark AF
This benchmark compares the following two kinds of aggregate functions:
- "hive udaf": Hive implementation of `percentile_approx`, without partial aggregation supports, evaluated using `SortAggregateExec`.
- "spark af": Spark native implementation of `percentile_approx`, with partial aggregation support, evaluated using `ObjectHashAggregateExec`
The performance differences are mostly due to faster implementation and partial aggregation support in the Spark native version of `percentile_approx`.
This benchmark basically shows the performance differences between the worst case, where an aggregate function without partial aggregation support is evaluated using `SortAggregateExec`, and the best case, where a `TypedImperativeAggregate` with partial aggregation support is evaluated using `ObjectHashAggregateExec`.
```
Java HotSpot(TM) 64-Bit Server VM 1.8.0_92-b14 on Mac OS X 10.10.5
Intel(R) Core(TM) i7-4960HQ CPU 2.60GHz
hive udaf vs spark af: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
hive udaf w/o group by 5326 / 5408 0.0 81264.2 1.0X
spark af w/o group by 93 / 111 0.7 1415.6 57.4X
hive udaf w/ group by 3804 / 3946 0.0 58050.1 1.4X
spark af w/ group by w/o fallback 71 / 90 0.9 1085.7 74.8X
spark af w/ group by w/ fallback 98 / 111 0.7 1501.6 54.1X
```
### Real world benchmark
We also did a relatively large benchmark using a real world query involving `percentile_approx`:
- Hive UDAF implementation, sort-based aggregation, w/o partial aggregation support
24.77 minutes
- Native implementation, sort-based aggregation, w/ partial aggregation support
4.64 minutes
- Native implementation, object hash aggregator, w/ partial aggregation support
1.80 minutes
## How was this patch tested?
New unit tests and randomized test cases are added in `ObjectAggregateFunctionSuite`.
Author: Cheng Lian <lian@databricks.com>
Closes#15590 from liancheng/obj-hash-agg.
### What changes were proposed in this pull request?
When `FilterExec` contains `isNotNull`, which could be inferred and pushed down or users specified, we convert the nullability of the involved columns if the top-layer expression is null-intolerant. However, this is not correct, if the top-layer expression is not a leaf expression, it could still tolerate the null when it has null-tolerant child expressions.
For example, `cast(coalesce(a#5, a#15) as double)`. Although `cast` is a null-intolerant expression, but obviously`coalesce` is null-tolerant. Thus, it could eat null.
When the nullability is wrong, we could generate incorrect results in different cases. For example,
``` Scala
val df1 = Seq((1, 2), (2, 3)).toDF("a", "b")
val df2 = Seq((2, 5), (3, 4)).toDF("a", "c")
val joinedDf = df1.join(df2, Seq("a"), "outer").na.fill(0)
val df3 = Seq((3, 1)).toDF("a", "d")
joinedDf.join(df3, "a").show
```
The optimized plan is like
```
Project [a#29, b#30, c#31, d#42]
+- Join Inner, (a#29 = a#41)
:- Project [cast(coalesce(cast(coalesce(a#5, a#15) as double), 0.0) as int) AS a#29, cast(coalesce(cast(b#6 as double), 0.0) as int) AS b#30, cast(coalesce(cast(c#16 as double), 0.0) as int) AS c#31]
: +- Filter isnotnull(cast(coalesce(cast(coalesce(a#5, a#15) as double), 0.0) as int))
: +- Join FullOuter, (a#5 = a#15)
: :- LocalRelation [a#5, b#6]
: +- LocalRelation [a#15, c#16]
+- LocalRelation [a#41, d#42]
```
Without the fix, it returns an empty result. With the fix, it can return a correct answer:
```
+---+---+---+---+
| a| b| c| d|
+---+---+---+---+
| 3| 0| 4| 1|
+---+---+---+---+
```
### How was this patch tested?
Added test cases to verify the nullability changes in FilterExec. Also added a test case for verifying the reported incorrect result.
Author: gatorsmile <gatorsmile@gmail.com>
Closes#15523 from gatorsmile/nullabilityFilterExec.
## What changes were proposed in this pull request?
This patch moves the new commit protocol API from sql/core to core module, so we can use it in the future in the RDD API.
As part of this patch, I also moved the speficiation of the random uuid for the write path out of the commit protocol, and instead pass in a job id.
## How was this patch tested?
N/A
Author: Reynold Xin <rxin@databricks.com>
Closes#15731 from rxin/SPARK-18219.
## What changes were proposed in this pull request?
This PR proposes to change the documentation for functions. Please refer the discussion from https://github.com/apache/spark/pull/15513
The changes include
- Re-indent the documentation
- Add examples/arguments in `extended` where the arguments are multiple or specific format (e.g. xml/ json).
For examples, the documentation was updated as below:
### Functions with single line usage
**Before**
- `pow`
``` sql
Usage: pow(x1, x2) - Raise x1 to the power of x2.
Extended Usage:
> SELECT pow(2, 3);
8.0
```
- `current_timestamp`
``` sql
Usage: current_timestamp() - Returns the current timestamp at the start of query evaluation.
Extended Usage:
No example for current_timestamp.
```
**After**
- `pow`
``` sql
Usage: pow(expr1, expr2) - Raises `expr1` to the power of `expr2`.
Extended Usage:
Examples:
> SELECT pow(2, 3);
8.0
```
- `current_timestamp`
``` sql
Usage: current_timestamp() - Returns the current timestamp at the start of query evaluation.
Extended Usage:
No example/argument for current_timestamp.
```
### Functions with (already) multiple line usage
**Before**
- `approx_count_distinct`
``` sql
Usage: approx_count_distinct(expr) - Returns the estimated cardinality by HyperLogLog++.
approx_count_distinct(expr, relativeSD=0.05) - Returns the estimated cardinality by HyperLogLog++
with relativeSD, the maximum estimation error allowed.
Extended Usage:
No example for approx_count_distinct.
```
- `percentile_approx`
``` sql
Usage:
percentile_approx(col, percentage [, accuracy]) - Returns the approximate percentile value of numeric
column `col` at the given percentage. The value of percentage must be between 0.0
and 1.0. The `accuracy` parameter (default: 10000) is a positive integer literal which
controls approximation accuracy at the cost of memory. Higher value of `accuracy` yields
better accuracy, `1.0/accuracy` is the relative error of the approximation.
percentile_approx(col, array(percentage1 [, percentage2]...) [, accuracy]) - Returns the approximate
percentile array of column `col` at the given percentage array. Each value of the
percentage array must be between 0.0 and 1.0. The `accuracy` parameter (default: 10000) is
a positive integer literal which controls approximation accuracy at the cost of memory.
Higher value of `accuracy` yields better accuracy, `1.0/accuracy` is the relative error of
the approximation.
Extended Usage:
No example for percentile_approx.
```
**After**
- `approx_count_distinct`
``` sql
Usage:
approx_count_distinct(expr[, relativeSD]) - Returns the estimated cardinality by HyperLogLog++.
`relativeSD` defines the maximum estimation error allowed.
Extended Usage:
No example/argument for approx_count_distinct.
```
- `percentile_approx`
``` sql
Usage:
percentile_approx(col, percentage [, accuracy]) - Returns the approximate percentile value of numeric
column `col` at the given percentage. The value of percentage must be between 0.0
and 1.0. The `accuracy` parameter (default: 10000) is a positive numeric literal which
controls approximation accuracy at the cost of memory. Higher value of `accuracy` yields
better accuracy, `1.0/accuracy` is the relative error of the approximation.
When `percentage` is an array, each value of the percentage array must be between 0.0 and 1.0.
In this case, returns the approximate percentile array of column `col` at the given
percentage array.
Extended Usage:
Examples:
> SELECT percentile_approx(10.0, array(0.5, 0.4, 0.1), 100);
[10.0,10.0,10.0]
> SELECT percentile_approx(10.0, 0.5, 100);
10.0
```
## How was this patch tested?
Manually tested
**When examples are multiple**
``` sql
spark-sql> describe function extended reflect;
Function: reflect
Class: org.apache.spark.sql.catalyst.expressions.CallMethodViaReflection
Usage: reflect(class, method[, arg1[, arg2 ..]]) - Calls a method with reflection.
Extended Usage:
Examples:
> SELECT reflect('java.util.UUID', 'randomUUID');
c33fb387-8500-4bfa-81d2-6e0e3e930df2
> SELECT reflect('java.util.UUID', 'fromString', 'a5cf6c42-0c85-418f-af6c-3e4e5b1328f2');
a5cf6c42-0c85-418f-af6c-3e4e5b1328f2
```
**When `Usage` is in single line**
``` sql
spark-sql> describe function extended min;
Function: min
Class: org.apache.spark.sql.catalyst.expressions.aggregate.Min
Usage: min(expr) - Returns the minimum value of `expr`.
Extended Usage:
No example/argument for min.
```
**When `Usage` is already in multiple lines**
``` sql
spark-sql> describe function extended percentile_approx;
Function: percentile_approx
Class: org.apache.spark.sql.catalyst.expressions.aggregate.ApproximatePercentile
Usage:
percentile_approx(col, percentage [, accuracy]) - Returns the approximate percentile value of numeric
column `col` at the given percentage. The value of percentage must be between 0.0
and 1.0. The `accuracy` parameter (default: 10000) is a positive numeric literal which
controls approximation accuracy at the cost of memory. Higher value of `accuracy` yields
better accuracy, `1.0/accuracy` is the relative error of the approximation.
When `percentage` is an array, each value of the percentage array must be between 0.0 and 1.0.
In this case, returns the approximate percentile array of column `col` at the given
percentage array.
Extended Usage:
Examples:
> SELECT percentile_approx(10.0, array(0.5, 0.4, 0.1), 100);
[10.0,10.0,10.0]
> SELECT percentile_approx(10.0, 0.5, 100);
10.0
```
**When example/argument is missing**
``` sql
spark-sql> describe function extended rank;
Function: rank
Class: org.apache.spark.sql.catalyst.expressions.Rank
Usage:
rank() - Computes the rank of a value in a group of values. The result is one plus the number
of rows preceding or equal to the current row in the ordering of the partition. The values
will produce gaps in the sequence.
Extended Usage:
No example/argument for rank.
```
Author: hyukjinkwon <gurwls223@gmail.com>
Closes#15677 from HyukjinKwon/SPARK-17963-1.
## What changes were proposed in this pull request?
Due to a limitation of hive metastore(table location must be directory path, not file path), we always store `path` for data source table in storage properties, instead of the `locationUri` field. However, we should not expose this difference to `CatalogTable` level, but just treat it as a hack in `HiveExternalCatalog`, like we store table schema of data source table in table properties.
This PR unifies `path` and `locationUri` outside of `HiveExternalCatalog`, both data source table and hive serde table should use the `locationUri` field.
This PR also unifies the way we handle default table location for managed table. Previously, the default table location of hive serde managed table is set by external catalog, but the one of data source table is set by command. After this PR, we follow the hive way and the default table location is always set by external catalog.
For managed non-file-based tables, we will assign a default table location and create an empty directory for it, the table location will be removed when the table is dropped. This is reasonable as metastore doesn't care about whether a table is file-based or not, and an empty table directory has no harm.
For external non-file-based tables, ideally we can omit the table location, but due to a hive metastore issue, we will assign a random location to it, and remove it right after the table is created. See SPARK-15269 for more details. This is fine as it's well isolated in `HiveExternalCatalog`.
To keep the existing behaviour of the `path` option, in this PR we always add the `locationUri` to storage properties using key `path`, before passing storage properties to `DataSource` as data source options.
## How was this patch tested?
existing tests.
Author: Wenchen Fan <wenchen@databricks.com>
Closes#15024 from cloud-fan/path.
## What changes were proposed in this pull request?
When a user appended a column using a "nondeterministic" function to a DataFrame, e.g., `rand`, `randn`, and `monotonically_increasing_id`, the expected semantic is the following:
- The value in each row should remain unchanged, as if we materialize the column immediately, regardless of later DataFrame operations.
However, since we use `TaskContext.getPartitionId` to get the partition index from the current thread, the values from nondeterministic columns might change if we call `union` or `coalesce` after. `TaskContext.getPartitionId` returns the partition index of the current Spark task, which might not be the corresponding partition index of the DataFrame where we defined the column.
See the unit tests below or JIRA for examples.
This PR uses the partition index from `RDD.mapPartitionWithIndex` instead of `TaskContext` and fixes the partition initialization logic in whole-stage codegen, normal codegen, and codegen fallback. `initializeStatesForPartition(partitionIndex: Int)` was added to `Projection`, `Nondeterministic`, and `Predicate` (codegen) and initialized right after object creation in `mapPartitionWithIndex`. `newPredicate` now returns a `Predicate` instance rather than a function for proper initialization.
## How was this patch tested?
Unit tests. (Actually I'm not very confident that this PR fixed all issues without introducing new ones ...)
cc: rxin davies
Author: Xiangrui Meng <meng@databricks.com>
Closes#15567 from mengxr/SPARK-14393.
## What changes were proposed in this pull request?
Simplify struct creation, especially the aspect of `CleanupAliases` which missed some aliases when handling trees created by `CreateStruct`.
This PR includes:
1. A failing test (create struct with nested aliases, some of the aliases survive `CleanupAliases`).
2. A fix that transforms `CreateStruct` into a `CreateNamedStruct` constructor, effectively eliminating `CreateStruct` from all expression trees.
3. A `NamePlaceHolder` used by `CreateStruct` when column names cannot be extracted from unresolved `NamedExpression`.
4. A new Analyzer rule that resolves `NamePlaceHolder` into a string literal once the `NamedExpression` is resolved.
5. `CleanupAliases` code was simplified as it no longer has to deal with `CreateStruct`'s top level columns.
## How was this patch tested?
Running all tests-suits in package org.apache.spark.sql, especially including the analysis suite, making sure added test initially fails, after applying suggested fix rerun the entire analysis package successfully.
Modified few tests that expected `CreateStruct` which is now transformed into `CreateNamedStruct`.
Author: eyal farago <eyal farago>
Author: Herman van Hovell <hvanhovell@databricks.com>
Author: eyal farago <eyal.farago@gmail.com>
Author: Eyal Farago <eyal.farago@actimize.com>
Author: Hyukjin Kwon <gurwls223@gmail.com>
Author: eyalfa <eyal.farago@gmail.com>
Closes#15718 from hvanhovell/SPARK-16839-2.
## What changes were proposed in this pull request?
Fix `Locale.US` for all usages of `DateFormat`, `NumberFormat`
## How was this patch tested?
Existing tests.
Author: Sean Owen <sowen@cloudera.com>
Closes#15610 from srowen/SPARK-18076.
## What changes were proposed in this pull request?
The PR fixes the bug that the QueryStartedEvent is not logged
the postToAll() in the original code is actually calling StreamingQueryListenerBus.postToAll() which has no listener at all....we shall post by sparkListenerBus.postToAll(s) and this.postToAll() to trigger local listeners as well as the listeners registered in LiveListenerBus
zsxwing
## How was this patch tested?
The following snapshot shows that QueryStartedEvent has been logged correctly
![image](https://cloud.githubusercontent.com/assets/678008/19821553/007a7d28-9d2d-11e6-9f13-49851559cdaa.png)
Author: CodingCat <zhunansjtu@gmail.com>
Closes#15675 from CodingCat/SPARK-18144.
## What changes were proposed in this pull request?
This patch adds support for all file formats in structured streaming sinks. This is actually a very small change thanks to all the previous refactoring done using the new internal commit protocol API.
## How was this patch tested?
Updated FileStreamSinkSuite to add test cases for json, text, and parquet.
Author: Reynold Xin <rxin@databricks.com>
Closes#15711 from rxin/SPARK-18192.
## What changes were proposed in this pull request?
There are a couple issues with the current 2.1 behavior when inserting into Datasource tables with partitions managed by Hive.
(1) OVERWRITE TABLE ... PARTITION will actually overwrite the entire table instead of just the specified partition.
(2) INSERT|OVERWRITE does not work with partitions that have custom locations.
This PR fixes both of these issues for Datasource tables managed by Hive. The behavior for legacy tables or when `manageFilesourcePartitions = false` is unchanged.
There is one other issue in that INSERT OVERWRITE with dynamic partitions will overwrite the entire table instead of just the updated partitions, but this behavior is pretty complicated to implement for Datasource tables. We should address that in a future release.
## How was this patch tested?
Unit tests.
Author: Eric Liang <ekl@databricks.com>
Closes#15705 from ericl/sc-4942.
## What changes were proposed in this pull request?
When the metadata logs for various parts of Structured Streaming are stored on non-HDFS filesystems such as NFS or ext4, the HDFSMetadataLog class leaves hidden HDFS-style checksum (CRC) files in the log directory, one file per batch. This PR modifies HDFSMetadataLog so that it detects the use of a filesystem that doesn't use CRC files and removes the CRC files.
## How was this patch tested?
Modified an existing test case in HDFSMetadataLogSuite to check whether HDFSMetadataLog correctly removes CRC files on the local POSIX filesystem. Ran the entire regression suite.
Author: frreiss <frreiss@us.ibm.com>
Closes#15027 from frreiss/fred-17475.