Migrate PR builders from Jenkins to Github Actions #4

HyukjinKwon · 2020-07-07T07:34:12Z

No description provided.

.github/workflows/master.yml

…and dependencies

### What changes were proposed in this pull request? This PR aims to run the Spark tests in Github Actions. To briefly explain the main idea: - Reuse `dev/run-tests.py` with SBT build - Reuse the modules in `dev/sparktestsupport/modules.py` to test each module - Pass the modules to test into `dev/run-tests.py` directly via `TEST_ONLY_MODULES` environment variable. For example, `pyspark-sql,core,sql,hive`. - `dev/run-tests.py` _does not_ take the dependent modules into account but solely the specified modules to test. Another thing to note might be `SlowHiveTest` annotation. Running the tests in Hive modules takes too much so the slow tests are extracted and it runs as a separate job. It was extracted from the actual elapsed time in Jenkins: ![Screen Shot 2020-07-09 at 7 48 13 PM](https://user-images.githubusercontent.com/6477701/87050238-f6098e80-c238-11ea-9c4a-ab505af61381.png) So, Hive tests are separated into to jobs. One is slow test cases, and the other one is the other test cases. _Note that_ the current GitHub Actions build virtually copies what the default PR builder on Jenkins does (without other profiles such as JDK 11, Hadoop 2, etc.). The only exception is Kinesis https://github.com/apache/spark/pull/29057/files#diff-04eb107ee163a50b61281ca08f4e4c7bR23 ### Why are the changes needed? Last week and onwards, the Jenkins machines became very unstable for many reasons: - Apparently, the machines became extremely slow. Almost all tests can't pass. - One machine (worker 4) started to have the corrupt `.m2` which fails the build. - Documentation build fails time to time for an unknown reason in Jenkins machine specifically. This is disabled for now at #29017. - Almost all PRs are basically blocked by this instability currently. The advantages of using Github Actions: - To avoid depending on few persons who can access to the cluster. - To reduce the elapsed time in the build - we could split the tests (e.g., SQL, ML, CORE), and run them in parallel so the total build time will significantly reduce. - To control the environment more flexibly. - Other contributors can test and propose to fix Github Actions configurations so we can distribute this build management cost. Note that: - The current build in Jenkins takes _more than 7 hours_. With Github actions it takes _less than 2 hours_ - We can now control the environments especially for Python easily. - The test and build look more stable than the Jenkins'. ### Does this PR introduce _any_ user-facing change? No, dev-only change. ### How was this patch tested? Tested at HyukjinKwon#4 Closes #29057 from HyukjinKwon/migrate-to-github-actions. Authored-by: HyukjinKwon <gurwls223@apache.org> Signed-off-by: Dongjoon Hyun <dongjoon@apache.org>

This PR aims to run the Spark tests in Github Actions. To briefly explain the main idea: - Reuse `dev/run-tests.py` with SBT build - Reuse the modules in `dev/sparktestsupport/modules.py` to test each module - Pass the modules to test into `dev/run-tests.py` directly via `TEST_ONLY_MODULES` environment variable. For example, `pyspark-sql,core,sql,hive`. - `dev/run-tests.py` _does not_ take the dependent modules into account but solely the specified modules to test. Another thing to note might be `SlowHiveTest` annotation. Running the tests in Hive modules takes too much so the slow tests are extracted and it runs as a separate job. It was extracted from the actual elapsed time in Jenkins: ![Screen Shot 2020-07-09 at 7 48 13 PM](https://user-images.githubusercontent.com/6477701/87050238-f6098e80-c238-11ea-9c4a-ab505af61381.png) So, Hive tests are separated into to jobs. One is slow test cases, and the other one is the other test cases. _Note that_ the current GitHub Actions build virtually copies what the default PR builder on Jenkins does (without other profiles such as JDK 11, Hadoop 2, etc.). The only exception is Kinesis https://github.com/apache/spark/pull/29057/files#diff-04eb107ee163a50b61281ca08f4e4c7bR23 Last week and onwards, the Jenkins machines became very unstable for many reasons: - Apparently, the machines became extremely slow. Almost all tests can't pass. - One machine (worker 4) started to have the corrupt `.m2` which fails the build. - Documentation build fails time to time for an unknown reason in Jenkins machine specifically. This is disabled for now at apache#29017. - Almost all PRs are basically blocked by this instability currently. The advantages of using Github Actions: - To avoid depending on few persons who can access to the cluster. - To reduce the elapsed time in the build - we could split the tests (e.g., SQL, ML, CORE), and run them in parallel so the total build time will significantly reduce. - To control the environment more flexibly. - Other contributors can test and propose to fix Github Actions configurations so we can distribute this build management cost. Note that: - The current build in Jenkins takes _more than 7 hours_. With Github actions it takes _less than 2 hours_ - We can now control the environments especially for Python easily. - The test and build look more stable than the Jenkins'. No, dev-only change. Tested at #4 Closes apache#29057 from HyukjinKwon/migrate-to-github-actions. Authored-by: HyukjinKwon <gurwls223@apache.org> Signed-off-by: Dongjoon Hyun <dongjoon@apache.org>

### What changes were proposed in this pull request? This PR introduces sasl retry count in RetryingBlockTransferor. ### Why are the changes needed? Previously a boolean variable, saslTimeoutSeen, was used. However, the boolean variable wouldn't cover the following scenario: 1. SaslTimeoutException 2. IOException 3. SaslTimeoutException 4. IOException Even though IOException at #2 is retried (resulting in increment of retryCount), the retryCount would be cleared at step #4. Since the intention of saslTimeoutSeen is to undo the increment due to retrying SaslTimeoutException, we should keep a counter for SaslTimeoutException retries and subtract the value of this counter from retryCount. ### Does this PR introduce _any_ user-facing change? No ### How was this patch tested? New test is added, courtesy of Mridul. Closes apache#39611 from tedyu/sasl-cnt. Authored-by: Ted Yu <yuzhihong@gmail.com> Signed-off-by: Mridul Muralidharan <mridul<at>gmail.com>

…edExpression() ### What changes were proposed in this pull request? In `EquivalentExpressions.addExpr()`, add a guard `supportedExpression()` to make it consistent with `addExprTree()` and `getExprState()`. ### Why are the changes needed? This fixes a regression caused by apache#39010 which added the `supportedExpression()` to `addExprTree()` and `getExprState()` but not `addExpr()`. One example of a use case affected by the inconsistency is the `PhysicalAggregation` pattern in physical planning. There, it calls `addExpr()` to deduplicate the aggregate expressions, and then calls `getExprState()` to deduplicate the result expressions. Guarding inconsistently will cause the aggregate and result expressions go out of sync, eventually resulting in query execution error (or whole-stage codegen error). ### Does this PR introduce _any_ user-facing change? This fixes a regression affecting Spark 3.3.2+, where it may manifest as an error running aggregate operators with higher-order functions. Example running the SQL command: ```sql select max(transform(array(id), x -> x)), max(transform(array(id), x -> x)) from range(2) ``` example error message before the fix: ``` java.lang.IllegalStateException: Couldn't find max(transform(array(id#0L), lambdafunction(lambda x#2L, lambda x#2L, false)))#4 in [max(transform(array(id#0L), lambdafunction(lambda x#1L, lambda x#1L, false)))#3] ``` after the fix this error is gone. ### How was this patch tested? Added new test cases to `SubexpressionEliminationSuite` for the immediate issue, and to `DataFrameAggregateSuite` for an example of user-visible symptom. Closes apache#40473 from rednaxelafx/spark-42851. Authored-by: Kris Mok <kris.mok@databricks.com> Signed-off-by: Wenchen Fan <wenchen@databricks.com>

…edExpression() ### What changes were proposed in this pull request? In `EquivalentExpressions.addExpr()`, add a guard `supportedExpression()` to make it consistent with `addExprTree()` and `getExprState()`. ### Why are the changes needed? This fixes a regression caused by apache#39010 which added the `supportedExpression()` to `addExprTree()` and `getExprState()` but not `addExpr()`. One example of a use case affected by the inconsistency is the `PhysicalAggregation` pattern in physical planning. There, it calls `addExpr()` to deduplicate the aggregate expressions, and then calls `getExprState()` to deduplicate the result expressions. Guarding inconsistently will cause the aggregate and result expressions go out of sync, eventually resulting in query execution error (or whole-stage codegen error). ### Does this PR introduce _any_ user-facing change? This fixes a regression affecting Spark 3.3.2+, where it may manifest as an error running aggregate operators with higher-order functions. Example running the SQL command: ```sql select max(transform(array(id), x -> x)), max(transform(array(id), x -> x)) from range(2) ``` example error message before the fix: ``` java.lang.IllegalStateException: Couldn't find max(transform(array(id#0L), lambdafunction(lambda x#2L, lambda x#2L, false)))#4 in [max(transform(array(id#0L), lambdafunction(lambda x#1L, lambda x#1L, false)))#3] ``` after the fix this error is gone. ### How was this patch tested? Added new test cases to `SubexpressionEliminationSuite` for the immediate issue, and to `DataFrameAggregateSuite` for an example of user-visible symptom. Closes apache#40473 from rednaxelafx/spark-42851. Authored-by: Kris Mok <kris.mok@databricks.com> Signed-off-by: Wenchen Fan <wenchen@databricks.com> (cherry picked from commit ef0a76e) Signed-off-by: Wenchen Fan <wenchen@databricks.com>

### What changes were proposed in this pull request? In the `Window` node, both `partitionSpec` and `orderSpec` must be orderable, but the current type check only verifies `orderSpec` is orderable. This can cause an error in later optimizing phases. Given a query: ``` with t as (select id, map(id, id) as m from range(0, 10)) select rank() over (partition by m order by id) from t ``` Before the PR, it fails with an `INTERNAL_ERROR`: ``` org.apache.spark.SparkException: [INTERNAL_ERROR] grouping/join/window partition keys cannot be map type. SQLSTATE: XX000 at org.apache.spark.SparkException$.internalError(SparkException.scala:92) at org.apache.spark.SparkException$.internalError(SparkException.scala:96) at org.apache.spark.sql.catalyst.optimizer.NormalizeFloatingNumbers$.needNormalize(NormalizeFloatingNumbers.scala:103) at org.apache.spark.sql.catalyst.optimizer.NormalizeFloatingNumbers$.org$apache$spark$sql$catalyst$optimizer$NormalizeFloatingNumbers$$needNormalize(NormalizeFloatingNumbers.scala:94) ... ``` After the PR, it fails with a `EXPRESSION_TYPE_IS_NOT_ORDERABLE`, which is expected: ``` org.apache.spark.sql.catalyst.ExtendedAnalysisException: [EXPRESSION_TYPE_IS_NOT_ORDERABLE] Column expression "m" cannot be sorted because its type "MAP<BIGINT, BIGINT>" is not orderable. SQLSTATE: 42822; line 2 pos 53; Project [RANK() OVER (PARTITION BY m ORDER BY id ASC NULLS FIRST ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)#4] +- Project [id#1L, m#0, RANK() OVER (PARTITION BY m ORDER BY id ASC NULLS FIRST ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)#4, RANK() OVER (PARTITION BY m ORDER BY id ASC NULLS FIRST ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)#4] +- Window [rank(id#1L) windowspecdefinition(m#0, id#1L ASC NULLS FIRST, specifiedwindowframe(RowFrame, unboundedpreceding$(), currentrow$())) AS RANK() OVER (PARTITION BY m ORDER BY id ASC NULLS FIRST ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)#4], [m#0], [id#1L ASC NULLS FIRST] +- Project [id#1L, m#0] +- SubqueryAlias t +- SubqueryAlias t +- Project [id#1L, map(id#1L, id#1L) AS m#0] +- Range (0, 10, step=1, splits=None) at org.apache.spark.sql.catalyst.analysis.package$AnalysisErrorAt.failAnalysis(package.scala:52) ... ``` ### How was this patch tested? Unit test. Closes apache#45730 from chenhao-db/SPARK-47572. Authored-by: Chenhao Li <chenhao.li@databricks.com> Signed-off-by: Wenchen Fan <wenchen@databricks.com>

… throw internal error ### What changes were proposed in this pull request? This PR fixes the error messages and classes when Python UDFs are used in higher order functions. ### Why are the changes needed? To show the proper user-facing exceptions with error classes. ### Does this PR introduce _any_ user-facing change? Yes, previously it threw internal error such as: ```python from pyspark.sql.functions import transform, udf, col, array spark.range(1).select(transform(array("id"), lambda x: udf(lambda y: y)(x))).collect() ``` Before: ``` py4j.protocol.Py4JJavaError: An error occurred while calling o74.collectToPython. : org.apache.spark.SparkException: Job aborted due to stage failure: Task 15 in stage 0.0 failed 1 times, most recent failure: Lost task 15.0 in stage 0.0 (TID 15) (ip-192-168-123-103.ap-northeast-2.compute.internal executor driver): org.apache.spark.SparkException: [INTERNAL_ERROR] Cannot evaluate expression: <lambda>(lambda x_0#3L)#2 SQLSTATE: XX000 at org.apache.spark.SparkException$.internalError(SparkException.scala:92) at org.apache.spark.SparkException$.internalError(SparkException.scala:96) ``` After: ``` pyspark.errors.exceptions.captured.AnalysisException: [INVALID_LAMBDA_FUNCTION_CALL.UNEVALUABLE] Invalid lambda function call. Python UDFs should be used in a lambda function at a higher order function. However, "<lambda>(lambda x_0#3L)" was a Python UDF. SQLSTATE: 42K0D; Project [transform(array(id#0L), lambdafunction(<lambda>(lambda x_0#3L)#2, lambda x_0#3L, false)) AS transform(array(id), lambdafunction(<lambda>(lambda x_0#3L), namedlambdavariable()))#4] +- Range (0, 1, step=1, splits=Some(16)) ``` ### How was this patch tested? Unittest was added ### Was this patch authored or co-authored using generative AI tooling? No. Closes apache#47079 from HyukjinKwon/SPARK-48706. Authored-by: Hyukjin Kwon <gurwls223@apache.org> Signed-off-by: Kent Yao <yao@apache.org>

gengliangwang reviewed Jul 7, 2020

View reviewed changes

.github/workflows/master.yml Outdated Show resolved Hide resolved

HyukjinKwon force-pushed the migrate-to-github-actions branch 2 times, most recently from c3fa962 to 7c376df Compare July 7, 2020 12:27

gengliangwang reviewed Jul 8, 2020

View reviewed changes

.github/workflows/master.yml Outdated Show resolved Hide resolved

HyukjinKwon force-pushed the migrate-to-github-actions branch from 891c10d to dc2018b Compare July 8, 2020 13:30

HyukjinKwon mentioned this pull request Jul 8, 2020

[SPARK-32233][TESTS] Disable SBT unidoc generation testing in Jenkins apache/spark#29017

Closed

dongjoon-hyun reviewed Jul 8, 2020

View reviewed changes

.github/workflows/master.yml Show resolved Hide resolved

HyukjinKwon force-pushed the migrate-to-github-actions branch 7 times, most recently from bdb56e2 to ba8029e Compare July 9, 2020 13:26

HyukjinKwon mentioned this pull request Jul 9, 2020

[SPARK-32245][INFRA] Run Spark tests in Github Actions apache/spark#29057

Closed

HyukjinKwon force-pushed the migrate-to-github-actions branch from a267705 to a2afb94 Compare July 9, 2020 14:30

Migrate PR builders from Jenkins to Github Actions

f0962bb

HyukjinKwon force-pushed the migrate-to-github-actions branch from a2afb94 to f0962bb Compare July 9, 2020 14:41

HyukjinKwon added 4 commits July 10, 2020 00:53

Add "pypy" for Jenkins run

f9c3f5a

Fix condition to only search sql and pyspark when it installs Python …

de93af6

…and dependencies

Address comments

a25dcd0

Make job's title prettier

801a740

HyukjinKwon closed this Jul 10, 2020

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Migrate PR builders from Jenkins to Github Actions #4

Migrate PR builders from Jenkins to Github Actions #4

HyukjinKwon commented Jul 7, 2020

Migrate PR builders from Jenkins to Github Actions #4

Migrate PR builders from Jenkins to Github Actions #4

Conversation

HyukjinKwon commented Jul 7, 2020