[SPARK-1522] : YARN ClientBase throws a NPE if there is no YARN Application CP #433
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
Can one of the admins verify this patch? |
| @@ -52,7 +52,7 @@ object SparkBuild extends Build { | |||
| val SCALAC_JVM_VERSION = "jvm-1.6" | |||
| val JAVAC_JVM_VERSION = "1.6" | |||
|
|
|||
| lazy val root = Project("root", file("."), settings = rootSettings) aggregate(allProjects: _*) | |||
There was a problem hiding this comment.
When importing to IntelliJ Idea as an SBT project it uses the name of the Projects and "root" lacks a bit of context. I presume the usage of the word "root" has been based on the SBT multimodule example and lacks a real reason.
There was a problem hiding this comment.
So better?
lazy val spark = Project("spark", file("."), settings = rootSettings) aggregate(allProjects: _*)
|
Most of the changes in the diff look unrelated to what is mentioned in the summary. Please cleanup the diffs and include only what is required to fix the issue without unrelated changes. |
berngp
changed the title
|
@mridulm reverted the changes not related with the issue. |
| } catch { | ||
| case err: NoSuchFieldError => null | ||
| case err: NoSuchFieldException => null | ||
| protected[yarn] def getAppClasspathForKey(key:String, conf:Configuration) |
There was a problem hiding this comment.
please follow style guide for indentations: https://cwiki.apache.org/confluence/display/SPARK/Spark+Code+Style+Guide
|
@berngp Thanks for doing this. I literally ran into this NPE yesterday in my own YARN cluster. It turns out I forgot to point YARN_CONF_DIR to the proper place, but running into a NPE did not leave any clue as to what the problem is (until I dug into the code, which is bad user experience). This PR is a much needed fix. I left a couple of comments. As @tgraves mentioned, the style of this PR is inconsistent with the Spark style guide. Further, it would be good if we could remove several levels of indirection to make the code clearer. |
| protected[yarn] def getMRAppClasspath(conf: Configuration) = | ||
| getAppClasspathForKey("mapreduce.application.classpath", conf)(getDefaultMRApplicationClasspath) | ||
|
|
||
| protected[yarn] def addToAppClasspath(env: HashMap[String, String], elements : Iterable[String]) { |
There was a problem hiding this comment.
Also, in Spark we try not to use protected[*], since the semantics of that aren't intuitive at all. I think these can just be private?
|
@andrewor14 thanks for the review and feedback! |
|
Jenkins, test this please |
|
Merged build triggered. |
|
Merged build started. |
|
Merged build finished. All automated tests passed. |
|
All automated tests passed. |
|
Not convinced that the build should be failing due hive/test taking too long, at least in Travis CI. |
|
Don't worry about Travis for now |
| File.pathSeparator) | ||
| } | ||
| classPathElementsToAdd |
There was a problem hiding this comment.
Since populateHadoopClasspath has a side effect on env by convention I am used to return the actual side effect that was applied. It aids testing/asserting the expectation of such side effect for consumers of the API and since populateHadoopClasspath is part of the public API of the ClientBase:Object I think is a good idea to provide it.
There was a problem hiding this comment.
Better yet @andrewor14 I could change populateHadoopClasspath to avoid a side effect and return the new env. I don't see any consumer of populateHadoopClasspath outside the ClientBase.
There was a problem hiding this comment.
Oops sorry I accidentally deleted my comment "No need to return this"
There was a problem hiding this comment.
Yeah I also did a quick grep for it and I didn't see any usages outside of ClientBase. I didn't realize it was a public API, which seems super strange to me that this returns the classpaths, since this is more like a setter than a getter. I think for now it's OK to leave the public API the same, as you suggest.
There was a problem hiding this comment.
It is strange, public APIs with side effects make me anxious.
|
@berngp This is looking good. I will do a quick test of this on a YARN cluster, and provided that I don't run into anything I think this is good to go. |
|
Thank you @andrewor14, and again I appreciate very much all the feedback. |
| } | ||
| } | ||
|
|
||
| def getDefaultYarnApplicationClasspath: Option[Array[String]] = Try[Array[String]] { |
There was a problem hiding this comment.
By the way I just noticed, it looks like we're changing the public API here. It used to return Array[String], but now it returns an Option. A second point is that using Try here changes the semantics a little bit. Before we propagate any exception that's not NoSuchField*, but now we swallow everything.
I think we should revert this method to what it was before, and have the caller deal with the fact that this can potentially be null.
pwendell
added a commit
to pwendell/spark
that referenced
this pull request
May 12, 2014
Updated Debian packaging
|
@tgravescs any thoughts around this, should I just close this pull request? |
|
@berngp sorry for the delay I think everyone has been busy with the spark 1.0 release. I think this should still be fixed in spark 1.1. Lets leave this pr as is and I'll review it this week. |
|
There is still a lot going on with spark 1.0. I'm going to wait for that to setting down and then review this. |
|
@berngp can you upmerge to the latest master? |
|
@tgravescs done, thank you again for following this PR. |
| object Fixtures { | ||
|
|
||
| val knownDefYarnAppCP: Option[Seq[String]] = | ||
| Some(YarnConfiguration.DEFAULT_YARN_APPLICATION_CLASSPATH) |
There was a problem hiding this comment.
DEFAULT_YARN_APPLICATION_CLASSPATH doesn't exist in hadoop 0.23 so we can't use it directly. That is why we use the reflection in the ClientBase.
There was a problem hiding this comment.
Thanks for caching this.
I will wrap the call in a Try. e.g.
Try(YarnConfiguration.DEFAULT_YARN_APPLICATION_CLASSPATH).toOption
|
Mostly looks good. Fix those couple minor test issues for hadoop 0.23 and I'll commit it. |
…ation specific CP
The current implementation of ClientBase.getDefaultYarnApplicationClasspath inspects
the MRJobConfig class for the field DEFAULT_YARN_APPLICATION_CLASSPATH when it should
be really looking into YarnConfiguration. If the Application Configuration has no
yarn.application.classpath defined a NPE exception will be thrown.
Public API Changes
===========================
* YARN ClientBase getDefault*ApplicationClasspath returns Option[Seq[String]]
This commit depicts how the `ClientBase` API could change the
`getDefaultYarnApplicationClasspath` and `getDefaultMRApplicationClasspath`
to return a `Option[Seq[String]]` while recovering from `NoSuchFieldException`.
Both methods that return the default application's *classpath*, for both *YARN*
as well as *Map Reduce (MR)*, use reflection and per the Java API
documentation they can throw the following exceptions:
* Class:getField(String name):
* NoSuchFieldException - if a field with the specified name is not found.
* NullPointerException - if name is null
* SecurityException - If a security manager, s, is present and any of
the following conditions is met:
1. Invocation of s.checkMemberAccess(this, Member.PUBLIC) denies
access to the field.
2. The caller's class loader is not the same as or an ancestor of
the class loader for the current class and invocation of
`s.checkPackageAccess()` denies access to the package of this class.
* Field:Object get(Object obj):
* IllegalAccessException - if this Field object is enforcing Java language access
control and the underlying field is inaccessible.
* IllegalArgumentException - if the specified object is not an instance of the class
or interface declaring the underlying field (or a subclass or implementor thereof).
* NullPointerException - if the specified object is null and the field
is an instance field.
* ExceptionInInitializerError - if the initialization provoked
by this method fails.
**NOTE**: The above is based on the *Java API for JDK 1.7*
An interesting thing to notice is that the official JDK doesn't mention
the occurrence of the `NoSuchFieldError`. This is completely acceptable
per the JDK spec. The reason is that it is an *Error* and as
described by the Java Language Specification and depicted in
the *Error Class* documentation.
An `Error` "indicates serious problems that a reasonable
application should not try to catch."
While
An `Exception` "indicates conditions that a reasonable
application might want to catch."
If we actually dig deeper according to the *JVM SE7 Specification*
"While Loading, Linking, and Initializing, if an error occurs during resolution
of a symbolic reference, then an instance of
IncompatibleClassChangeError (or a subclass) must be thrown..."
"If an attempt by the Java Virtual Machine to resolve a symbolic reference fails
because an error is thrown that is an instance of LinkageError (or a subclass),
then subsequent attempts to resolve the reference always fail with the same error
that was thrown as a result of the initial resolution attempt."
Now `NoSuchFieldError` extends `LinkageError` which in turn is a `IncompatibleClassChangeError`
and according to its documentation, the *LinkageError Class*,
"indicates that a class has some dependency on another class;
however, the latter class has incompatibly changed after the
compilation of the former class."
Why all these is important and how it relates with a couple of lines of
code?
Well, the original approach catches the two most probable problems
you might encounter if you access, using reflection, a field that you
are almost sure that if it exist it will be of _public_ access but you
are not sure it will always be there. Interesting enough the original
implementation addresses one of the _exceptions_ as well as a potential
_linkage error_ but as mentioned neglects a documented _security exception_,
probably due its unlikeliness to occur.
Fact is that if an error _bubbles_ up the Spark YARN Client doesn't handle it
will terminate, in a probably obscure fashion. The current call stack is
as follows.
Client >> run >> runApp >> ClientBase.setupLaunchEnv >> populateClasspath
In my opinion it is questionable to let an exception escape of this context, the _ClientBase Object_.
In my opinion such _ClientBase Object_ should fail gracefully by handling the potential
_exceptions_ and _linkage error_ while providing enough logging
to let a user know and identify what happened. Yet again, in my opinion
the implementation in this commit handles it in a better, more
resilient, manner than the previous implementation while adding logging
that will help clarify the issues in case of an _exception_.
Additional Changes include:
===========================
* Test Suite for ClientBase added
* Coding Style:
* [Spark Style Guidelines](https://cwiki.apache.org/confluence/display/SPARK/Spark+Code+Style+Guide)
* [Scala Official Style Guidelines](http://docs.scala-lang.org/style/)
* [Scalariform](https://github.com/mdr/scalariform)
* Code refactoring and cleanup per review by andrewor14
Ref.
"JVM SE7 Specification" http://docs.oracle.com/javase/specs/jvms/se7/html/jvms-5.html#jvms-5.4
"Java API for JDK 1.7" http://docs.oracle.com/javase/7/docs/api/
[ticket: SPARK-1522] : https://issues.apache.org/jira/browse/SPARK-1522
Author : berngp
Reviewer : andrewor14, tgravescs
Testing : ?
|
|
||
| def flatten(a: Option[Seq[String]], b: Option[Seq[String]]) = (a ++ b).flatten.toArray | ||
|
|
||
| def getFieldValue[A, B](clazz: Class[_], field: String, defaults: => B)(mapTo: A => B): B = |
There was a problem hiding this comment.
@tgravescs tested for Hadoop 0.23 and 2.4.0. A bit silly that I made such mistake when the main issue was actually accessing through reflection such fields within the ClientBase classs.
|
@tgravescs I fixed the test, squashed the commit and repointed the branch. The latest commit addresses the test failure you were seeing while running it for Hadoop 0.23. |
|
Looks good. Thanks @berngp |
pdeyhim
pushed a commit
to pdeyhim/spark-1
that referenced
this pull request
Jun 25, 2014
…cation CP The current implementation of ClientBase.getDefaultYarnApplicationClasspath inspects the MRJobConfig class for the field DEFAULT_YARN_APPLICATION_CLASSPATH when it should be really looking into YarnConfiguration. If the Application Configuration has no yarn.application.classpath defined a NPE exception will be thrown. Additional Changes include: * Test Suite for ClientBase added [ticket: SPARK-1522] : https://issues.apache.org/jira/browse/SPARK-1522 Author : [email protected] Testing : SPARK_HADOOP_VERSION=2.3.0 SPARK_YARN=true ./sbt/sbt test Author: Bernardo Gomez Palacio <[email protected]> Closes apache#433 from berngp/feature/SPARK-1522 and squashes the following commits: 2c2e118 [Bernardo Gomez Palacio] [SPARK-1522]: YARN ClientBase throws a NPE if there is no YARN Application specific CP
berngp
added a commit
to ThalesGroup/spark
that referenced
this pull request
Jul 22, 2014
…cation CP The current implementation of ClientBase.getDefaultYarnApplicationClasspath inspects the MRJobConfig class for the field DEFAULT_YARN_APPLICATION_CLASSPATH when it should be really looking into YarnConfiguration. If the Application Configuration has no yarn.application.classpath defined a NPE exception will be thrown. Additional Changes include: * Test Suite for ClientBase added [ticket: SPARK-1522] : https://issues.apache.org/jira/browse/SPARK-1522 Author : [email protected] Testing : SPARK_HADOOP_VERSION=2.3.0 SPARK_YARN=true ./sbt/sbt test Author: Bernardo Gomez Palacio <[email protected]> Closes apache#433 from berngp/feature/SPARK-1522 and squashes the following commits: 2c2e118 [Bernardo Gomez Palacio] [SPARK-1522]: YARN ClientBase throws a NPE if there is no YARN Application specific CP
xiliu82
pushed a commit
to xiliu82/spark
that referenced
this pull request
Sep 4, 2014
…cation CP The current implementation of ClientBase.getDefaultYarnApplicationClasspath inspects the MRJobConfig class for the field DEFAULT_YARN_APPLICATION_CLASSPATH when it should be really looking into YarnConfiguration. If the Application Configuration has no yarn.application.classpath defined a NPE exception will be thrown. Additional Changes include: * Test Suite for ClientBase added [ticket: SPARK-1522] : https://issues.apache.org/jira/browse/SPARK-1522 Author : [email protected] Testing : SPARK_HADOOP_VERSION=2.3.0 SPARK_YARN=true ./sbt/sbt test Author: Bernardo Gomez Palacio <[email protected]> Closes apache#433 from berngp/feature/SPARK-1522 and squashes the following commits: 2c2e118 [Bernardo Gomez Palacio] [SPARK-1522]: YARN ClientBase throws a NPE if there is no YARN Application specific CP
andrewor14
pushed a commit
to andrewor14/spark
that referenced
this pull request
Jan 8, 2015
Updated Debian packaging (cherry picked from commit 494d3c0) Signed-off-by: Patrick Wendell <[email protected]>
markhamstra
pushed a commit
to markhamstra/spark
that referenced
this pull request
Nov 7, 2017
This fixes local integration testing
bzhaoopenstack
pushed a commit
to bzhaoopenstack/spark
that referenced
this pull request
Sep 11, 2019
Fix trove deployment failing
arjunshroff
pushed a commit
to arjunshroff/spark
that referenced
this pull request
Nov 24, 2020
apache#433) * SPARK-417 impersonation fixes for spark executor. Impersonation is moved from HadoopRDD.compute() method to org.apache.spark.executor.Executor.run() method * SPARK-363 Hive version changed to '1.2.0-mapr-spark-MEP-6.0.0'
Agirish
pushed a commit
to HPEEzmeral/apache-spark
that referenced
this pull request
May 5, 2022
…cript K8S-1077 (apache#598) * K8S-1077 - use single k8s secret with user info MapR [SPARK-651] Replacing joda-time-*.jar with joda-time-2.10.3.jar. MapR [SPARK-638] Wrong permissions when creating files under directory with GID bit set. MapR [SPARK-627] SparkHistoryServer-2.4 is getting 403 Unauthorized home page for users(spark.ui.view.acls) via spark-submit MapR [SPARK-639] Default headers are adding two times MapR [SPARK-629] Spark UI for job lose CSS styles MapR [MS-925] After upgrade to MEP 6.2 (Spark 2.4.0) can no longer consume Kafka / MapR Streams. MapR [SPARK-626] Update kafka dependencies for Spark 2.4.4.0 in release MEP-6.3.0 MapR [SPARK-340] Jetty web server version at Spark should be updated tp v9.4.X MapR [SPARK-617] an't use ssl via spark beeline MapR [SPARK-617] Can't use ssl via spark beeline MapR [SPARK-620] Replace core dependency in Spark-2.4.4 MapR [SPARK-621] Fix multiple XML configuration initialization for (apache#575) custom headers. Use X-XSS-Protection, X-Content-Type-Options Content-Security-Policy and Strict-Transport-Security configuration only in case: cluster security is enabled OR spark.ui.security.headers.enabled set to true. MapR [SPARK-595] Spark cannot access hs2 through zookeeper Revert "MapR [SPARK-595] Spark cannot access hs2 through zookeeper (apache#577)" MapR [SPARK-595] Spark cannot access hs2 through zookeeper MapR [SPARK-620] Replace core dependency in Spark-2.4. MapR [SPARK-619] Move absent commits from 2.4.3 branch to 2.4.4 (apache#574) * Adding SQL API to write to kafka from Spark (apache#567) * Branch 2.4.3 extended kafka and examples (apache#569) * The v2 API is in its own package - the v2 api is in a different package - the old functionality is available in a separated package * v2 API examples - All the examples are using the newest API. - I have removed the old examples since they are not relevant any more and the same functionality is shown in the new examples usin the new API. * MapR [SPARK-619] Move absent commits from 2.4.3 branch to 2.4.4 CORE-321. Add custom http header support for jetty. MapR [SPARK-609] Port Apache Spark-2.4.4 changes to the MapR Spark-2.4.4 branch Adding multi table loader (apache#560) * Adding multi table loader - This allows us to load multiple matching tables into one Union DataFrame. If we have the fallowing MFS structure: ``` /clients/client_1/data.table /clients/client_2/data.table ``` we can load a union dataframe by doing `loadFromMapRDB("/clients/*/*.table")` * Fixing the path to the reader MapR [SPARK-588] Spark thriftserver fails when work with hive-maprdb json table MapR [SPARK-598] Spark can't add needed properties to hive-site.xml MAPR-SPARK-596: Change HBase compatible version for Spark 2.4.3 MapR [SPARK-592] Add possibility to use start-thriftserver.sh script with 2304 port MapR [SPARK-584] MaprDB connector's setHintUsingIndex method doesn't work as expected MapR [SPARK-583] MaprDB connector's loadFromMaprDB function for Java API doesn't work as expected SPARK-579 info about ssl_trustore is added for metrics MapR [SPARK-552] Failed to get broadcast_11_piece0 of broadcast_11 SPARK-569 Generation of SSL ceritificates for spark UI MapR [SPARK-575] Warning messages in spark workspace after the second attempt to login to job's UI Update zookeeper version Adding `joinWithMapRDBTable` function (apache#529) The related documentation of this function is here https://github.com/anicolaspp/MapRDBConnector#joinwithmaprdbtable. The main idea is that having a dataframe (no matter how was it constructed) we can join it with a MapR-DB table. This functions looks at the join query and load only those records from MapR-DB that will join instead of loading the full table and then join in memory. In other words, we only load what we know will be joint. Adding DataSource Reader Support (apache#525) * Adding DataSource Reader Support * Update SparkSessionExt.scala * creating a package object * Update MapRDBSpark.scala * fully path to avoid name collition * refactorings MapR [SPARK-451] Spark hadoop/core dependency updates MapR [SPARK-566] Move absent commits from 2.4.0 branch MapR [SPARK-561] Spark 2.4.3 porting to MapR MapR [SPARK-561] Spark 2.4.3 porting to MapR MapR [SPARK-558] Render application UI init page if driver is not up MapR [SPARK-541] Avoid duplication of the first unexpired record MapR [COLD-150][K8S] Fix metrics copy MapR [K8S-893] Hide plain text password from logs MapR [SPARK-540] Include 'avro' artifacts MapR [SPARK-536] PySpark streaming package for kafka-0-10 added K8S-853: Enable spark metrics for external tenant MapR [SPARK-531] Remove duplicating entries from classpath in ClasspathFilter MapR [SPARK-516] Spark jobs failure using yarn mode on kerberos fixed MapR [SPARK-462] Spark and SparkHistoryServer allow week ciphers, which can allow man in the middle attack [SPARK-508] MapR-DB OJAI Connector for Spark isNull condition returns incorrect result MapR [SPARK-510] nonmapr "admin" users not able to view other user logs in SHS SPARK-460: Spark Metrics for CollectD Configuration for collecting Spark metrics SPARK-463 MAPR_MAVEN_REPO variable for specifying mapR repository MapR [SPARK-492] Spark 2.4.0.0 configure.sh has error messages MapR [SPARK-515][K8S] Remove configure.sh call for k8s MapR [SPARK-515] Move configuring spark-env.sh back to the private-pkg MapR [SPARK-515] Move configuring spark-env.sh back to the private-pkg MapR [SPARK-514] Recovery from checkpoint is broken MapR [SPARK-445] Messages loss fixed by reverting [MAPR-32290] changes from kafka09 package (apache#460) * MapR [SPARK-445] Revert "[MAPR-32290] Spark processing offsets when messages are already TTL in the first batch (apache#376)" This reverts commit e8d59b9. * MapR [SPARK-445] Revert "[MAPR-32290] Spark processing offsets when messages are already ttl in first batch (apache#368)" This reverts commit b282a8b. MapR [SPARK-445] Messages loss fixed by reverting [MAPR-32290] changes from kafka10 package MapR [SPARK-469] Fix NPE in generated classes by reverting "[SPARK-23466][SQL] Remove redundant null checks in generated Java code by GenerateUnsafeProjection" (apache#455) This reverts commit c5583fd. MapR [SPARK-482] Spark streaming app fails to start by UnknownTopicOrPartitionException with checkpoint MapR [SPARK-496] Spark HS UI doesn't work MapR [SPARK-416] CVE-2018-1320 vulnerability in Apache Thrift MapR [SPARK-486][K8S] Fix sasl encryption error on Kubernetes MapR [SPARK-481] Cannot run spark configure.sh on Client node MapR [K8S-637][K8S] Add configure.sh configuration in spark-defaults.conf for job runtime MapR [SPARK-465] Error messages after update of spark 2.4 MapR [SPARK-465] Error messages after update of spark 2.4 MapR [SPARK-464] Can't submit spark 2.4 jobs from mapr-client [SPARK-466] SparkR errors fixed MapR [SPARK-456] Spark shell can't be started SPARK-417 impersonation fixes for spark executor. Impersonation is mo… (apache#433) * SPARK-417 impersonation fixes for spark executor. Impersonation is moved from HadoopRDD.compute() method to org.apache.spark.executor.Executor.run() method * SPARK-363 Hive version changed to '1.2.0-mapr-spark-MEP-6.0.0' [SPARK-449] Kafka offset commit issue fixed MapR [SPARK-287] Move logic of creating /apps/spark folder from installer's scripts to the configure.sh MapR [SPARK-221] Investigate possibility to move creating of the spark-env.sh from private-pkg to configure.sh MapR [SPARK-430] PID files should be under /opt/mapr/pid MapR [SPARK-446] Spark configure.sh doesn't start/stop Spark services MapR [SPARK-434] Move absent commits from 2.3.2 branch (apache#425) * MapR [SPARK-352] Spark shell fails with "NoClassDefFoundError: org/apache/hadoop/fs/FSDataInputStream" if java is not available in PATH * MapR [SPARK-350] Deprecate Spark Kafka-09 package * MapR [SPARK-326] Investigate possibility of writing Java example for the MapRDB OJAI connector * [SPARK-356] Merge mapr changes from kafka-09 package into the kafka-10 * SPARK-319 Fix for sparkR version check * MapR [SPARK-349] Update OJAI client to v3 for Spark MapR-DB JSON connector * MapR [SPARK-367] Move absent commits from 2.3.1 branch * MapR [SPARK-137] Analyze the warning during compilation of OJAI connector * MapR [SPARK-369] Spark 2.3.2 fails with error related to zookeeper * [MAPR-26258] hbasecontext.HBaseDistributedScanExample fails * [SPARK-24355] Spark external shuffle server improvement to better handle block fetch requests * MapR [SPARK-374] Spark Hive example fails when we submit job from another(simple) cluster user * MapR [SPARK-434] Move absent commits from 2.3.2 branch * MapR [SPARK-434] Move absent commits from 2.3.2 branch * MapR [SPARK-373] Unexpected behavior during job running in standalone cluster mode * MapR [SPARK-419] Update hive-maprdb-json-handler jar for spark 2.3.2.0 and spark 2.2.1 * MapR [SPARK-396] Interface change of sendToKafka * MapR [SPARK-357] consumer groups are prepeneded with a "service_" prefix * MapR [SPARK-429] Changes in maprdb connector are the cause of broken backward compatibility * MapR [SPARK-427] Update kafka in Spark-2.4.0 to the 1.1.1-mapr * MapR [SPARK-434] Move absent commits from 2.3.2 branch * Move absent commits from 2.3.2 branch * MapR [SPARK-434] Move absent commits from 2.3.2 branch * Move absent commits from 2.3.2 branch * Move absent commits from 2.3.2 branch MapR [SPARK-427] Update kafka in Spark-2.4.0 to the 1.1.1-mapr MapR [SPARK-379] Spark 2.4 4-gidit version MapR [PIC-48][K8S] Port k8s changes to 2.4.0 [PIC-48] Create user for k8s driver and executor if required [PIC-48] Create user for k8s driver and executor if required Revert "Remove spark.ui.filters property" This reverts commit d8941ba36c3451cdce15d18d6c1a52991de3b971. [SPARK-351] Copy kubernetes start scripts anyway PIC-34: Rename default configmap name to be consistent with mapr-kubernetes [SPARK-23668][K8S] Add config option for passing through k8s Pod.spec.imagePullSecrets (apache#355) Pass through the `imagePullSecrets` option to the k8s pod in order to allow user to access private image registries. See https://kubernetes.io/docs/tasks/configure-pod-container/pull-image-private-registry/ Unit tests + manual testing. Manual testing procedure: 1. Have private image registry. 2. Spark-submit application with no `spark.kubernetes.imagePullSecret` set. Do `kubectl describe pod ...`. See the error message: ``` Error syncing pod, skipping: failed to "StartContainer" for "spark-kubernetes-driver" with ErrImagePull: "rpc error: code = 2 desc = Error: Status 400 trying to pull repository ...: \"{\\n \\\"errors\\\" : [ {\\n \\\"status\\\" : 400,\\n \\\"message\\\" : \\\"Unsupported docker v1 repository request for '...'\\\"\\n } ]\\n}\"" ``` 3. Create secret `kubectl create secret docker-registry ...` 4. Spark-submit with `spark.kubernetes.imagePullSecret` set to the new secret. See that deployment was successful. Author: Andrew Korzhuev <[email protected]> Author: Andrew Korzhuev <[email protected]> Closes apache#20811 from andrusha/spark-23668-image-pull-secrets. [SPARK-321] Change default value of spark.mapr.ssl.secret.prefix property [PIC-32] Spark on k8s with MapR secure cluster Update entrypoint.sh with correct spark version (apache#340) This PR has minor fix to correct the spark version string [SPARK-274] Create home directory for user who submitted job [MAPR-SPARK-230] Implement security for Spark on Kubernetes Run Spark job with specify the username for driver and executor Read cluster configs from configMap Run configure.sh script form entrypoint.sh Remove spark.kubernetes.driver.pod.commands property Add Spark properties for executor and driver environment variable MapR [SPARK-296] Structured Streaming memory leak Revert "[MAPR-SPARK-210] Rename sprk-defaults.conf to spark-defaults.conf.tem…" (apache#252) * Revert "[MAPR-SPARK-176] Fix Spark Project Catalyst unit tests (apache#251)" This reverts commit 5de05075cd14abf8ac65046a57a5d76617818fbe. * Revert "[MAPR-SPARK-210] Rename sprk-defaults.conf to spark-defaults.conf.template (apache#249)" This reverts commit 1baa677d727e89db7c605ffbae9a9eba00337ad0. [MAPR-SPARK-210] Rename sprk-defaults.conf to spark-defaults.conf.template MapR [SPARK-379] Port Spark to 2.4.0 MapR [SPARK-341] Spark 2.3.2 porting [MAPR-32290] Spark processing offsets when messages are already TTL in the first batch * Bug 32263 - Seek called on unsubscribed partitions [MSPARK-331] Remove snapshot versions of mapr dependencies from Spark-2.3.1 [MAPR-32290] Spark processing offsets when messages are already ttl in first batch MapR [SPARK-325] Add examples for work with the MapRDB JSON connector into the Spark project [ATS-449] Unit test for EBF 32013 created. MAPR-SPARK-311: Spark beeline uses default ssl truststore instead of mapr ssl truststore Bug 32355 - Executor tab empty on Spark UI [SPARK-318] Submitting Spark jobs from Oozie fails due to ClassNotFoundException Bug 32014 - Spark Consumer fails with java.lang.AssertionError Revert "[SPARK-306] Kafka clients 1.0.1 present in jars directory for Spark 2.3.1" (apache#341) * Revert "[SPARK-306] Kafka clients 1.0.1 present in jars directory for Spark 2.3.1 (apache#335)" This reverts commit 832411e. Bug 32014 - Spark Consumer fails with java.lang.AssertionError (apache#326) (apache#336) * MapR [32014] Spark Consumer fails with java.lang.AssertionError [SPARK-306] Kafka clients 1.0.1 present in jars directory for Spark 2.3.1 DEVOPS-2768 temporarily removed curl for file downloading [SPARK-302] Local privilege escalation MapR [SPARK-297] Added unit test for empty value conversion MapR [SPARK-297] Empty values are loaded as non-null MapR [SPARK-296] Structured Streaming memory leak 2.3.1 spark 289 (apache#318) * MapR [SPARK-289] Fix unit test for Spark-2.3.1 [SPARK-130] MapRDB connector - NPE while saving Pair RDD with 'null' values MapR [SPARK-283] Unit tests fail during initialization SSL properties. [SPARK-212] SparkHiveExample fails when we run it twice MapR [SPARK-282] Remove maprfs and hadoop jars from mapr spark package MapR [SPARK-278] Spark submit fails for jobs with python MapR [SPARK-279] Can't connect to spark thrift server with new spark and hive packages MapR [SPARK-276] Update zookeeper dependency to v.3.4.11 for spark 2.3.1 MapR [SPARK-272] Use only client passwords from ssl-client.xml MapR [SPARK-266] Spark jobs can't finish correctly, when there is an error during job running MapR [SPARK-263] Add possibility to use keyPassword which is different from keyStorePassword [MSPARK-31632] RM UI showing broken page for Spark jobs MapR [SPARK-261] Use mapr-security-web for getting passwords. MapR [SPARK-259] Spark application doesn't finish correctly MapR [SPARK-268] Update Spark version for Warden change project version to 2.3.1-mapr-SNAPSHOT MapR [SPARK-256] Spark doesn't work on yarn mode MapR [SPARK-255] Installer fresh install 610/600 secure fails to start "mapr-spark-thriftserver", "mapr-spark-historyserver" Mapr [SPARK-248] MapRDBTableScanRDD fails to convert to Scala Dataframe when using where clause MapR [SPARK-225] Hadoop credentials provider usage for hiding passwords at spark-defaults MapR [SPARK-214] Hive-2.1 poperties can't be read from a hive-site.xml as Spark uses Hive-1.2 MapR [SPARK-216] Spark thriftserver fails when work with hive-maprdb json table SPARK-244 (apache#278) Provide ability to use MapR-Negotiation authentication for Spark HistoryServer MapR [SPARK-226] Spark - pySpark Security Vulnerability MapR [SPARK-220] SparkR fails with UDF functions bug fixed MapR [SPARK-227] KafkaUtils.createDirectStream fails with kafka-09 MapR [SPARK-183] Spark Integration for Kafka 0.10 unit tests disabled MapR [SPARK-182] Spark Project External Kafka Producer v09 unit tests fixed MapR [SPARK-179] Spark Integration for Kafka 0.9 unit tests fixed MapR [SPARK-181] Kafka 0.10 Structured Streaming unit tests fixed [MSPARK-31305] Spark History server NOT loading applications submitted by users other than 'mapr' MapR [SPARK-175] Fix Spark Project Streaming unit tests [MAPR-SPARK-176] Fix Spark Project Catalyst unit tests [MAPR-SPARK-178] Fix Spark Project Hive unit tests MapR [SPARK-174] Spark Core unit tests fixed Changed version for spark-kafka connector. MapR [SPARK-202] Update MapR Spark to 2.3.0 Fixed compile time errors in tests Change project version [SPARK-198] Update hadoop dependency version to 2.7.0-mapr-1803 for Spark 2.2.1 MapR [SPARK-188] Couldn't connect to thrift server via spark beeline on kerberos cluster MapR [SPARK-143] Spark History Server does not require login for secured-by-default clusters MapR [SPARK-186] Update OJAI versions to the latest for Spark-2.2.1 OJAI Connector MapR [SPARK-191] Incorrect work of MapR-DB Sink 'complete' and 'update' modes fixed MapR [SPARK-170] StackOverflowException in equals method in DBMapValue 2.2.1 build fixed (apache#231) * MapR [SPARK-164] Update Kafka version to 1.0.1-mapr in Spark Kafka Producer module MapR [SPARK-161] Include Kafka Structured streaming jar to Spark package. MapR [SPARK-155] Change Spark Master port from 8080 MapR [SPARK-153] Exception in spark job with configured labels on yarn-client mode MapR [SPARK-152] Incorrect date string parsing fixed MapR [SPARK-21] Structured Streaming MapR-DB Sink created MapR [SPARK-135] Spark 2.2 with MapR Streams ( Kafka 1.0) (apache#218) * MapR [SPARK-135] Spark 2.2 with MapR Streams (Kafka 1.0) Added functionality of MapR-Streams specific EOF handling. MapR [SPARK-143] Spark History Server does not require login for secured-by-default clusters Disable build failing if scalastyle checking is fall. MapR [SPARK-16] Change Spark version in Warden files and configure.sh MapR [SPARK-144] Add insertToMapRDB method for rdd for Java API [MAPR-30536] Spark SQL queries on Map column fails after upgrade MapR [SPARK-139] Remove "update" related APIs from connector MapR [SPARK-140] Change the option name "tableName" to "tablePath" in the Spark/MapR-DB connectors. MapR [SPARK-121] Spark OJAI JAVA: update functionality removed MapR [SPARK-118] Spark OJAI Python: missed DataFrame import while moving imports in order to fix MapR [ZEP-101] interpreter issue MapR [SPARK-118] Spark OJAI Python: move MapR DB Connector class importing in order to fix MapR [ZEP-101] interpreter issue MapR [SPARK-117] Spark OJAI Python: Save functionality implementation MapR [SPARK-131] Exception when try to save JSON table with Binary _id field Spark OJAI JAVA: load to RDD, save from RDD implementation (apache#195) * MapR [SPARK-124] Loading to JavaRDD implemented * MapR [SPARK-124] MapRDBJavaSparkContext constructor changed * MapR [SPARK-124] implemented RDD[Row] saving MapR [SPARK-118] Spark OJAI Python: Read implementation MapR [SPARK-128] MapRDB connector - wrong handle of null fields when nullable is false * MapR [SPARK-121] Spark OJAI JAVA: Read to Dataset functionality implementation * Minor refactoring MapR [SPARK-125] Default value of idFieldPath parameter is not handle MapR [SPARK-113] Hit java.lang.UnsupportedOperationException: empty.reduceLeft during loadFromMapRDB Spark Mapr-DB connector was refactored according to Scala style Removed code duplication [MSPARK-107]idField information is lost in MapRDBDataFrameWriterFunctions.saveToMapRDB configure.sh takes options to change ports Kafka client excluded from package because correct version is located in "mapr classpath" Changed Kafka version in Kafka producer module. Branch spark 69 (apache#170) * Fixing the wrong type casting of TimeStamp to OTimeStamp when read from spark dataFrame. * SPARK-69: Problem with license when we try to read from json and write to maprdb remove creatin /usr/local/spark link from configure.sh. This link will be creates by private-pkg remove include-maprdb from default profiles added profiles in maprdb pom file instead of two pom files Fixed maprdb connector dependencies. Fixing the wrong type casting of TimeStamp to OTimeStamp when read from spark dataFrame. changed port for spark-thriftserver as it conflicts with hive server changed port for spark-thriftserver as it conflicts with hive server remove .not_configured_yet file after success Ojai connector fixed required java version [MSPARK-45] Move Spark-OJAI connector code to Spark github repo (apache#132) * SPARK-45 Move Spark-OJAI connector code to Spark github repo * Fixing pom versions for maprdb spark connector. * Changes made to the connector code to be compatible with 5.2.* and 6.0 clients. Spark 2.1.0 mapr 29106 (apache#150) * [SPARK-20922][CORE] Add whitelist of classes that can be deserialized by the launcher. Blindly deserializing classes using Java serialization opens the code up to issues in other libraries, since just deserializing data from a stream may end up execution code (think readObject()). Since the launcher protocol is pretty self-contained, there's just a handful of classes it legitimately needs to deserialize, and they're in just two packages, so add a filter that throws errors if classes from any other package show up in the stream. This also maintains backwards compatibility (the updated launcher code can still communicate with the backend code in older Spark releases). Tested with new and existing unit tests. Author: Marcelo Vanzin <[email protected]> Closes apache#18166 from vanzin/SPARK-20922. (cherry picked from commit 8efc6e9) Signed-off-by: Marcelo Vanzin <[email protected]> (cherry picked from commit 772a9b9) * [SPARK-20922][CORE][HOTFIX] Don't use Java 8 lambdas in older branches. Author: Marcelo Vanzin <[email protected]> Closes apache#18178 from vanzin/SPARK-20922-hotfix. Added security by default for historyserver use waitForConsumerAssignment() instead of consumer.poll(0) for spark-29052 change MAPR_HADOOP_CLASSPATH in configure.sh for creating it by mapr-classpath.sh change MAPR_HADOOP_CLASSPATH in configure.sh for creating it by mapr-classpath.sh changes for mapr-classpath.sh changes for mapr-classpath.sh configure.sh changes [SPARK-39] Classpath filter was added Fixed impersonation when data read from MapR-DB via Spark-Hive. added configure.sh and warden.spark-thriftserver.conf hive-hbase-handler added to Spark jars Fixed "Single message comes late" 28339 bug fixed Spark streaming skipped message with zero offset from Kafka 0.9 [MSPARK-9] Initial fix for Spark unit tests Bump dependencies after ECO-1703 release [SPARK-33] Streaming example fixed [MAPR-26060] Fixed case when mapr-streams make gaps in offsets ported features from kafka 10 to kafka 9 [MAPR-26289][SPARK-2.1] Streaming general improvements (apache#93) * Added include-kafka-09 profile to Assembly * Set default poll timeout to 120s Set default HBase verison to 1.1.8 Changes from Kafka10 package were ported to Kafka09 package. [MAPR-26053] Include MapR Classes to the default value of spark.sql.hive.metastore.sharedPrefixes [MAPR-25807] Spark-Warehouse path computes incorrectly Add MapR-SASL support for Thrift Server Adding scala library. [MAPR-25713] Spark might try to load MapR Class Loader multiple times and fail [MAPR-25311] Bump Spark dependencies after ECO-1611 release [MINOR] Fix spark-jars.sh script [MAPR-24603] Could not launch beeline shell after starting spark thrift server fixed syntax error in V09DirectKafkaWordCount example Spark 2.0.1 MAPR-streams Python API [MAPR-24415] SPARK_JAVA_OPTS is deprecated Kafka streaming producer added. Minor fix for previous commit Added script for MAPR-24374 Some minor changes to spark-defaults.conf Changed default HBase version to 1.1.1 in compatibility.version Streaming example was refactored [MAPR-24470] HiveFromSpark test fails in yarn-cluster mode Added MapR Repo [MAPR-22940] Failed to connect spark beeline (after spark thrift server is started) on Kerberos cluster [MAPR-18865] Unable to submit spark apps from Windows client Skip maven clean task on the parent module New: Issue with running Hive commands in Spark This is fixed in SPARK-7819 Isolated Hive Client Loader appears to cause Native Library libMapRClient.4.0.2-mapr.so already loaded in another classloader error Spark warden.services.conf should have dependency on cldb Remove DFS shuffle settings. These settings are not used right now. Copy every file in the conf directory into the distribution package. Create spark-defaults.conf for MapR Settings to enable DFS shuffle on MapR. Support hbase classpath computation in util script. Adding external conf and scripts. Enable SPARK_HIVE mode while building. This is needed to bundle datanucleus jars needed for hive table creation. Build Spark on MapR. - make-distribution.sh takes an environment variable to enable profiles - MVN_PROFILE_ARG - Added warden conf files under ext-conf. - Updated pom.xml to use right set of jars and version. Spark Master failed to start in HA mode Updated Apache Curator version Added spark streaming integration with kafka 0.9 and mapr-streams Added MapR Repo
RolatZhang
added a commit
to RolatZhang/spark
that referenced
this pull request
Aug 15, 2022
AL-5546 upgrade commons-io,mesos,kafka-client,protobuf-java,gson,guava version
udaynpusa
pushed a commit
to mapr/spark
that referenced
this pull request
Jan 30, 2024
…cript K8S-1077 (apache#598) * K8S-1077 - use single k8s secret with user info MapR [SPARK-651] Replacing joda-time-*.jar with joda-time-2.10.3.jar. MapR [SPARK-638] Wrong permissions when creating files under directory with GID bit set. MapR [SPARK-627] SparkHistoryServer-2.4 is getting 403 Unauthorized home page for users(spark.ui.view.acls) via spark-submit MapR [SPARK-639] Default headers are adding two times MapR [SPARK-629] Spark UI for job lose CSS styles MapR [MS-925] After upgrade to MEP 6.2 (Spark 2.4.0) can no longer consume Kafka / MapR Streams. MapR [SPARK-626] Update kafka dependencies for Spark 2.4.4.0 in release MEP-6.3.0 MapR [SPARK-340] Jetty web server version at Spark should be updated tp v9.4.X MapR [SPARK-617] an't use ssl via spark beeline MapR [SPARK-617] Can't use ssl via spark beeline MapR [SPARK-620] Replace core dependency in Spark-2.4.4 MapR [SPARK-621] Fix multiple XML configuration initialization for (apache#575) custom headers. Use X-XSS-Protection, X-Content-Type-Options Content-Security-Policy and Strict-Transport-Security configuration only in case: cluster security is enabled OR spark.ui.security.headers.enabled set to true. MapR [SPARK-595] Spark cannot access hs2 through zookeeper Revert "MapR [SPARK-595] Spark cannot access hs2 through zookeeper (apache#577)" MapR [SPARK-595] Spark cannot access hs2 through zookeeper MapR [SPARK-620] Replace core dependency in Spark-2.4. MapR [SPARK-619] Move absent commits from 2.4.3 branch to 2.4.4 (apache#574) * Adding SQL API to write to kafka from Spark (apache#567) * Branch 2.4.3 extended kafka and examples (apache#569) * The v2 API is in its own package - the v2 api is in a different package - the old functionality is available in a separated package * v2 API examples - All the examples are using the newest API. - I have removed the old examples since they are not relevant any more and the same functionality is shown in the new examples usin the new API. * MapR [SPARK-619] Move absent commits from 2.4.3 branch to 2.4.4 CORE-321. Add custom http header support for jetty. MapR [SPARK-609] Port Apache Spark-2.4.4 changes to the MapR Spark-2.4.4 branch Adding multi table loader (apache#560) * Adding multi table loader - This allows us to load multiple matching tables into one Union DataFrame. If we have the fallowing MFS structure: ``` /clients/client_1/data.table /clients/client_2/data.table ``` we can load a union dataframe by doing `loadFromMapRDB("/clients/*/*.table")` * Fixing the path to the reader MapR [SPARK-588] Spark thriftserver fails when work with hive-maprdb json table MapR [SPARK-598] Spark can't add needed properties to hive-site.xml MAPR-SPARK-596: Change HBase compatible version for Spark 2.4.3 MapR [SPARK-592] Add possibility to use start-thriftserver.sh script with 2304 port MapR [SPARK-584] MaprDB connector's setHintUsingIndex method doesn't work as expected MapR [SPARK-583] MaprDB connector's loadFromMaprDB function for Java API doesn't work as expected SPARK-579 info about ssl_trustore is added for metrics MapR [SPARK-552] Failed to get broadcast_11_piece0 of broadcast_11 SPARK-569 Generation of SSL ceritificates for spark UI MapR [SPARK-575] Warning messages in spark workspace after the second attempt to login to job's UI Update zookeeper version Adding `joinWithMapRDBTable` function (apache#529) The related documentation of this function is here https://github.com/anicolaspp/MapRDBConnector#joinwithmaprdbtable. The main idea is that having a dataframe (no matter how was it constructed) we can join it with a MapR-DB table. This functions looks at the join query and load only those records from MapR-DB that will join instead of loading the full table and then join in memory. In other words, we only load what we know will be joint. Adding DataSource Reader Support (apache#525) * Adding DataSource Reader Support * Update SparkSessionExt.scala * creating a package object * Update MapRDBSpark.scala * fully path to avoid name collition * refactorings MapR [SPARK-451] Spark hadoop/core dependency updates MapR [SPARK-566] Move absent commits from 2.4.0 branch MapR [SPARK-561] Spark 2.4.3 porting to MapR MapR [SPARK-561] Spark 2.4.3 porting to MapR MapR [SPARK-558] Render application UI init page if driver is not up MapR [SPARK-541] Avoid duplication of the first unexpired record MapR [COLD-150][K8S] Fix metrics copy MapR [K8S-893] Hide plain text password from logs MapR [SPARK-540] Include 'avro' artifacts MapR [SPARK-536] PySpark streaming package for kafka-0-10 added K8S-853: Enable spark metrics for external tenant MapR [SPARK-531] Remove duplicating entries from classpath in ClasspathFilter MapR [SPARK-516] Spark jobs failure using yarn mode on kerberos fixed MapR [SPARK-462] Spark and SparkHistoryServer allow week ciphers, which can allow man in the middle attack [SPARK-508] MapR-DB OJAI Connector for Spark isNull condition returns incorrect result MapR [SPARK-510] nonmapr "admin" users not able to view other user logs in SHS SPARK-460: Spark Metrics for CollectD Configuration for collecting Spark metrics SPARK-463 MAPR_MAVEN_REPO variable for specifying mapR repository MapR [SPARK-492] Spark 2.4.0.0 configure.sh has error messages MapR [SPARK-515][K8S] Remove configure.sh call for k8s MapR [SPARK-515] Move configuring spark-env.sh back to the private-pkg MapR [SPARK-515] Move configuring spark-env.sh back to the private-pkg MapR [SPARK-514] Recovery from checkpoint is broken MapR [SPARK-445] Messages loss fixed by reverting [MAPR-32290] changes from kafka09 package (apache#460) * MapR [SPARK-445] Revert "[MAPR-32290] Spark processing offsets when messages are already TTL in the first batch (apache#376)" This reverts commit e8d59b9. * MapR [SPARK-445] Revert "[MAPR-32290] Spark processing offsets when messages are already ttl in first batch (apache#368)" This reverts commit b282a8b. MapR [SPARK-445] Messages loss fixed by reverting [MAPR-32290] changes from kafka10 package MapR [SPARK-469] Fix NPE in generated classes by reverting "[SPARK-23466][SQL] Remove redundant null checks in generated Java code by GenerateUnsafeProjection" (apache#455) This reverts commit c5583fd. MapR [SPARK-482] Spark streaming app fails to start by UnknownTopicOrPartitionException with checkpoint MapR [SPARK-496] Spark HS UI doesn't work MapR [SPARK-416] CVE-2018-1320 vulnerability in Apache Thrift MapR [SPARK-486][K8S] Fix sasl encryption error on Kubernetes MapR [SPARK-481] Cannot run spark configure.sh on Client node MapR [K8S-637][K8S] Add configure.sh configuration in spark-defaults.conf for job runtime MapR [SPARK-465] Error messages after update of spark 2.4 MapR [SPARK-465] Error messages after update of spark 2.4 MapR [SPARK-464] Can't submit spark 2.4 jobs from mapr-client [SPARK-466] SparkR errors fixed MapR [SPARK-456] Spark shell can't be started SPARK-417 impersonation fixes for spark executor. Impersonation is mo… (apache#433) * SPARK-417 impersonation fixes for spark executor. Impersonation is moved from HadoopRDD.compute() method to org.apache.spark.executor.Executor.run() method * SPARK-363 Hive version changed to '1.2.0-mapr-spark-MEP-6.0.0' [SPARK-449] Kafka offset commit issue fixed MapR [SPARK-287] Move logic of creating /apps/spark folder from installer's scripts to the configure.sh MapR [SPARK-221] Investigate possibility to move creating of the spark-env.sh from private-pkg to configure.sh MapR [SPARK-430] PID files should be under /opt/mapr/pid MapR [SPARK-446] Spark configure.sh doesn't start/stop Spark services MapR [SPARK-434] Move absent commits from 2.3.2 branch (apache#425) * MapR [SPARK-352] Spark shell fails with "NoClassDefFoundError: org/apache/hadoop/fs/FSDataInputStream" if java is not available in PATH * MapR [SPARK-350] Deprecate Spark Kafka-09 package * MapR [SPARK-326] Investigate possibility of writing Java example for the MapRDB OJAI connector * [SPARK-356] Merge mapr changes from kafka-09 package into the kafka-10 * SPARK-319 Fix for sparkR version check * MapR [SPARK-349] Update OJAI client to v3 for Spark MapR-DB JSON connector * MapR [SPARK-367] Move absent commits from 2.3.1 branch * MapR [SPARK-137] Analyze the warning during compilation of OJAI connector * MapR [SPARK-369] Spark 2.3.2 fails with error related to zookeeper * [MAPR-26258] hbasecontext.HBaseDistributedScanExample fails * [SPARK-24355] Spark external shuffle server improvement to better handle block fetch requests * MapR [SPARK-374] Spark Hive example fails when we submit job from another(simple) cluster user * MapR [SPARK-434] Move absent commits from 2.3.2 branch * MapR [SPARK-434] Move absent commits from 2.3.2 branch * MapR [SPARK-373] Unexpected behavior during job running in standalone cluster mode * MapR [SPARK-419] Update hive-maprdb-json-handler jar for spark 2.3.2.0 and spark 2.2.1 * MapR [SPARK-396] Interface change of sendToKafka * MapR [SPARK-357] consumer groups are prepeneded with a "service_" prefix * MapR [SPARK-429] Changes in maprdb connector are the cause of broken backward compatibility * MapR [SPARK-427] Update kafka in Spark-2.4.0 to the 1.1.1-mapr * MapR [SPARK-434] Move absent commits from 2.3.2 branch * Move absent commits from 2.3.2 branch * MapR [SPARK-434] Move absent commits from 2.3.2 branch * Move absent commits from 2.3.2 branch * Move absent commits from 2.3.2 branch MapR [SPARK-427] Update kafka in Spark-2.4.0 to the 1.1.1-mapr MapR [SPARK-379] Spark 2.4 4-gidit version MapR [PIC-48][K8S] Port k8s changes to 2.4.0 [PIC-48] Create user for k8s driver and executor if required [PIC-48] Create user for k8s driver and executor if required Revert "Remove spark.ui.filters property" This reverts commit d8941ba36c3451cdce15d18d6c1a52991de3b971. [SPARK-351] Copy kubernetes start scripts anyway PIC-34: Rename default configmap name to be consistent with mapr-kubernetes [SPARK-23668][K8S] Add config option for passing through k8s Pod.spec.imagePullSecrets (apache#355) Pass through the `imagePullSecrets` option to the k8s pod in order to allow user to access private image registries. See https://kubernetes.io/docs/tasks/configure-pod-container/pull-image-private-registry/ Unit tests + manual testing. Manual testing procedure: 1. Have private image registry. 2. Spark-submit application with no `spark.kubernetes.imagePullSecret` set. Do `kubectl describe pod ...`. See the error message: ``` Error syncing pod, skipping: failed to "StartContainer" for "spark-kubernetes-driver" with ErrImagePull: "rpc error: code = 2 desc = Error: Status 400 trying to pull repository ...: \"{\\n \\\"errors\\\" : [ {\\n \\\"status\\\" : 400,\\n \\\"message\\\" : \\\"Unsupported docker v1 repository request for '...'\\\"\\n } ]\\n}\"" ``` 3. Create secret `kubectl create secret docker-registry ...` 4. Spark-submit with `spark.kubernetes.imagePullSecret` set to the new secret. See that deployment was successful. Author: Andrew Korzhuev <[email protected]> Author: Andrew Korzhuev <[email protected]> Closes apache#20811 from andrusha/spark-23668-image-pull-secrets. [SPARK-321] Change default value of spark.mapr.ssl.secret.prefix property [PIC-32] Spark on k8s with MapR secure cluster Update entrypoint.sh with correct spark version (apache#340) This PR has minor fix to correct the spark version string [SPARK-274] Create home directory for user who submitted job [MAPR-SPARK-230] Implement security for Spark on Kubernetes Run Spark job with specify the username for driver and executor Read cluster configs from configMap Run configure.sh script form entrypoint.sh Remove spark.kubernetes.driver.pod.commands property Add Spark properties for executor and driver environment variable MapR [SPARK-296] Structured Streaming memory leak Revert "[MAPR-SPARK-210] Rename sprk-defaults.conf to spark-defaults.conf.tem…" (apache#252) * Revert "[MAPR-SPARK-176] Fix Spark Project Catalyst unit tests (apache#251)" This reverts commit 5de05075cd14abf8ac65046a57a5d76617818fbe. * Revert "[MAPR-SPARK-210] Rename sprk-defaults.conf to spark-defaults.conf.template (apache#249)" This reverts commit 1baa677d727e89db7c605ffbae9a9eba00337ad0. [MAPR-SPARK-210] Rename sprk-defaults.conf to spark-defaults.conf.template MapR [SPARK-379] Port Spark to 2.4.0 MapR [SPARK-341] Spark 2.3.2 porting [MAPR-32290] Spark processing offsets when messages are already TTL in the first batch * Bug 32263 - Seek called on unsubscribed partitions [MSPARK-331] Remove snapshot versions of mapr dependencies from Spark-2.3.1 [MAPR-32290] Spark processing offsets when messages are already ttl in first batch MapR [SPARK-325] Add examples for work with the MapRDB JSON connector into the Spark project [ATS-449] Unit test for EBF 32013 created. MAPR-SPARK-311: Spark beeline uses default ssl truststore instead of mapr ssl truststore Bug 32355 - Executor tab empty on Spark UI [SPARK-318] Submitting Spark jobs from Oozie fails due to ClassNotFoundException Bug 32014 - Spark Consumer fails with java.lang.AssertionError Revert "[SPARK-306] Kafka clients 1.0.1 present in jars directory for Spark 2.3.1" (apache#341) * Revert "[SPARK-306] Kafka clients 1.0.1 present in jars directory for Spark 2.3.1 (apache#335)" This reverts commit 832411e. Bug 32014 - Spark Consumer fails with java.lang.AssertionError (apache#326) (apache#336) * MapR [32014] Spark Consumer fails with java.lang.AssertionError [SPARK-306] Kafka clients 1.0.1 present in jars directory for Spark 2.3.1 DEVOPS-2768 temporarily removed curl for file downloading [SPARK-302] Local privilege escalation MapR [SPARK-297] Added unit test for empty value conversion MapR [SPARK-297] Empty values are loaded as non-null MapR [SPARK-296] Structured Streaming memory leak 2.3.1 spark 289 (apache#318) * MapR [SPARK-289] Fix unit test for Spark-2.3.1 [SPARK-130] MapRDB connector - NPE while saving Pair RDD with 'null' values MapR [SPARK-283] Unit tests fail during initialization SSL properties. [SPARK-212] SparkHiveExample fails when we run it twice MapR [SPARK-282] Remove maprfs and hadoop jars from mapr spark package MapR [SPARK-278] Spark submit fails for jobs with python MapR [SPARK-279] Can't connect to spark thrift server with new spark and hive packages MapR [SPARK-276] Update zookeeper dependency to v.3.4.11 for spark 2.3.1 MapR [SPARK-272] Use only client passwords from ssl-client.xml MapR [SPARK-266] Spark jobs can't finish correctly, when there is an error during job running MapR [SPARK-263] Add possibility to use keyPassword which is different from keyStorePassword [MSPARK-31632] RM UI showing broken page for Spark jobs MapR [SPARK-261] Use mapr-security-web for getting passwords. MapR [SPARK-259] Spark application doesn't finish correctly MapR [SPARK-268] Update Spark version for Warden change project version to 2.3.1-mapr-SNAPSHOT MapR [SPARK-256] Spark doesn't work on yarn mode MapR [SPARK-255] Installer fresh install 610/600 secure fails to start "mapr-spark-thriftserver", "mapr-spark-historyserver" Mapr [SPARK-248] MapRDBTableScanRDD fails to convert to Scala Dataframe when using where clause MapR [SPARK-225] Hadoop credentials provider usage for hiding passwords at spark-defaults MapR [SPARK-214] Hive-2.1 poperties can't be read from a hive-site.xml as Spark uses Hive-1.2 MapR [SPARK-216] Spark thriftserver fails when work with hive-maprdb json table SPARK-244 (apache#278) Provide ability to use MapR-Negotiation authentication for Spark HistoryServer MapR [SPARK-226] Spark - pySpark Security Vulnerability MapR [SPARK-220] SparkR fails with UDF functions bug fixed MapR [SPARK-227] KafkaUtils.createDirectStream fails with kafka-09 MapR [SPARK-183] Spark Integration for Kafka 0.10 unit tests disabled MapR [SPARK-182] Spark Project External Kafka Producer v09 unit tests fixed MapR [SPARK-179] Spark Integration for Kafka 0.9 unit tests fixed MapR [SPARK-181] Kafka 0.10 Structured Streaming unit tests fixed [MSPARK-31305] Spark History server NOT loading applications submitted by users other than 'mapr' MapR [SPARK-175] Fix Spark Project Streaming unit tests [MAPR-SPARK-176] Fix Spark Project Catalyst unit tests [MAPR-SPARK-178] Fix Spark Project Hive unit tests MapR [SPARK-174] Spark Core unit tests fixed Changed version for spark-kafka connector. MapR [SPARK-202] Update MapR Spark to 2.3.0 Fixed compile time errors in tests Change project version [SPARK-198] Update hadoop dependency version to 2.7.0-mapr-1803 for Spark 2.2.1 MapR [SPARK-188] Couldn't connect to thrift server via spark beeline on kerberos cluster MapR [SPARK-143] Spark History Server does not require login for secured-by-default clusters MapR [SPARK-186] Update OJAI versions to the latest for Spark-2.2.1 OJAI Connector MapR [SPARK-191] Incorrect work of MapR-DB Sink 'complete' and 'update' modes fixed MapR [SPARK-170] StackOverflowException in equals method in DBMapValue 2.2.1 build fixed (apache#231) * MapR [SPARK-164] Update Kafka version to 1.0.1-mapr in Spark Kafka Producer module MapR [SPARK-161] Include Kafka Structured streaming jar to Spark package. MapR [SPARK-155] Change Spark Master port from 8080 MapR [SPARK-153] Exception in spark job with configured labels on yarn-client mode MapR [SPARK-152] Incorrect date string parsing fixed MapR [SPARK-21] Structured Streaming MapR-DB Sink created MapR [SPARK-135] Spark 2.2 with MapR Streams ( Kafka 1.0) (apache#218) * MapR [SPARK-135] Spark 2.2 with MapR Streams (Kafka 1.0) Added functionality of MapR-Streams specific EOF handling. MapR [SPARK-143] Spark History Server does not require login for secured-by-default clusters Disable build failing if scalastyle checking is fall. MapR [SPARK-16] Change Spark version in Warden files and configure.sh MapR [SPARK-144] Add insertToMapRDB method for rdd for Java API [MAPR-30536] Spark SQL queries on Map column fails after upgrade MapR [SPARK-139] Remove "update" related APIs from connector MapR [SPARK-140] Change the option name "tableName" to "tablePath" in the Spark/MapR-DB connectors. MapR [SPARK-121] Spark OJAI JAVA: update functionality removed MapR [SPARK-118] Spark OJAI Python: missed DataFrame import while moving imports in order to fix MapR [ZEP-101] interpreter issue MapR [SPARK-118] Spark OJAI Python: move MapR DB Connector class importing in order to fix MapR [ZEP-101] interpreter issue MapR [SPARK-117] Spark OJAI Python: Save functionality implementation MapR [SPARK-131] Exception when try to save JSON table with Binary _id field Spark OJAI JAVA: load to RDD, save from RDD implementation (apache#195) * MapR [SPARK-124] Loading to JavaRDD implemented * MapR [SPARK-124] MapRDBJavaSparkContext constructor changed * MapR [SPARK-124] implemented RDD[Row] saving MapR [SPARK-118] Spark OJAI Python: Read implementation MapR [SPARK-128] MapRDB connector - wrong handle of null fields when nullable is false * MapR [SPARK-121] Spark OJAI JAVA: Read to Dataset functionality implementation * Minor refactoring MapR [SPARK-125] Default value of idFieldPath parameter is not handle MapR [SPARK-113] Hit java.lang.UnsupportedOperationException: empty.reduceLeft during loadFromMapRDB Spark Mapr-DB connector was refactored according to Scala style Removed code duplication [MSPARK-107]idField information is lost in MapRDBDataFrameWriterFunctions.saveToMapRDB configure.sh takes options to change ports Kafka client excluded from package because correct version is located in "mapr classpath" Changed Kafka version in Kafka producer module. Branch spark 69 (apache#170) * Fixing the wrong type casting of TimeStamp to OTimeStamp when read from spark dataFrame. * SPARK-69: Problem with license when we try to read from json and write to maprdb remove creatin /usr/local/spark link from configure.sh. This link will be creates by private-pkg remove include-maprdb from default profiles added profiles in maprdb pom file instead of two pom files Fixed maprdb connector dependencies. Fixing the wrong type casting of TimeStamp to OTimeStamp when read from spark dataFrame. changed port for spark-thriftserver as it conflicts with hive server changed port for spark-thriftserver as it conflicts with hive server remove .not_configured_yet file after success Ojai connector fixed required java version [MSPARK-45] Move Spark-OJAI connector code to Spark github repo (apache#132) * SPARK-45 Move Spark-OJAI connector code to Spark github repo * Fixing pom versions for maprdb spark connector. * Changes made to the connector code to be compatible with 5.2.* and 6.0 clients. Spark 2.1.0 mapr 29106 (apache#150) * [SPARK-20922][CORE] Add whitelist of classes that can be deserialized by the launcher. Blindly deserializing classes using Java serialization opens the code up to issues in other libraries, since just deserializing data from a stream may end up execution code (think readObject()). Since the launcher protocol is pretty self-contained, there's just a handful of classes it legitimately needs to deserialize, and they're in just two packages, so add a filter that throws errors if classes from any other package show up in the stream. This also maintains backwards compatibility (the updated launcher code can still communicate with the backend code in older Spark releases). Tested with new and existing unit tests. Author: Marcelo Vanzin <[email protected]> Closes apache#18166 from vanzin/SPARK-20922. (cherry picked from commit 8efc6e9) Signed-off-by: Marcelo Vanzin <[email protected]> (cherry picked from commit 772a9b9) * [SPARK-20922][CORE][HOTFIX] Don't use Java 8 lambdas in older branches. Author: Marcelo Vanzin <[email protected]> Closes apache#18178 from vanzin/SPARK-20922-hotfix. Added security by default for historyserver use waitForConsumerAssignment() instead of consumer.poll(0) for spark-29052 change MAPR_HADOOP_CLASSPATH in configure.sh for creating it by mapr-classpath.sh change MAPR_HADOOP_CLASSPATH in configure.sh for creating it by mapr-classpath.sh changes for mapr-classpath.sh changes for mapr-classpath.sh configure.sh changes [SPARK-39] Classpath filter was added Fixed impersonation when data read from MapR-DB via Spark-Hive. added configure.sh and warden.spark-thriftserver.conf hive-hbase-handler added to Spark jars Fixed "Single message comes late" 28339 bug fixed Spark streaming skipped message with zero offset from Kafka 0.9 [MSPARK-9] Initial fix for Spark unit tests Bump dependencies after ECO-1703 release [SPARK-33] Streaming example fixed [MAPR-26060] Fixed case when mapr-streams make gaps in offsets ported features from kafka 10 to kafka 9 [MAPR-26289][SPARK-2.1] Streaming general improvements (apache#93) * Added include-kafka-09 profile to Assembly * Set default poll timeout to 120s Set default HBase verison to 1.1.8 Changes from Kafka10 package were ported to Kafka09 package. [MAPR-26053] Include MapR Classes to the default value of spark.sql.hive.metastore.sharedPrefixes [MAPR-25807] Spark-Warehouse path computes incorrectly Add MapR-SASL support for Thrift Server Adding scala library. [MAPR-25713] Spark might try to load MapR Class Loader multiple times and fail [MAPR-25311] Bump Spark dependencies after ECO-1611 release [MINOR] Fix spark-jars.sh script [MAPR-24603] Could not launch beeline shell after starting spark thrift server fixed syntax error in V09DirectKafkaWordCount example Spark 2.0.1 MAPR-streams Python API [MAPR-24415] SPARK_JAVA_OPTS is deprecated Kafka streaming producer added. Minor fix for previous commit Added script for MAPR-24374 Some minor changes to spark-defaults.conf Changed default HBase version to 1.1.1 in compatibility.version Streaming example was refactored [MAPR-24470] HiveFromSpark test fails in yarn-cluster mode Added MapR Repo [MAPR-22940] Failed to connect spark beeline (after spark thrift server is started) on Kerberos cluster [MAPR-18865] Unable to submit spark apps from Windows client Skip maven clean task on the parent module New: Issue with running Hive commands in Spark This is fixed in SPARK-7819 Isolated Hive Client Loader appears to cause Native Library libMapRClient.4.0.2-mapr.so already loaded in another classloader error Spark warden.services.conf should have dependency on cldb Remove DFS shuffle settings. These settings are not used right now. Copy every file in the conf directory into the distribution package. Create spark-defaults.conf for MapR Settings to enable DFS shuffle on MapR. Support hbase classpath computation in util script. Adding external conf and scripts. Enable SPARK_HIVE mode while building. This is needed to bundle datanucleus jars needed for hive table creation. Build Spark on MapR. - make-distribution.sh takes an environment variable to enable profiles - MVN_PROFILE_ARG - Added warden conf files under ext-conf. - Updated pom.xml to use right set of jars and version. Spark Master failed to start in HA mode Updated Apache Curator version Added spark streaming integration with kafka 0.9 and mapr-streams Added MapR Repo
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
The current implementation of ClientBase.getDefaultYarnApplicationClasspath inspects
the MRJobConfig class for the field DEFAULT_YARN_APPLICATION_CLASSPATH when it should
be really looking into YarnConfiguration. If the Application Configuration has no
yarn.application.classpath defined a NPE exception will be thrown.
Additional Changes include:
[ticket: SPARK-1522] : https://issues.apache.org/jira/browse/SPARK-1522
Author : [email protected]
Testing : SPARK_HADOOP_VERSION=2.3.0 SPARK_YARN=true ./sbt/sbt test