## What changes were proposed in this pull request?
spark.kafka.sasl.kerberos.service.name is an optional parameter but most of the time value `kafka` has to be set. As I've written in the jira the following reasoning is behind:
* Kafka's configuration guide suggest the same value: https://kafka.apache.org/documentation/#security_sasl_kerberos_brokerconfig
* It would be easier for spark users by providing less configuration
* Other streaming engines are doing the same
In this PR I've changed the parameter from optional to `WithDefault` and set `kafka` as default value.
## How was this patch tested?
Available unit tests + on cluster.
Closes#23254 from gaborgsomogyi/SPARK-26304.
Authored-by: Gabor Somogyi <gabor.g.somogyi@gmail.com>
Signed-off-by: Marcelo Vanzin <vanzin@cloudera.com>
## What changes were proposed in this pull request?
This is the first step of the data source v2 API refactor [proposal](https://docs.google.com/document/d/1uUmKCpWLdh9vHxP7AWJ9EgbwB_U6T3EJYNjhISGmiQg/edit?usp=sharing)
It adds the new API for batch read, without removing the old APIs, as they are still needed for streaming sources.
More concretely, it adds
1. `TableProvider`, works like an anonymous catalog
2. `Table`, represents a structured data set.
3. `ScanBuilder` and `Scan`, a logical represents of data source scan
4. `Batch`, a physical representation of data source batch scan.
## How was this patch tested?
existing tests
Closes#23086 from cloud-fan/refactor-batch.
Authored-by: Wenchen Fan <wenchen@databricks.com>
Signed-off-by: gatorsmile <gatorsmile@gmail.com>
## What changes were proposed in this pull request?
It adds kafka delegation token support for structured streaming. Please see the relevant [SPIP](https://docs.google.com/document/d/1ouRayzaJf_N5VQtGhVq9FURXVmRpXzEEWYHob0ne3NY/edit?usp=sharing)
What this PR contains:
* Configuration parameters for the feature
* Delegation token fetching from broker
* Usage of token through dynamic JAAS configuration
* Minor refactoring in the existing code
What this PR doesn't contain:
* Documentation changes because design can change
## How was this patch tested?
Existing tests + added small amount of additional unit tests.
Because it's an external service integration mainly tested on cluster.
* 4 node cluster
* Kafka broker version 1.1.0
* Topic with 4 partitions
* security.protocol = SASL_SSL
* sasl.mechanism = SCRAM-SHA-256
An example of obtaining a token:
```
18/10/01 01:07:49 INFO kafka010.TokenUtil: TOKENID HMAC OWNER RENEWERS ISSUEDATE EXPIRYDATE MAXDATE
18/10/01 01:07:49 INFO kafka010.TokenUtil: D1-v__Q5T_uHx55rW16Jwg [hidden] User:user [] 2018-10-01T01:07 2018-10-02T01:07 2018-10-08T01:07
18/10/01 01:07:49 INFO security.KafkaDelegationTokenProvider: Get token from Kafka: Kind: KAFKA_DELEGATION_TOKEN, Service: kafka.server.delegation.token, Ident: 44 31 2d 76 5f 5f 51 35 54 5f 75 48 78 35 35 72 57 31 36 4a 77 67
```
An example token usage:
```
18/10/01 01:08:07 INFO kafka010.KafkaSecurityHelper: Scram JAAS params: org.apache.kafka.common.security.scram.ScramLoginModule required tokenauth=true serviceName="kafka" username="D1-v__Q5T_uHx55rW16Jwg" password="[hidden]";
18/10/01 01:08:07 INFO kafka010.KafkaSourceProvider: Delegation token detected, using it for login.
```
Closes#22598 from gaborgsomogyi/SPARK-25501.
Authored-by: Gabor Somogyi <gabor.g.somogyi@gmail.com>
Signed-off-by: Marcelo Vanzin <vanzin@cloudera.com>
## What changes were proposed in this pull request?
Allow the Spark Structured Streaming user to specify the prefix of the consumer group (group.id), compared to force consumer group ids of the form `spark-kafka-source-*`
## How was this patch tested?
Unit tests provided by Spark (backwards compatible change, i.e., user can optionally use the functionality)
`mvn test -pl external/kafka-0-10`
Closes#23103 from zouzias/SPARK-26121.
Authored-by: Anastasios Zouzias <anastasios@sqooba.io>
Signed-off-by: cody koeninger <cody@koeninger.org>
## What changes were proposed in this pull request?
The build has a lot of deprecation warnings. Some are new in Scala 2.12 and Java 11. We've fixed some, but I wanted to take a pass at fixing lots of easy miscellaneous ones here.
They're too numerous and small to list here; see the pull request. Some highlights:
- `BeanInfo` is deprecated in 2.12, and BeanInfo classes are pretty ancient in Java. Instead, case classes can explicitly declare getters
- Eta expansion of zero-arg methods; foo() becomes () => foo() in many cases
- Floating-point Range is inexact and deprecated, like 0.0 to 100.0 by 1.0
- finalize() is finally deprecated (just needs to be suppressed)
- StageInfo.attempId was deprecated and easiest to remove here
I'm not now going to touch some chunks of deprecation warnings:
- Parquet deprecations
- Hive deprecations (particularly serde2 classes)
- Deprecations in generated code (mostly Thriftserver CLI)
- ProcessingTime deprecations (we may need to revive this class as internal)
- many MLlib deprecations because they concern methods that may be removed anyway
- a few Kinesis deprecations I couldn't figure out
- Mesos get/setRole, which I don't know well
- Kafka/ZK deprecations (e.g. poll())
- Kinesis
- a few other ones that will probably resolve by deleting a deprecated method
## How was this patch tested?
Existing tests, including manual testing with the 2.11 build and Java 11.
Closes#23065 from srowen/SPARK-26090.
Authored-by: Sean Owen <sean.owen@databricks.com>
Signed-off-by: Sean Owen <sean.owen@databricks.com>
## What changes were proposed in this pull request?
Deprecated in Java 11, replace Class.newInstance with Class.getConstructor.getInstance, and primtive wrapper class constructors with valueOf or equivalent
## How was this patch tested?
Existing tests.
Closes#22988 from srowen/SPARK-25984.
Authored-by: Sean Owen <sean.owen@databricks.com>
Signed-off-by: Sean Owen <sean.owen@databricks.com>
## What changes were proposed in this pull request?
In this test, i have reduced the test time to 20 secs from 1 minute while reducing the sleep time from 1 sec to 100 milliseconds.
With this change, i was able to run the test in 20+ seconds consistently on my laptop. I would like see if it passes in jenkins consistently.
## How was this patch tested?
Its a test fix.
Closes#22900 from dilipbiswal/SPARK-25618.
Authored-by: Dilip Biswal <dbiswal@us.ibm.com>
Signed-off-by: Dongjoon Hyun <dongjoon@apache.org>
## What changes were proposed in this pull request?
`Literal.value` should have a value a value corresponding to `dataType`. This pr added code to verify it and fixed the existing tests to do so.
## How was this patch tested?
Modified the existing tests.
Closes#22724 from maropu/SPARK-25734.
Authored-by: Takeshi Yamamuro <yamamuro@apache.org>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
## What changes were proposed in this pull request?
Specify `kafka.max.block.ms` to 10 seconds while creating the kafka writer. In the absence of this overridden config, by default it uses a default time out of 60 seconds.
With this change the test completes in close to 10 seconds as opposed to 1 minute.
## How was this patch tested?
This is a test fix.
Closes#22671 from dilipbiswal/SPARK-25615.
Authored-by: Dilip Biswal <dbiswal@us.ibm.com>
Signed-off-by: Sean Owen <sean.owen@databricks.com>
## What changes were proposed in this pull request?
This PR fixes the Scala-2.12 build error due to ambiguity in `foreachBatch` test cases.
- https://amplab.cs.berkeley.edu/jenkins/view/Spark%20QA%20Test%20(Dashboard)/job/spark-master-test-maven-hadoop-2.7-ubuntu-scala-2.12/428/console
```scala
[error] /home/jenkins/workspace/spark-master-test-maven-hadoop-2.7-ubuntu-scala-2.12/sql/core/src/test/scala/org/apache/spark/sql/execution/streaming/sources/ForeachBatchSinkSuite.scala:102: ambiguous reference to overloaded definition,
[error] both method foreachBatch in class DataStreamWriter of type (function: org.apache.spark.api.java.function.VoidFunction2[org.apache.spark.sql.Dataset[Int],Long])org.apache.spark.sql.streaming.DataStreamWriter[Int]
[error] and method foreachBatch in class DataStreamWriter of type (function: (org.apache.spark.sql.Dataset[Int], Long) => Unit)org.apache.spark.sql.streaming.DataStreamWriter[Int]
[error] match argument types ((org.apache.spark.sql.Dataset[Int], Any) => Unit)
[error] ds.writeStream.foreachBatch((_, _) => {}).trigger(Trigger.Continuous("1 second")).start()
[error] ^
[error] /home/jenkins/workspace/spark-master-test-maven-hadoop-2.7-ubuntu-scala-2.12/sql/core/src/test/scala/org/apache/spark/sql/execution/streaming/sources/ForeachBatchSinkSuite.scala:106: ambiguous reference to overloaded definition,
[error] both method foreachBatch in class DataStreamWriter of type (function: org.apache.spark.api.java.function.VoidFunction2[org.apache.spark.sql.Dataset[Int],Long])org.apache.spark.sql.streaming.DataStreamWriter[Int]
[error] and method foreachBatch in class DataStreamWriter of type (function: (org.apache.spark.sql.Dataset[Int], Long) => Unit)org.apache.spark.sql.streaming.DataStreamWriter[Int]
[error] match argument types ((org.apache.spark.sql.Dataset[Int], Any) => Unit)
[error] ds.writeStream.foreachBatch((_, _) => {}).partitionBy("value").start()
[error] ^
```
## How was this patch tested?
Manual.
Since this failure occurs in Scala-2.12 profile and test cases, Jenkins will not test this. We need to build with Scala-2.12 and run the tests.
Closes#22649 from dongjoon-hyun/SPARK-SCALA212.
Authored-by: Dongjoon Hyun <dongjoon@apache.org>
Signed-off-by: Dongjoon Hyun <dongjoon@apache.org>
## What changes were proposed in this pull request?
`FetchedData.reset` should reset `_nextOffsetInFetchedData` and `_offsetAfterPoll`. Otherwise it will cause inconsistent cached data and may make Kafka connector return wrong results.
## How was this patch tested?
The new unit test.
Closes#22507 from zsxwing/fix-kafka-reset.
Lead-authored-by: Shixiong Zhu <zsxwing@gmail.com>
Co-authored-by: Shixiong Zhu <shixiong@databricks.com>
Signed-off-by: Shixiong Zhu <zsxwing@gmail.com>
## What changes were proposed in this pull request?
For self-join/self-union, Spark will produce a physical plan which has multiple `DataSourceV2ScanExec` instances referring to the same `ReadSupport` instance. In this case, the streaming source is indeed scanned multiple times, and the `numInputRows` metrics should be counted for each scan.
Actually we already have 2 test cases to verify the behavior:
1. `StreamingQuerySuite.input row calculation with same V2 source used twice in self-join`
2. `KafkaMicroBatchSourceSuiteBase.ensure stream-stream self-join generates only one offset in log and correct metrics`.
However, in these 2 tests, the expected result is different, which is super confusing. It turns out that, the first test doesn't trigger exchange reuse, so the source is scanned twice. The second test triggers exchange reuse, and the source is scanned only once.
This PR proposes to improve these 2 tests, to test with/without exchange reuse.
## How was this patch tested?
test only change
Closes#22402 from cloud-fan/bug.
Authored-by: Wenchen Fan <wenchen@databricks.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
## What changes were proposed in this pull request?
This PR ensures to call `super.afterAll()` in `override afterAll()` method for test suites.
* Some suites did not call `super.afterAll()`
* Some suites may call `super.afterAll()` only under certain condition
* Others never call `super.afterAll()`.
This PR also ensures to call `super.beforeAll()` in `override beforeAll()` for test suites.
## How was this patch tested?
Existing UTs
Closes#22337 from kiszk/SPARK-25338.
Authored-by: Kazuaki Ishizaki <ishizaki@jp.ibm.com>
Signed-off-by: Dongjoon Hyun <dongjoon@apache.org>
## What changes were proposed in this pull request?
Fix unused imports & outdated comments on `kafka-0-10-sql` module. (Found while I was working on [SPARK-23539](https://github.com/apache/spark/pull/22282))
## How was this patch tested?
Existing unit tests.
Closes#22342 from dongjinleekr/feature/fix-kafka-sql-trivials.
Authored-by: Lee Dongjin <dongjin@apache.org>
Signed-off-by: Sean Owen <sean.owen@databricks.com>
## What changes were proposed in this pull request?
Revert SPARK-24863 (#21819) and SPARK-24748 (#21721) as per discussion in #21721. We will revisit them when the data source v2 APIs are out.
## How was this patch tested?
Jenkins
Closes#22334 from zsxwing/revert-SPARK-24863-SPARK-24748.
Authored-by: Shixiong Zhu <zsxwing@gmail.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
## What changes were proposed in this pull request?
As the user uses Kafka transactions to write data, the offsets in Kafka will be non-consecutive. It will contains some transaction (commit or abort) markers. In addition, if the consumer's `isolation.level` is `read_committed`, `poll` will not return aborted messages either. Hence, we will see non-consecutive offsets in the date returned by `poll`. However, as `seekToEnd` may move the offset point to these missing offsets, there are 4 possible corner cases we need to support:
- The whole batch contains no data messages
- The first offset in a batch is not a committed data message
- The last offset in a batch is not a committed data message
- There is a gap in the middle of a batch
They are all covered by the new unit tests.
## How was this patch tested?
The new unit tests.
Closes#22042 from zsxwing/kafka-transaction-read.
Authored-by: Shixiong Zhu <zsxwing@gmail.com>
Signed-off-by: Shixiong Zhu <zsxwing@gmail.com>
## What changes were proposed in this pull request?
Fix flaky synchronization in Kafka tests - we need to use the scan config that was persisted rather than reconstructing it to identify the stream's current configuration.
We caught most instances of this in the original PR, but this one slipped through.
## How was this patch tested?
n/a
Closes#22245 from jose-torres/fixflake.
Authored-by: Jose Torres <torres.joseph.f+github@gmail.com>
Signed-off-by: Shixiong Zhu <zsxwing@gmail.com>
## What changes were proposed in this pull request?
This is a follow up PR for #22207 to fix a potential flaky test. `processAllAvailable` doesn't work for continuous processing so we should not use it for a continuous query.
## How was this patch tested?
Jenkins.
Closes#22230 from zsxwing/SPARK-25214-2.
Authored-by: Shixiong Zhu <zsxwing@gmail.com>
Signed-off-by: Shixiong Zhu <zsxwing@gmail.com>
## What changes were proposed in this pull request?
When there are missing offsets, Kafka v2 source may return duplicated records when `failOnDataLoss=false` because it doesn't skip missing offsets.
This PR fixes the issue and also adds regression tests for all Kafka readers.
## How was this patch tested?
New tests.
Closes#22207 from zsxwing/SPARK-25214.
Authored-by: Shixiong Zhu <zsxwing@gmail.com>
Signed-off-by: Shixiong Zhu <zsxwing@gmail.com>
## What changes were proposed in this pull request?
This pr proposed to show RDD/relation names in RDD/Hive table scan nodes.
This change made these names show up in the webUI and explain results.
For example;
```
scala> sql("CREATE TABLE t(c1 int) USING hive")
scala> sql("INSERT INTO t VALUES(1)")
scala> spark.table("t").explain()
== Physical Plan ==
Scan hive default.t [c1#8], HiveTableRelation `default`.`t`, org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe, [c1#8]
^^^^^^^^^^^
```
<img width="212" alt="spark-pr-hive" src="https://user-images.githubusercontent.com/692303/44501013-51264c80-a6c6-11e8-94f8-0704aee83bb6.png">
Closes#20226
## How was this patch tested?
Added tests in `DataFrameSuite`, `DatasetSuite`, and `HiveExplainSuite`
Closes#22153 from maropu/pr20226.
Lead-authored-by: Takeshi Yamamuro <yamamuro@apache.org>
Co-authored-by: Tejas Patil <tejasp@fb.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
## What changes were proposed in this pull request?
The race condition that caused test failure is between 2 threads.
- The MicrobatchExecution thread that processes inputs to produce answers and then generates progress events.
- The test thread that generates some input data, checked the answer and then verified the query generated progress event.
The synchronization structure between these threads is as follows
1. MicrobatchExecution thread, in every batch, does the following in order.
a. Processes batch input to generate answer.
b. Signals `awaitProgressLockCondition` to wake up threads waiting for progress using `awaitOffset`
c. Generates progress event
2. Test execution thread
a. Calls `awaitOffset` to wait for progress, which waits on `awaitProgressLockCondition`.
b. As soon as `awaitProgressLockCondition` is signaled, it would move on the in the test to check answer.
c. Finally, it would verify the last generated progress event.
What can happen is the following sequence of events: 2a -> 1a -> 1b -> 2b -> 2c -> 1c.
In other words, the progress event may be generated after the test tries to verify it.
The solution has two steps.
1. Signal the waiting thread after the progress event has been generated, that is, after `finishTrigger()`.
2. Increase the timeout of `awaitProgressLockCondition.await(100 ms)` to a large value.
This latter is to ensure that test thread for keeps waiting on `awaitProgressLockCondition`until the MicroBatchExecution thread explicitly signals it. With the existing small timeout of 100ms the following sequence can occur.
- MicroBatchExecution thread updates committed offsets
- Test thread waiting on `awaitProgressLockCondition` accidentally times out after 100 ms, finds that the committed offsets have been updated, therefore returns from `awaitOffset` and moves on to the progress event tests.
- MicroBatchExecution thread then generates progress event and signals. But the test thread has already attempted to verify the event and failed.
By increasing the timeout to large (e.g., `streamingTimeoutMs = 60 seconds`, similar to `awaitInitialization`), this above type of race condition is also avoided.
## How was this patch tested?
Ran locally many times.
Closes#22182 from tdas/SPARK-25184.
Authored-by: Tathagata Das <tathagata.das1565@gmail.com>
Signed-off-by: Tathagata Das <tathagata.das1565@gmail.com>
## What changes were proposed in this pull request?
Improve the data source v2 API according to the [design doc](https://docs.google.com/document/d/1DDXCTCrup4bKWByTalkXWgavcPdvur8a4eEu8x1BzPM/edit?usp=sharing)
summary of the changes
1. rename `ReadSupport` -> `DataSourceReader` -> `InputPartition` -> `InputPartitionReader` to `BatchReadSupportProvider` -> `BatchReadSupport` -> `InputPartition`/`PartitionReaderFactory` -> `PartitionReader`. Similar renaming also happens at streaming and write APIs.
2. create `ScanConfig` to store query specific information like operator pushdown result, streaming offsets, etc. This makes batch and streaming `ReadSupport`(previouslly named `DataSourceReader`) immutable. All other methods take `ScanConfig` as input, which implies applying operator pushdown and getting streaming offsets happen before all other things(get input partitions, report statistics, etc.).
3. separate `InputPartition` to `InputPartition` and `PartitionReaderFactory`. This is a natural separation, data splitting and reading are orthogonal and we should not mix them in one interfaces. This also makes the naming consistent between read and write API: `PartitionReaderFactory` vs `DataWriterFactory`.
4. separate the batch and streaming interfaces. Sometimes it's painful to force the streaming interface to extend batch interface, as we may need to override some batch methods to return false, or even leak the streaming concept to batch API(e.g. `DataWriterFactory#createWriter(partitionId, taskId, epochId)`)
Some follow-ups we should do after this PR (tracked by https://issues.apache.org/jira/browse/SPARK-25186 ):
1. Revisit the life cycle of `ReadSupport` instances. Currently I keep it same as the previous `DataSourceReader`, i.e. the life cycle is bound to the batch/stream query. This fits streaming very well but may not be perfect for batch source. We can also consider to let `ReadSupport.newScanConfigBuilder` take `DataSourceOptions` as parameter, if we decide to change the life cycle.
2. Add `WriteConfig`. This is similar to `ScanConfig` and makes the write API more flexible. But it's only needed when we add the `replaceWhere` support, and it needs to change the streaming execution engine for this new concept, which I think is better to be done in another PR.
3. Refine the document. This PR adds/changes a lot of document and it's very likely that some people may have better ideas.
4. Figure out the life cycle of `CustomMetrics`. It looks to me that it should be bound to a `ScanConfig`, but we need to change `ProgressReporter` to get the `ScanConfig`. Better to be done in another PR.
5. Better operator pushdown API. This PR keeps the pushdown API as it was, i.e. using the `SupportsPushdownXYZ` traits. We can design a better API using build pattern, but this is a complicated design and deserves an individual JIRA ticket and design doc.
6. Improve the continuous streaming engine to only create a new `ScanConfig` when re-configuring.
7. Remove `SupportsPushdownCatalystFilter`. This is actually not a must-have for file source, we can change the hive partition pruning to use the public `Filter`.
## How was this patch tested?
existing tests.
Closes#22009 from cloud-fan/redesign.
Authored-by: Wenchen Fan <wenchen@databricks.com>
Signed-off-by: Xiao Li <gatorsmile@gmail.com>
## What changes were proposed in this pull request?
This builds on top of SPARK-24748 to report 'offset lag' as a custom metrics for Kafka structured streaming source.
This lag is the difference between the latest offsets in Kafka the time the metrics is reported (just after a micro-batch completes) and the latest offset Spark has processed. It can be 0 (or close to 0) if spark keeps up with the rate at which messages are ingested into Kafka topics in steady state. This measures how far behind the spark source has fallen behind (per partition) and can aid in tuning the application.
## How was this patch tested?
Existing and new unit tests
Please review http://spark.apache.org/contributing.html before opening a pull request.
Closes#21819 from arunmahadevan/SPARK-24863.
Authored-by: Arun Mahadevan <arunm@apache.org>
Signed-off-by: hyukjinkwon <gurwls223@apache.org>
## What changes were proposed in this pull request?
KafkaContinuousSinkSuite leaks a Kafka cluster because both KafkaSourceTest and KafkaContinuousSinkSuite create a Kafka cluster but `afterAll` only shuts down one cluster. This leaks a Kafka cluster and causes that some Kafka thread crash and kill JVM when SBT is trying to clean up tests.
This PR fixes the leak and also adds a shut down hook to detect Kafka cluster leak.
In additions, it also fixes `AdminClient` leak and cleans up cached producers (When a record is writtn using a producer, the producer will keep refreshing the topic and I don't find an API to clear it except closing the producer) to eliminate the following annoying logs:
```
8/13 15:34:42.568 kafka-admin-client-thread | adminclient-4 WARN NetworkClient: [AdminClient clientId=adminclient-4] Connection to node 0 could not be established. Broker may not be available.
18/08/13 15:34:42.570 kafka-admin-client-thread | adminclient-6 WARN NetworkClient: [AdminClient clientId=adminclient-6] Connection to node 0 could not be established. Broker may not be available.
18/08/13 15:34:42.606 kafka-admin-client-thread | adminclient-8 WARN NetworkClient: [AdminClient clientId=adminclient-8] Connection to node -1 could not be established. Broker may not be available.
18/08/13 15:34:42.729 kafka-producer-network-thread | producer-797 WARN NetworkClient: [Producer clientId=producer-797] Connection to node -1 could not be established. Broker may not be available.
18/08/13 15:34:42.906 kafka-producer-network-thread | producer-1598 WARN NetworkClient: [Producer clientId=producer-1598] Connection to node 0 could not be established. Broker may not be available.
```
I also reverted b5eb54244e introduced by #22097 since it doesn't help.
## How was this patch tested?
Jenkins
Closes#22106 from zsxwing/SPARK-25116.
Authored-by: Shixiong Zhu <zsxwing@gmail.com>
Signed-off-by: Shixiong Zhu <zsxwing@gmail.com>
## What changes were proposed in this pull request?
I'm still seeing the Kafka tests failed randomly due to `kafka.zookeeper.ZooKeeperClientTimeoutException: Timed out waiting for connection while in state: CONNECTING`. I checked the test output and saw zookeeper picked up an ipv6 address. Most details can be found in https://issues.apache.org/jira/browse/KAFKA-7193
This PR just uses `127.0.0.1` rather than `localhost` to make sure zookeeper will never use an ipv6 address.
## How was this patch tested?
Jenkins
Closes#22097 from zsxwing/fix-zookeeper-connect.
Authored-by: Shixiong Zhu <zsxwing@gmail.com>
Signed-off-by: Shixiong Zhu <zsxwing@gmail.com>
## What changes were proposed in this pull request?
This PR fixes typo regarding `auxiliary verb + verb[s]`. This is a follow-on of #21956.
## How was this patch tested?
N/A
Closes#22040 from kiszk/spellcheck1.
Authored-by: Kazuaki Ishizaki <ishizaki@jp.ibm.com>
Signed-off-by: hyukjinkwon <gurwls223@apache.org>
## What changes were proposed in this pull request?
A follow up of #21118
Since we use `InternalRow` in the read API of data source v2, we should do the same thing for the write API.
## How was this patch tested?
existing tests.
Author: Wenchen Fan <wenchen@databricks.com>
Closes#21948 from cloud-fan/row-write.
## What changes were proposed in this pull request?
This small fix adds a `consumer.release()` call to `KafkaSourceRDD` in the case where we've retrieved offsets from Kafka, but the `fromOffset` is equal to the `lastOffset`, meaning there is no new data to read for a particular topic partition. Up until now, we'd just return an empty iterator without closing the consumer which would cause a FD leak.
If accepted, this pull request should be merged into master as well.
## How was this patch tested?
Haven't ran any specific tests, would love help on how to test methods running inside `RDD.compute`.
Author: Yuval Itzchakov <yuval.itzchakov@clicktale.com>
Closes#21997 from YuvalItzchakov/master.
## What changes were proposed in this pull request?
Increase ZK timeout and harmonize configs across Kafka tests to resol…ve potentially flaky test failure
## How was this patch tested?
Existing tests
Author: Sean Owen <srowen@gmail.com>
Closes#21995 from srowen/SPARK-18057.3.
## What changes were proposed in this pull request?
This PR addresses issues 2,3 in this [document](https://docs.google.com/document/d/1fbkjEL878witxVQpOCbjlvOvadHtVjYXeB-2mgzDTvk).
* We modified the closure cleaner to identify closures that are implemented via the LambdaMetaFactory mechanism (serializedLambdas) (issue2).
* We also fix the issue due to scala/bug#11016. There are two options for solving the Unit issue, either add () at the end of the closure or use the trick described in the doc. Otherwise overloading resolution does not work (we are not going to eliminate either of the methods) here. Compiler tries to adapt to Unit and makes these two methods candidates for overloading, when there is polymorphic overloading there is no ambiguity (that is the workaround implemented). This does not look that good but it serves its purpose as we need to support two different uses for method: `addTaskCompletionListener`. One that passes a TaskCompletionListener and one that passes a closure that is wrapped with a TaskCompletionListener later on (issue3).
Note: regarding issue 1 in the doc the plan is:
> Do Nothing. Don’t try to fix this as this is only a problem for Java users who would want to use 2.11 binaries. In that case they can cast to MapFunction to be able to utilize lambdas. In Spark 3.0.0 the API should be simplified so that this issue is removed.
## How was this patch tested?
This was manually tested:
```./dev/change-scala-version.sh 2.12
./build/mvn -DskipTests -Pscala-2.12 clean package
./build/mvn -Pscala-2.12 clean package -DwildcardSuites=org.apache.spark.serializer.ProactiveClosureSerializationSuite -Dtest=None
./build/mvn -Pscala-2.12 clean package -DwildcardSuites=org.apache.spark.util.ClosureCleanerSuite -Dtest=None
./build/mvn -Pscala-2.12 clean package -DwildcardSuites=org.apache.spark.streaming.DStreamClosureSuite -Dtest=None```
Author: Stavros Kontopoulos <stavros.kontopoulos@lightbend.com>
Closes#21930 from skonto/scala2.12-sup.
## What changes were proposed in this pull request?
This PR upgrades to the Kafka 2.0.0 release where KIP-266 is integrated.
## How was this patch tested?
This PR uses existing Kafka related unit tests
(Please explain how this patch was tested. E.g. unit tests, integration tests, manual tests)
(If this patch involves UI changes, please attach a screenshot; otherwise, remove this)
Please review http://spark.apache.org/contributing.html before opening a pull request.
Author: tedyu <yuzhihong@gmail.com>
Closes#21488 from tedyu/master.
## What changes were proposed in this pull request?
This updates the DataSourceV2 API to use InternalRow instead of Row for the default case with no scan mix-ins.
Support for readers that produce Row is added through SupportsDeprecatedScanRow, which matches the previous API. Readers that used Row now implement this class and should be migrated to InternalRow.
Readers that previously implemented SupportsScanUnsafeRow have been migrated to use no SupportsScan mix-ins and produce InternalRow.
## How was this patch tested?
This uses existing tests.
Author: Ryan Blue <blue@apache.org>
Closes#21118 from rdblue/SPARK-23325-datasource-v2-internal-row.
## What changes were proposed in this pull request?
As stated in https://github.com/apache/spark/pull/21321, in the error messages we should use `catalogString`. This is not the case, as SPARK-22893 used `simpleString` in order to have the same representation everywhere and it missed some places.
The PR unifies the messages using alway the `catalogString` representation of the dataTypes in the messages.
## How was this patch tested?
existing/modified UTs
Author: Marco Gaido <marcogaido91@gmail.com>
Closes#21804 from mgaido91/SPARK-24268_catalog.
## What changes were proposed in this pull request?
SPARK-22893 tried to unify error messages about dataTypes. Unfortunately, still many places were missing the `simpleString` method in other to have the same representation everywhere.
The PR unified the messages using alway the simpleString representation of the dataTypes in the messages.
## How was this patch tested?
existing/modified UTs
Author: Marco Gaido <marcogaido91@gmail.com>
Closes#21321 from mgaido91/SPARK-24268.
This passes the unique task attempt id instead of attempt number to v2 data sources because attempt number is reused when stages are retried. When attempt numbers are reused, sources that track data by partition id and attempt number may incorrectly clean up data because the same attempt number can be both committed and aborted.
For v1 / Hadoop writes, generate a unique ID based on available attempt numbers to avoid a similar problem.
Closes#21558
Author: Marcelo Vanzin <vanzin@cloudera.com>
Author: Ryan Blue <blue@apache.org>
Closes#21606 from vanzin/SPARK-24552.2.
## What changes were proposed in this pull request?
This PR replaces `getTimeAsMs` with `getTimeAsSeconds` to fix the issue that reading "spark.network.timeout" using a wrong time unit when the user doesn't specify a time out.
## How was this patch tested?
Jenkins
Author: Shixiong Zhu <zsxwing@gmail.com>
Closes#21382 from zsxwing/fix-network-timeout-conf.
## What changes were proposed in this pull request?
`CachedKafkaConsumer` in the project streaming-kafka-0-10 is designed to maintain a pool of KafkaConsumers that can be reused. However, it was built with the assumption there will be only one thread trying to read the same Kafka TopicPartition at the same time. This assumption is not true all the time and this can inadvertently lead to ConcurrentModificationException.
Here is a better way to design this. The consumer pool should be smart enough to avoid concurrent use of a cached consumer. If there is another request for the same TopicPartition as a currently in-use consumer, the pool should automatically return a fresh consumer.
- There are effectively two kinds of consumer that may be generated
- Cached consumer - this should be returned to the pool at task end
- Non-cached consumer - this should be closed at task end
- A trait called `KafkaDataConsumer` is introduced to hide this difference from the users of the consumer so that the client code does not have to reason about whether to stop and release. They simply call `val consumer = KafkaDataConsumer.acquire` and then `consumer.release`.
- If there is request for a consumer that is in-use, then a new consumer is generated.
- If there is request for a consumer which is a task reattempt, then already existing cached consumer will be invalidated and a new consumer is generated. This could fix potential issues if the source of the reattempt is a malfunctioning consumer.
- In addition, I renamed the `CachedKafkaConsumer` class to `KafkaDataConsumer` because is a misnomer given that what it returns may or may not be cached.
## How was this patch tested?
A new stress test that verifies it is safe to concurrently get consumers for the same TopicPartition from the consumer pool.
Author: Gabor Somogyi <gabor.g.somogyi@gmail.com>
Closes#20997 from gaborgsomogyi/SPARK-19185.
## What changes were proposed in this pull request?
SPARK-24073 renames DataReaderFactory -> InputPartition and DataReader -> InputPartitionReader. Some classes still reflects the old name and causes confusion. This patch renames the left over classes to reflect the new interface and fixes a few comments.
## How was this patch tested?
Existing unit tests.
Please review http://spark.apache.org/contributing.html before opening a pull request.
Author: Arun Mahadevan <arunm@apache.org>
Closes#21355 from arunmahadevan/SPARK-24308.
## What changes were proposed in this pull request?
Renames:
* `DataReaderFactory` to `InputPartition`
* `DataReader` to `InputPartitionReader`
* `createDataReaderFactories` to `planInputPartitions`
* `createUnsafeDataReaderFactories` to `planUnsafeInputPartitions`
* `createBatchDataReaderFactories` to `planBatchInputPartitions`
This fixes the changes in SPARK-23219, which renamed ReadTask to
DataReaderFactory. The intent of that change was to make the read and
write API match (write side uses DataWriterFactory), but the underlying
problem is that the two classes are not equivalent.
ReadTask/DataReader function as Iterable/Iterator. One InputPartition is
a specific partition of the data to be read, in contrast to
DataWriterFactory where the same factory instance is used in all write
tasks. InputPartition's purpose is to manage the lifecycle of the
associated reader, which is now called InputPartitionReader, with an
explicit create operation to mirror the close operation. This was no
longer clear from the API because DataReaderFactory appeared to be more
generic than it is and it isn't clear why a set of them is produced for
a read.
## How was this patch tested?
Existing tests, which have been updated to use the new name.
Author: Ryan Blue <blue@apache.org>
Closes#21145 from rdblue/SPARK-24073-revert-data-reader-factory-rename.
## What changes were proposed in this pull request?
This makes it easy to understand at runtime which version is running. Great for debugging production issues.
## How was this patch tested?
Not necessary.
Author: Tathagata Das <tathagata.das1565@gmail.com>
Closes#21160 from tdas/SPARK-24094.
## What changes were proposed in this pull request?
In some streaming queries, the input and processing rates are not calculated at all (shows up as zero) because MicroBatchExecution fails to associated metrics from the executed plan of a trigger with the sources in the logical plan of the trigger. The way this executed-plan-leaf-to-logical-source attribution works is as follows. With V1 sources, there was no way to identify which execution plan leaves were generated by a streaming source. So did a best-effort attempt to match logical and execution plan leaves when the number of leaves were same. In cases where the number of leaves is different, we just give up and report zero rates. An example where this may happen is as follows.
```
val cachedStaticDF = someStaticDF.union(anotherStaticDF).cache()
val streamingInputDF = ...
val query = streamingInputDF.join(cachedStaticDF).writeStream....
```
In this case, the `cachedStaticDF` has multiple logical leaves, but in the trigger's execution plan it only has leaf because a cached subplan is represented as a single InMemoryTableScanExec leaf. This leads to a mismatch in the number of leaves causing the input rates to be computed as zero.
With DataSourceV2, all inputs are represented in the executed plan using `DataSourceV2ScanExec`, each of which has a reference to the associated logical `DataSource` and `DataSourceReader`. So its easy to associate the metrics to the original streaming sources.
In this PR, the solution is as follows. If all the streaming sources in a streaming query as v2 sources, then use a new code path where the execution-metrics-to-source mapping is done directly. Otherwise we fall back to existing mapping logic.
## How was this patch tested?
- New unit tests using V2 memory source
- Existing unit tests using V1 source
Author: Tathagata Das <tathagata.das1565@gmail.com>
Closes#21126 from tdas/SPARK-24050.
## What changes were proposed in this pull request?
Currently, the driver side of the Kafka source (i.e. KafkaMicroBatchReader) eagerly creates a consumer as soon as the Kafk aMicroBatchReader is created. However, we create dummy KafkaMicroBatchReader to get the schema and immediately stop it. Its better to make the consumer creation lazy, it will be created on the first attempt to fetch offsets using the KafkaOffsetReader.
## How was this patch tested?
Existing unit tests
Author: Tathagata Das <tathagata.das1565@gmail.com>
Closes#21134 from tdas/SPARK-24056.
## What changes were proposed in this pull request?
There was no check on nullability for arguments of `Tuple`s. This could lead to have weird behavior when a null value had to be deserialized into a non-nullable Scala object: in those cases, the `null` got silently transformed in a valid value (like `-1` for `Int`), corresponding to the default value we are using in the SQL codebase. This situation was very likely to happen when deserializing to a Tuple of primitive Scala types (like Double, Int, ...).
The PR adds the `AssertNotNull` to arguments of tuples which have been asked to be converted to non-nullable types.
## How was this patch tested?
added UT
Author: Marco Gaido <marcogaido91@gmail.com>
Closes#20976 from mgaido91/SPARK-23835.
## What changes were proposed in this pull request?
This PR implemented the following cleanups related to `UnsafeWriter` class:
- Remove code duplication between `UnsafeRowWriter` and `UnsafeArrayWriter`
- Make `BufferHolder` class internal by delegating its accessor methods to `UnsafeWriter`
- Replace `UnsafeRow.setTotalSize(...)` with `UnsafeRowWriter.setTotalSize()`
## How was this patch tested?
Tested by existing UTs
Author: Kazuaki Ishizaki <ishizaki@jp.ibm.com>
Closes#20850 from kiszk/SPARK-23713.
## What changes were proposed in this pull request?
CacheKafkaConsumer in the project `kafka-0-10-sql` is designed to maintain a pool of KafkaConsumers that can be reused. However, it was built with the assumption there will be only one task using trying to read the same Kafka TopicPartition at the same time. Hence, the cache was keyed by the TopicPartition a consumer is supposed to read. And any cases where this assumption may not be true, we have SparkPlan flag to disable the use of a cache. So it was up to the planner to correctly identify when it was not safe to use the cache and set the flag accordingly.
Fundamentally, this is the wrong way to approach the problem. It is HARD for a high-level planner to reason about the low-level execution model, whether there will be multiple tasks in the same query trying to read the same partition. Case in point, 2.3.0 introduced stream-stream joins, and you can build a streaming self-join query on Kafka. It's pretty non-trivial to figure out how this leads to two tasks reading the same partition twice, possibly concurrently. And due to the non-triviality, it is hard to figure this out in the planner and set the flag to avoid the cache / consumer pool. And this can inadvertently lead to ConcurrentModificationException ,or worse, silent reading of incorrect data.
Here is a better way to design this. The planner shouldnt have to understand these low-level optimizations. Rather the consumer pool should be smart enough avoid concurrent use of a cached consumer. Currently, it tries to do so but incorrectly (the flag inuse is not checked when returning a cached consumer, see [this](https://github.com/apache/spark/blob/master/external/kafka-0-10-sql/src/main/scala/org/apache/spark/sql/kafka010/CachedKafkaConsumer.scala#L403)). If there is another request for the same partition as a currently in-use consumer, the pool should automatically return a fresh consumer that should be closed when the task is done. Then the planner does not have to have a flag to avoid reuses.
This PR is a step towards that goal. It does the following.
- There are effectively two kinds of consumer that may be generated
- Cached consumer - this should be returned to the pool at task end
- Non-cached consumer - this should be closed at task end
- A trait called KafkaConsumer is introduced to hide this difference from the users of the consumer so that the client code does not have to reason about whether to stop and release. They simply called `val consumer = KafkaConsumer.acquire` and then `consumer.release()`.
- If there is request for a consumer that is in-use, then a new consumer is generated.
- If there is a concurrent attempt of the same task, then a new consumer is generated, and the existing cached consumer is marked for close upon release.
- In addition, I renamed the classes because CachedKafkaConsumer is a misnomer given that what it returns may or may not be cached.
This PR does not remove the planner flag to avoid reuse to make this patch safe enough for merging in branch-2.3. This can be done later in master-only.
## How was this patch tested?
A new stress test that verifies it is safe to concurrently get consumers for the same partition from the consumer pool.
Author: Tathagata Das <tathagata.das1565@gmail.com>
Closes#20767 from tdas/SPARK-23623.
## What changes were proposed in this pull request?
As discussion in #20675, we need add a new interface `ContinuousDataReaderFactory` to support the requirements of setting start offset in Continuous Processing.
## How was this patch tested?
Existing UT.
Author: Yuanjian Li <xyliyuanjian@gmail.com>
Closes#20689 from xuanyuanking/SPARK-23533.
## What changes were proposed in this pull request?
Add an epoch ID argument to DataWriterFactory for use in streaming. As a side effect of passing in this value, DataWriter will now have a consistent lifecycle; commit() or abort() ends the lifecycle of a DataWriter instance in any execution mode.
I considered making a separate streaming interface and adding the epoch ID only to that one, but I think it requires a lot of extra work for no real gain. I think it makes sense to define epoch 0 as the one and only epoch of a non-streaming query.
## How was this patch tested?
existing unit tests
Author: Jose Torres <jose@databricks.com>
Closes#20710 from jose-torres/api2.
## What changes were proposed in this pull request?
Currently, when the Kafka source reads from Kafka, it generates as many tasks as the number of partitions in the topic(s) to be read. In some case, it may be beneficial to read the data with greater parallelism, that is, with more number partitions/tasks. That means, offset ranges must be divided up into smaller ranges such the number of records in partition ~= total records in batch / desired partitions. This would also balance out any data skews between topic-partitions.
In this patch, I have added a new option called `minPartitions`, which allows the user to specify the desired level of parallelism.
## How was this patch tested?
New tests in KafkaMicroBatchV2SourceSuite.
Author: Tathagata Das <tathagata.das1565@gmail.com>
Closes#20698 from tdas/SPARK-23541.
## What changes were proposed in this pull request?
var `KafkaContinuousReader.knownPartitions` should be threadsafe as it is accessed from multiple threads - the query thread at the time of reader factory creation, and the epoch tracking thread at the time of `needsReconfiguration`.
## How was this patch tested?
Existing tests.
Author: Tathagata Das <tathagata.das1565@gmail.com>
Closes#20655 from tdas/SPARK-23484.
## What changes were proposed in this pull request?
SPARK-23203: DataSourceV2 should use immutable catalyst trees instead of wrapping a mutable DataSourceV2Reader. This commit updates DataSourceV2Relation and consolidates much of the DataSourceV2 API requirements for the read path in it. Instead of wrapping a reader that changes, the relation lazily produces a reader from its configuration.
This commit also updates the predicate and projection push-down. Instead of the implementation from SPARK-22197, this reuses the rule matching from the Hive and DataSource read paths (using `PhysicalOperation`) and copies most of the implementation of `SparkPlanner.pruneFilterProject`, with updates for DataSourceV2. By reusing the implementation from other read paths, this should have fewer regressions from other read paths and is less code to maintain.
The new push-down rules also supports the following edge cases:
* The output of DataSourceV2Relation should be what is returned by the reader, in case the reader can only partially satisfy the requested schema projection
* The requested projection passed to the DataSourceV2Reader should include filter columns
* The push-down rule may be run more than once if filters are not pushed through projections
## How was this patch tested?
Existing push-down and read tests.
Author: Ryan Blue <blue@apache.org>
Closes#20387 from rdblue/SPARK-22386-push-down-immutable-trees.
## What changes were proposed in this pull request?
Migrating KafkaSource (with data source v1) to KafkaMicroBatchReader (with data source v2).
Performance comparison:
In a unit test with in-process Kafka broker, I tested the read throughput of V1 and V2 using 20M records in a single partition. They were comparable.
## How was this patch tested?
Existing tests, few modified to be better tests than the existing ones.
Author: Tathagata Das <tathagata.das1565@gmail.com>
Closes#20554 from tdas/SPARK-23362.
## What changes were proposed in this pull request?
This is a followup of https://github.com/apache/spark/pull/20435.
While reorganizing the packages for streaming data source v2, the top level stream read/write support interfaces should not be in the reader/writer package, but should be in the `sources.v2` package, to follow the `ReadSupport`, `WriteSupport`, etc.
## How was this patch tested?
N/A
Author: Wenchen Fan <wenchen@databricks.com>
Closes#20509 from cloud-fan/followup.
## What changes were proposed in this pull request?
In the document of `ContinuousReader.setOffset`, we say this method is used to specify the start offset. We also have a `ContinuousReader.getStartOffset` to get the value back. I think it makes more sense to rename `ContinuousReader.setOffset` to `setStartOffset`.
## How was this patch tested?
N/A
Author: Wenchen Fan <wenchen@databricks.com>
Closes#20486 from cloud-fan/rename.
## What changes were proposed in this pull request?
1. create a new package for partitioning/distribution related classes.
As Spark will add new concrete implementations of `Distribution` in new releases, it is good to
have a new package for partitioning/distribution related classes.
2. move streaming related class to package `org.apache.spark.sql.sources.v2.reader/writer.streaming`, instead of `org.apache.spark.sql.sources.v2.streaming.reader/writer`.
So that the there won't be package reader/writer inside package streaming, which is quite confusing.
Before change:
```
v2
├── reader
├── streaming
│ ├── reader
│ └── writer
└── writer
```
After change:
```
v2
├── reader
│ └── streaming
└── writer
└── streaming
```
## How was this patch tested?
Unit test.
Author: Wang Gengliang <ltnwgl@gmail.com>
Closes#20435 from gengliangwang/new_pkg.
## What changes were proposed in this pull request?
All other classes in the reader/writer package doesn't have `V2` in their names, and the streaming reader/writer don't have `V2` either. It's more consistent to remove `V2` from `DataSourceV2Reader` and `DataSourceVWriter`.
Also rename `DataSourceV2Option` to remote the `V2`, we should only have `V2` in the root interface: `DataSourceV2`.
This PR also fixes some places that the mix-in interface doesn't extend the interface it aimed to mix in.
## How was this patch tested?
existing tests.
Author: Wenchen Fan <wenchen@databricks.com>
Closes#20427 from cloud-fan/ds-v2.
## What changes were proposed in this pull request?
Currently we have `ReadTask` in data source v2 reader, while in writer we have `DataWriterFactory`.
To make the naming consistent and better, renaming `ReadTask` to `DataReaderFactory`.
## How was this patch tested?
Unit test
Author: Wang Gengliang <ltnwgl@gmail.com>
Closes#20397 from gengliangwang/rename.
## What changes were proposed in this pull request?
Replace streaming V2 sinks with a unified StreamWriteSupport interface, with a shim to use it with microbatch execution.
Add a new SQL config to use for disabling V2 sinks, falling back to the V1 sink implementation.
## How was this patch tested?
Existing tests, which in the case of Kafka (the only existing continuous V2 sink) now use V2 for microbatch.
Author: Jose Torres <jose@databricks.com>
Closes#20369 from jose-torres/streaming-sink.
## What changes were proposed in this pull request?
KafkaSourceSuiteBase should be abstract class, otherwise KafkaSourceSuiteBase will also run.
## How was this patch tested?
Jenkins
Author: Shixiong Zhu <zsxwing@gmail.com>
Closes#20412 from zsxwing/SPARK-23242.
## What changes were proposed in this pull request?
Currently, `KafkaContinuousSourceStressForDontFailOnDataLossSuite` runs on `MicroBatchExecution`. It should test `ContinuousExecution`.
## How was this patch tested?
Pass the updated test suite.
Author: Dongjoon Hyun <dongjoon@apache.org>
Closes#20374 from dongjoon-hyun/SPARK-23198.
## What changes were proposed in this pull request?
Typo fixes
## How was this patch tested?
Local build / Doc-only changes
Author: Jacek Laskowski <jacek@japila.pl>
Closes#20344 from jaceklaskowski/typo-fixes.
## What changes were proposed in this pull request?
Continuous processing tasks will fail on any attempt number greater than 0. ContinuousExecution will catch these failures and restart globally from the last recorded checkpoints.
## How was this patch tested?
unit test
Author: Jose Torres <jose@databricks.com>
Closes#20225 from jose-torres/no-retry.
## What changes were proposed in this pull request?
The first commit added a new test, and the second refactored the class the test was in. The automatic merge put the test in the wrong place.
## How was this patch tested?
-
Author: Jose Torres <jose@databricks.com>
Closes#20289 from jose-torres/fix.
## What changes were proposed in this pull request?
The Kafka reader is now interruptible and can close itself.
## How was this patch tested?
I locally ran one of the ContinuousKafkaSourceSuite tests in a tight loop. Before the fix, my machine ran out of open file descriptors a few iterations in; now it works fine.
Author: Jose Torres <jose@databricks.com>
Closes#20253 from jose-torres/fix-data-reader.
## What changes were proposed in this pull request?
This problem reported by yanlin-Lynn ivoson and LiangchangZ. Thanks!
When we union 2 streams from kafka or other sources, while one of them have no continues data coming and in the same time task restart, this will cause an `IllegalStateException`. This mainly cause because the code in [MicroBatchExecution](https://github.com/apache/spark/blob/master/sql/core/src/main/scala/org/apache/spark/sql/execution/streaming/MicroBatchExecution.scala#L190) , while one stream has no continues data, its comittedOffset same with availableOffset during `populateStartOffsets`, and `currentPartitionOffsets` not properly handled in KafkaSource. Also, maybe we should also consider this scenario in other Source.
## How was this patch tested?
Add a UT in KafkaSourceSuite.scala
Author: Yuanjian Li <xyliyuanjian@gmail.com>
Closes#20150 from xuanyuanking/SPARK-22956.
## What changes were proposed in this pull request?
Add kafka source and sink for continuous processing. This involves two small changes to the execution engine:
* Bring data reader close() into the normal data reader thread to avoid thread safety issues.
* Fix up the semantics of the RECONFIGURING StreamExecution state. State updates are now atomic, and we don't have to deal with swallowing an exception.
## How was this patch tested?
new unit tests
Author: Jose Torres <jose@databricks.com>
Closes#20096 from jose-torres/continuous-kafka.
## What changes were proposed in this pull request?
Some users depend on source compatibility with the org.apache.spark.sql.execution.streaming.Offset class. Although this is not a stable interface, we can keep it in place for now to simplify upgrades to 2.3.
Author: Jose Torres <jose@databricks.com>
Closes#20012 from joseph-torres/binary-compat.
## What changes were proposed in this pull request?
This PR provides DataSourceV2 API support for structured streaming, including new pieces needed to support continuous processing [SPARK-20928]. High level summary:
- DataSourceV2 includes new mixins to support micro-batch and continuous reads and writes. For reads, we accept an optional user specified schema rather than using the ReadSupportWithSchema model, because doing so would severely complicate the interface.
- DataSourceV2Reader includes new interfaces to read a specific microbatch or read continuously from a given offset. These follow the same setter pattern as the existing Supports* mixins so that they can work with SupportsScanUnsafeRow.
- DataReader (the per-partition reader) has a new subinterface ContinuousDataReader only for continuous processing. This reader has a special method to check progress, and next() blocks for new input rather than returning false.
- Offset, an abstract representation of position in a streaming query, is ported to the public API. (Each type of reader will define its own Offset implementation.)
- DataSourceV2Writer has a new subinterface ContinuousWriter only for continuous processing. Commits to this interface come tagged with an epoch number, as the execution engine will continue to produce new epoch commits as the task continues indefinitely.
Note that this PR does not propose to change the existing DataSourceV2 batch API, or deprecate the existing streaming source/sink internal APIs in spark.sql.execution.streaming.
## How was this patch tested?
Toy implementations of the new interfaces with unit tests.
Author: Jose Torres <jose@databricks.com>
Closes#19925 from joseph-torres/continuous-api.
…a-2.12 and JDK9
## What changes were proposed in this pull request?
Some compile error after upgrading to scala-2.12
```javascript
spark_source/core/src/main/scala/org/apache/spark/executor/Executor.scala:455: ambiguous reference to overloaded definition, method limit in class ByteBuffer of type (x$1: Int)java.nio.ByteBuffer
method limit in class Buffer of type ()Int
match expected type ?
val resultSize = serializedDirectResult.limit
error
```
The limit method was moved from ByteBuffer to the superclass Buffer and it can no longer be called without (). The same reason for position method.
```javascript
/home/zly/prj/oss/jdk9_HOS_SOURCE/spark_source/sql/hive/src/main/scala/org/apache/spark/sql/hive/execution/ScriptTransformationExec.scala:427: ambiguous reference to overloaded definition, [error] both method putAll in class Properties of type (x$1: java.util.Map[_, _])Unit [error] and method putAll in class Hashtable of type (x$1: java.util.Map[_ <: Object, _ <: Object])Unit [error] match argument types (java.util.Map[String,String])
[error] props.putAll(outputSerdeProps.toMap.asJava)
[error] ^
```
This is because the key type is Object instead of String which is unsafe.
## How was this patch tested?
running tests
Please review http://spark.apache.org/contributing.html before opening a pull request.
Author: kellyzly <kellyzly@126.com>
Closes#19854 from kellyzly/SPARK-22660.
## What changes were proposed in this pull request?
Enable Scala 2.12 REPL. Fix most remaining issues with 2.12 compilation and warnings, including:
- Selecting Kafka 0.10.1+ for Scala 2.12 and patching over a minor API difference
- Fixing lots of "eta expansion of zero arg method deprecated" warnings
- Resolving the SparkContext.sequenceFile implicits compile problem
- Fixing an odd but valid jetty-server missing dependency in hive-thriftserver
## How was this patch tested?
Existing tests
Author: Sean Owen <sowen@cloudera.com>
Closes#19307 from srowen/Scala212.
During profiling of a structured streaming application with Kafka as the source, I came across this exception:
This is a 1 minute sample, which caused 106K `NonLocalReturnControl` exceptions to be thrown.
This happens because `CachedKafkaConsumer.get` is ran inside:
`private def runUninterruptiblyIfPossible[T](body: => T): T`
Where `body: => T` is the `get` method. Turning the method into a function means that in order to escape the `while` loop defined in `get` the runtime has to do dirty tricks which involve throwing the above exception.
## What changes were proposed in this pull request?
Instead of using `return` (which is generally not recommended in Scala), we place the result of the `fetchData` method inside a local variable and use a boolean flag to indicate the status of fetching data, which we monitor as our predicate to the `while` loop.
## How was this patch tested?
I've ran the `KafkaSourceSuite` to make sure regression passes. Since the exception isn't visible from user code, there is no way (at least that I could think of) to add this as a test to the existing suite.
Author: Yuval Itzchakov <yuval.itzchakov@clicktale.com>
Closes#19059 from YuvalItzchakov/master.
## What changes were proposed in this pull request?
All streaming logical plans will now have isStreaming set. This involved adding isStreaming as a case class arg in a few cases, since a node might be logically streaming depending on where it came from.
## How was this patch tested?
Existing unit tests - no functional change is intended in this PR.
Author: Jose Torres <joseph-torres@databricks.com>
Author: Tathagata Das <tathagata.das1565@gmail.com>
Closes#18973 from joseph-torres/SPARK-21765.
This version fixes a few issues in the import order checker; it provides
better error messages, and detects more improper ordering (thus the need
to change a lot of files in this patch). The main fix is that it correctly
complains about the order of packages vs. classes.
As part of the above, I moved some "SparkSession" import in ML examples
inside the "$example on$" blocks; that didn't seem consistent across
different source files to start with, and avoids having to add more on/off blocks
around specific imports.
The new scalastyle also seems to have a better header detector, so a few
license headers had to be updated to match the expected indentation.
Author: Marcelo Vanzin <vanzin@cloudera.com>
Closes#18943 from vanzin/SPARK-21731.
## What changes were proposed in this pull request?
Address scapegoat warnings for:
- BigDecimal double constructor
- Catching NPE
- Finalizer without super
- List.size is O(n)
- Prefer Seq.empty
- Prefer Set.empty
- reverse.map instead of reverseMap
- Type shadowing
- Unnecessary if condition.
- Use .log1p
- Var could be val
In some instances like Seq.empty, I avoided making the change even where valid in test code to keep the scope of the change smaller. Those issues are concerned with performance and it won't matter for tests.
## How was this patch tested?
Existing tests
Author: Sean Owen <sowen@cloudera.com>
Closes#18635 from srowen/Scapegoat1.
## What changes were proposed in this pull request?
Currently the `DataFrameWriter` operations have several problems:
1. non-file-format data source writing action doesn't show up in the SQL tab in Spark UI
2. file-format data source writing action shows a scan node in the SQL tab, without saying anything about writing. (streaming also have this issue, but not fixed in this PR)
3. Spark SQL CLI actions don't show up in the SQL tab.
This PR fixes all of them, by refactoring the `ExecuteCommandExec` to make it have children.
close https://github.com/apache/spark/pull/17540
## How was this patch tested?
existing tests.
Also test the UI manually. For a simple command: `Seq(1 -> "a").toDF("i", "j").write.parquet("/tmp/qwe")`
before this PR:
<img width="266" alt="qq20170523-035840 2x" src="https://cloud.githubusercontent.com/assets/3182036/26326050/24e18ba2-3f6c-11e7-8817-6dd275bf6ac5.png">
after this PR:
<img width="287" alt="qq20170523-035708 2x" src="https://cloud.githubusercontent.com/assets/3182036/26326054/2ad7f460-3f6c-11e7-8053-d68325beb28f.png">
Author: Wenchen Fan <wenchen@databricks.com>
Closes#18064 from cloud-fan/execution.
## What changes were proposed in this pull request?
In summary, cost of recreating a KafkaProducer for writing every batch is high as it starts a lot threads and make connections and then closes them. A KafkaProducer instance is promised to be thread safe in Kafka docs. Reuse of KafkaProducer instance while writing via multiple threads is encouraged.
Furthermore, I have performance improvement of 10x in latency, with this patch.
### These are times that addBatch took in ms. Without applying this patch
### These are times that addBatch took in ms. After applying this patch
## How was this patch tested?
Running distributed benchmarks comparing runs with this patch and without it.
Added relevant unit tests.
Author: Prashant Sharma <prashsh1@in.ibm.com>
Closes#17308 from ScrapCodes/cached-kafka-producer.
## What changes were proposed in this pull request?
I checked the logs of https://amplab.cs.berkeley.edu/jenkins/job/spark-branch-2.2-test-maven-hadoop-2.7/47/ and found it took several seconds to create Kafka internal topic `__consumer_offsets`. As Kafka creates this topic lazily, the topic creation happens in the first test `deserialization of initial offset with Spark 2.1.0` and causes it timeout.
This PR changes `offsets.topic.num.partitions` from the default value 50 to 1 to make creating `__consumer_offsets` (50 partitions -> 1 partition) much faster.
## How was this patch tested?
Jenkins
Author: Shixiong Zhu <shixiong@databricks.com>
Closes#17863 from zsxwing/fix-kafka-flaky-test.
## What changes were proposed in this pull request?
We didn't enforce analyzed plans in Spark 2.1 when writing out to Kafka.
## How was this patch tested?
New unit test.
Please review http://spark.apache.org/contributing.html before opening a pull request.
Author: Bill Chambers <bill@databricks.com>
Closes#17804 from anabranch/SPARK-20496-2.
## What changes were proposed in this pull request?
Cancel a batch Kafka query but one of task cannot be cancelled, and rerun the same DataFrame may cause ConcurrentModificationException because it may launch two tasks sharing the same group id.
This PR always create a new consumer when `reuseKafkaConsumer = false` to avoid ConcurrentModificationException. It also contains other minor fixes.
## How was this patch tested?
Jenkins.
Author: Shixiong Zhu <shixiong@databricks.com>
Closes#17752 from zsxwing/kafka-fix.
## What changes were proposed in this pull request?
This PR changes Executor's threads to `UninterruptibleThread` so that we can use `runUninterruptibly` in `CachedKafkaConsumer`. However, this is just best effort to avoid hanging forever. If the user uses`CachedKafkaConsumer` in another thread (e.g., create a new thread or Future), the potential hang may still happen.
## How was this patch tested?
The new added test.
Author: Shixiong Zhu <shixiong@databricks.com>
Closes#17761 from zsxwing/int.
## What changes were proposed in this pull request?
Add Locale.ROOT to internal calls to String `toLowerCase`, `toUpperCase`, to avoid inadvertent locale-sensitive variation in behavior (aka the "Turkish locale problem").
The change looks large but it is just adding `Locale.ROOT` (the locale with no country or language specified) to every call to these methods.
## How was this patch tested?
Existing tests.
Author: Sean Owen <sowen@cloudera.com>
Closes#17527 from srowen/SPARK-20156.
## What changes were proposed in this pull request?
For large trigger intervals (e.g. 10 minutes), if a batch takes 11 minutes, then it will wait for 9 mins before starting the next batch. This does not make sense. The processing time based trigger policy should be to do process batches as fast as possible, but no faster than 1 in every trigger interval. If batches are taking longer than trigger interval anyways, then no point waiting extra trigger interval.
In this PR, I modified the ProcessingTimeExecutor to do so. Another minor change I did was to extract our StreamManualClock into a separate class so that it can be used outside subclasses of StreamTest. For example, ProcessingTimeExecutorSuite does not need to create any context for testing, just needs the StreamManualClock.
## How was this patch tested?
Added new unit tests to comprehensively test this behavior.
Author: Tathagata Das <tathagata.das1565@gmail.com>
Closes#17525 from tdas/SPARK-20209.
## What changes were proposed in this pull request?
An additional trigger and trigger executor that will execute a single trigger only. One can use this OneTime trigger to have more control over the scheduling of triggers.
In addition, this patch requires an optimization to StreamExecution that logs a commit record at the end of successfully processing a batch. This new commit log will be used to determine the next batch (offsets) to process after a restart, instead of using the offset log itself to determine what batch to process next after restart; using the offset log to determine this would process the previously logged batch, always, thus not permitting a OneTime trigger feature.
## How was this patch tested?
A number of existing tests have been revised. These tests all assumed that when restarting a stream, the last batch in the offset log is to be re-processed. Given that we now have a commit log that will tell us if that last batch was processed successfully, the results/assumptions of those tests needed to be revised accordingly.
In addition, a OneTime trigger test was added to StreamingQuerySuite, which tests:
- The semantics of OneTime trigger (i.e., on start, execute a single batch, then stop).
- The case when the commit log was not able to successfully log the completion of a batch before restart, which would mean that we should fall back to what's in the offset log.
- A OneTime trigger execution that results in an exception being thrown.
marmbrus tdas zsxwing
Please review http://spark.apache.org/contributing.html before opening a pull request.
Author: Tyson Condie <tcondie@gmail.com>
Author: Tathagata Das <tathagata.das1565@gmail.com>
Closes#17219 from tcondie/stream-commit.
## Problem
There are several places where we write out version identifiers in various logs for structured streaming (usually `v1`). However, in the places where we check for this, we throw a confusing error message.
## What changes were proposed in this pull request?
This patch made two major changes:
1. added a `parseVersion(...)` method, and based on this method, fixed the following places the way they did version checking (no other place needed to do this checking):
```
HDFSMetadataLog
- CompactibleFileStreamLog ------------> fixed with this patch
- FileStreamSourceLog ---------------> inherited the fix of `CompactibleFileStreamLog`
- FileStreamSinkLog -----------------> inherited the fix of `CompactibleFileStreamLog`
- OffsetSeqLog ------------------------> fixed with this patch
- anonymous subclass in KafkaSource ---> fixed with this patch
```
2. changed the type of `FileStreamSinkLog.VERSION`, `FileStreamSourceLog.VERSION` etc. from `String` to `Int`, so that we can identify newer versions via `version > 1` instead of `version != "v1"`
- note this didn't break any backwards compatibility -- we are still writing out `"v1"` and reading back `"v1"`
## Exception message with this patch
```
java.lang.IllegalStateException: Failed to read log file /private/var/folders/nn/82rmvkk568sd8p3p8tb33trw0000gn/T/spark-86867b65-0069-4ef1-b0eb-d8bd258ff5b8/0. UnsupportedLogVersion: maximum supported log version is v1, but encountered v99. The log file was produced by a newer version of Spark and cannot be read by this version. Please upgrade.
at org.apache.spark.sql.execution.streaming.HDFSMetadataLog.get(HDFSMetadataLog.scala:202)
at org.apache.spark.sql.execution.streaming.OffsetSeqLogSuite$$anonfun$3$$anonfun$apply$mcV$sp$2.apply(OffsetSeqLogSuite.scala:78)
at org.apache.spark.sql.execution.streaming.OffsetSeqLogSuite$$anonfun$3$$anonfun$apply$mcV$sp$2.apply(OffsetSeqLogSuite.scala:75)
at org.apache.spark.sql.test.SQLTestUtils$class.withTempDir(SQLTestUtils.scala:133)
at org.apache.spark.sql.execution.streaming.OffsetSeqLogSuite.withTempDir(OffsetSeqLogSuite.scala:26)
at org.apache.spark.sql.execution.streaming.OffsetSeqLogSuite$$anonfun$3.apply$mcV$sp(OffsetSeqLogSuite.scala:75)
at org.apache.spark.sql.execution.streaming.OffsetSeqLogSuite$$anonfun$3.apply(OffsetSeqLogSuite.scala:75)
at org.apache.spark.sql.execution.streaming.OffsetSeqLogSuite$$anonfun$3.apply(OffsetSeqLogSuite.scala:75)
at org.scalatest.Transformer$$anonfun$apply$1.apply$mcV$sp(Transformer.scala:22)
at org.scalatest.OutcomeOf$class.outcomeOf(OutcomeOf.scala:85)
```
## How was this patch tested?
unit tests
Author: Liwei Lin <lwlin7@gmail.com>
Closes#17070 from lw-lin/better-msg.
When using the KafkaSource with Structured Streaming, consumer assignments are not what the user expects if startingOffsets is set to an explicit set of topics/partitions in JSON where the topic(s) happen to have uppercase characters. When StartingOffsets is constructed, the original string value from options is transformed toLowerCase to make matching on "earliest" and "latest" case insensitive. However, the toLowerCase JSON is passed to SpecificOffsets for the terminal condition, so topic names may not be what the user intended by the time assignments are made with the underlying KafkaConsumer.
KafkaSourceProvider.scala:
```
val startingOffsets = caseInsensitiveParams.get(STARTING_OFFSETS_OPTION_KEY).map(_.trim.toLowerCase) match {
case Some("latest") => LatestOffsets
case Some("earliest") => EarliestOffsets
case Some(json) => SpecificOffsets(JsonUtils.partitionOffsets(json))
case None => LatestOffsets
}
```
Thank cbowden for reporting.
Jenkins
Author: uncleGen <hustyugm@gmail.com>
Closes#17209 from uncleGen/SPARK-19853.
## What changes were proposed in this pull request?
Fix the `throw new IllegalStateException` if statement part.
## How is this patch tested
Regression test
Author: Burak Yavuz <brkyvz@gmail.com>
Closes#17228 from brkyvz/kafka-cause-fix.
## What changes were proposed in this pull request?
Add a new Kafka Sink and Kafka Relation for writing streaming and batch queries, respectively, to Apache Kafka.
### Streaming Kafka Sink
- When addBatch is called
-- If batchId is great than the last written batch
--- Write batch to Kafka
---- Topic will be taken from the record, if present, or from a topic option, which overrides topic in record.
-- Else ignore
### Batch Kafka Sink
- KafkaSourceProvider will implement CreatableRelationProvider
- CreatableRelationProvider#createRelation will write the passed in Dataframe to a Kafka
- Topic will be taken from the record, if present, or from topic option, which overrides topic in record.
- Save modes Append and ErrorIfExist supported under identical semantics. Other save modes result in an AnalysisException
tdas zsxwing
## How was this patch tested?
### The following unit tests will be included
- write to stream with topic field: valid stream write with data that includes an existing topic in the schema
- write structured streaming aggregation w/o topic field, with default topic: valid stream write with data that does not include a topic field, but the configuration includes a default topic
- write data with bad schema: various cases of writing data that does not conform to a proper schema e.g., 1. no topic field or default topic, and 2. no value field
- write data with valid schema but wrong types: data with a complete schema but wrong types e.g., key and value types are integers.
- write to non-existing topic: write a stream to a topic that does not exist in Kafka, which has been configured to not auto-create topics.
- write batch to kafka: simple write batch to Kafka, which goes through the same code path as streaming scenario, so validity checks will not be redone here.
### Examples
```scala
// Structured Streaming
val writer = inputStringStream.map(s => s.get(0).toString.getBytes()).toDF("value")
.selectExpr("value as key", "value as value")
.writeStream
.format("kafka")
.option("checkpointLocation", checkpointDir)
.outputMode(OutputMode.Append)
.option("kafka.bootstrap.servers", brokerAddress)
.option("topic", topic)
.queryName("kafkaStream")
.start()
// Batch
val df = spark
.sparkContext
.parallelize(Seq("1", "2", "3", "4", "5"))
.map(v => (topic, v))
.toDF("topic", "value")
df.write
.format("kafka")
.option("kafka.bootstrap.servers",brokerAddress)
.option("topic", topic)
.save()
```
Please review http://spark.apache.org/contributing.html before opening a pull request.
Author: Tyson Condie <tcondie@gmail.com>
Closes#17043 from tcondie/kafka-writer.
## What changes were proposed in this pull request?
This PR proposes to fix new test failures on WIndows as below:
**Before**
```
KafkaRelationSuite:
- test late binding start offsets *** FAILED *** (7 seconds, 679 milliseconds)
Cause: java.nio.file.FileSystemException: C:\projects\spark\target\tmp\spark-4c4b0cd1-4cb7-4908-949d-1b0cc8addb50\topic-4-0\00000000000000000000.log -> C:\projects\spark\target\tmp\spark-4c4b0cd1-4cb7-4908-949d-1b0cc8addb50\topic-4-0\00000000000000000000.log.deleted: The process cannot access the file because it is being used by another process.
KafkaSourceSuite:
- deserialization of initial offset with Spark 2.1.0 *** FAILED *** (3 seconds, 542 milliseconds)
java.io.IOException: Failed to delete: C:\projects\spark\target\tmp\spark-97ef64fc-ae61-4ce3-ac59-287fd38bd824
- deserialization of initial offset written by Spark 2.1.0 *** FAILED *** (60 milliseconds)
java.nio.file.InvalidPathException: Illegal char <:> at index 2: /C:/projects/spark/external/kafka-0-10-sql/target/scala-2.11/test-classes/kafka-source-initial-offset-version-2.1.0.b
HiveDDLSuite:
- partitioned table should always put partition columns at the end of table schema *** FAILED *** (657 milliseconds)
org.apache.spark.sql.AnalysisException: Path does not exist: file:/C:projectsspark arget mpspark-f1b83d09-850a-4bba-8e43-a2a28dfaa757;
DDLSuite:
- create a data source table without schema *** FAILED *** (94 milliseconds)
org.apache.spark.sql.AnalysisException: Path does not exist: file:/C:projectsspark arget mpspark-a3f3c161-afae-4d6f-9182-e8642f77062b;
- SET LOCATION for managed table *** FAILED *** (219 milliseconds)
org.apache.spark.sql.catalyst.errors.package$TreeNodeException: execute, tree:
Exchange SinglePartit
+- *HashAggregate(keys=[], functions=[partial_count(1)], output=[count#99367L])
+- *FileScan parquet default.tbl[] Batched: true, Format: Parquet, Location: InMemoryFileIndex[file:/C:projectsspark arget mpspark-15be2f2f-4ea9-4c47-bfee-1b7b49363033], PartitionFilters: [], PushedFilters: [], ReadSchema: struct<>
- insert data to a data source table which has a not existed location should succeed *** FAILED *** (16 milliseconds)
org.apache.spark.sql.AnalysisException: Path does not exist: file:/C:projectsspark arget mpspark-34987671-e8d1-4624-ba5b-db1012e1246b;
- insert into a data source table with no existed partition location should succeed *** FAILED *** (16 milliseconds)
org.apache.spark.sql.AnalysisException: Path does not exist: file:/C:projectsspark arget mpspark-4c6ccfbf-4091-4032-9fbc-3d40c58267d5;
- read data from a data source table which has a not existed location should succeed *** FAILED *** (0 milliseconds)
- read data from a data source table with no existed partition location should succeed *** FAILED *** (0 milliseconds)
org.apache.spark.sql.AnalysisException: Path does not exist: file:/C:projectsspark arget mpspark-6af39e37-abd1-44e8-ac68-e2dfcf67a2f3;
InputOutputMetricsSuite:
- output metrics on records written *** FAILED *** (0 milliseconds)
java.lang.IllegalArgumentException: Wrong FS: file://C:\projects\spark\target\tmp\spark-cd69ee77-88f2-4202-bed6-19c0ee05ef55\InputOutputMetricsSuite, expected: file:///
- output metrics on records written - new Hadoop API *** FAILED *** (16 milliseconds)
java.lang.IllegalArgumentException: Wrong FS: file://C:\projects\spark\target\tmp\spark-b69e8fcb-047b-4de8-9cdf-5f026efb6762\InputOutputMetricsSuite, expected: file:///
```
**After**
```
KafkaRelationSuite:
- test late binding start offsets !!! CANCELED !!! (62 milliseconds)
KafkaSourceSuite:
- deserialization of initial offset with Spark 2.1.0 (5 seconds, 341 milliseconds)
- deserialization of initial offset written by Spark 2.1.0 (910 milliseconds)
HiveDDLSuite:
- partitioned table should always put partition columns at the end of table schema (2 seconds)
DDLSuite:
- create a data source table without schema (828 milliseconds)
- SET LOCATION for managed table (406 milliseconds)
- insert data to a data source table which has a not existed location should succeed (406 milliseconds)
- insert into a data source table with no existed partition location should succeed (453 milliseconds)
- read data from a data source table which has a not existed location should succeed (94 milliseconds)
- read data from a data source table with no existed partition location should succeed (265 milliseconds)
InputOutputMetricsSuite:
- output metrics on records written (172 milliseconds)
- output metrics on records written - new Hadoop API (297 milliseconds)
```
## How was this patch tested?
Fixed tests in `InputOutputMetricsSuite`, `KafkaRelationSuite`, `KafkaSourceSuite`, `DDLSuite.scala` and `HiveDDLSuite`.
Manually tested via AppVeyor as below:
`InputOutputMetricsSuite`: https://ci.appveyor.com/project/spark-test/spark/build/633-20170219-windows-test/job/ex8nvwa6tsh7rmto
`KafkaRelationSuite`: https://ci.appveyor.com/project/spark-test/spark/build/633-20170219-windows-test/job/h8dlcowew52y8ncw
`KafkaSourceSuite`: https://ci.appveyor.com/project/spark-test/spark/build/634-20170219-windows-test/job/9ybgjl7yeubxcre4
`DDLSuite`: https://ci.appveyor.com/project/spark-test/spark/build/635-20170219-windows-test
`HiveDDLSuite`: https://ci.appveyor.com/project/spark-test/spark/build/633-20170219-windows-test/job/up6o9n47er087ltb
Author: hyukjinkwon <gurwls223@gmail.com>
Closes#16999 from HyukjinKwon/windows-fix.
## What changes were proposed in this pull request?
The streaming thread in StreamExecution uses the following ways to check if it should exit:
- Catch an InterruptException.
- `StreamExecution.state` is TERMINATED.
When starting and stopping a query quickly, the above two checks may both fail:
- Hit [HADOOP-14084](https://issues.apache.org/jira/browse/HADOOP-14084) and swallow InterruptException
- StreamExecution.stop is called before `state` becomes `ACTIVE`. Then [runBatches](dcc2d540a5/sql/core/src/main/scala/org/apache/spark/sql/execution/streaming/StreamExecution.scala (L252)) changes the state from `TERMINATED` to `ACTIVE`.
If the above cases both happen, the query will hang forever.
This PR changes `state` to `AtomicReference` and uses`compareAndSet` to make sure we only change the state from `INITIALIZING` to `ACTIVE`. It also removes the `runUninterruptibly` hack from ``HDFSMetadata`, because HADOOP-14084 won't cause any problem after we fix the race condition.
## How was this patch tested?
Jenkins
Author: Shixiong Zhu <shixiong@databricks.com>
Closes#16947 from zsxwing/SPARK-19617.
## What changes were proposed in this pull request?
This patch fixes a bug in `KafkaSource` with the (de)serialization of the length of the JSON string that contains the initial partition offsets.
## How was this patch tested?
I ran the test suite for spark-sql-kafka-0-10.
Author: Roberto Agostino Vitillo <ra.vitillo@gmail.com>
Closes#16857 from vitillo/kafka_source_fix.
- Move external/java8-tests tests into core, streaming, sql and remove
- Remove MaxPermGen and related options
- Fix some reflection / TODOs around Java 8+ methods
- Update doc references to 1.7/1.8 differences
- Remove Java 7/8 related build profiles
- Update some plugins for better Java 8 compatibility
- Fix a few Java-related warnings
For the future:
- Update Java 8 examples to fully use Java 8
- Update Java tests to use lambdas for simplicity
- Update Java internal implementations to use lambdas
## How was this patch tested?
Existing tests
Author: Sean Owen <sowen@cloudera.com>
Closes#16871 from srowen/SPARK-19493.
## What changes were proposed in this pull request?
In `KafkaOffsetReader`, when error occurs, we abort the existing consumer and create a new consumer. In our current implementation, the first consumer and the second consumer would be in the same group (which leads to SPARK-19559), **_violating our intention of the two consumers not being in the same group._**
The cause is that, in our current implementation, the first consumer is created before `groupId` and `nextId` are initialized in the constructor. Then even if `groupId` and `nextId` are increased during the creation of that first consumer, `groupId` and `nextId` would still be initialized to default values in the constructor for the second consumer.
We should make sure that `groupId` and `nextId` are initialized before any consumer is created.
## How was this patch tested?
Ran 100 times of `KafkaSourceSuite`; all passed
Author: Liwei Lin <lwlin7@gmail.com>
Closes#16902 from lw-lin/SPARK-19564-.
## What changes were proposed in this pull request?
Today, you can start a stream that reads from kafka. However, given kafka's configurable retention period, it seems like sometimes you might just want to read all of the data that is available now. As such we should add a version that works with spark.read as well.
The options should be the same as the streaming kafka source, with the following differences:
startingOffsets should default to earliest, and should not allow latest (which would always be empty).
endingOffsets should also be allowed and should default to latest. the same assign json format as startingOffsets should also be accepted.
It would be really good, if things like .limit(n) were enough to prevent all the data from being read (this might just work).
## How was this patch tested?
KafkaRelationSuite was added for testing batch queries via KafkaUtils.
Author: Tyson Condie <tcondie@gmail.com>
Closes#16686 from tcondie/SPARK-18682.
