[jira] [Commented] (HIVE-18301) Investigate to enable MapInput cache in Hive on Spark
[
https://issues.apache.org/jira/browse/HIVE-18301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16350077#comment-16350077
]
liyunzhang commented on HIVE-18301:
---
[~xuefuz]:
{quote}
Instead of putting the input patch in each row, like what the patch is
proposing, could we send a serialized IOContext object as a special row
whenever the content of the object changes?
{quote}
This is good idea. IOContext is bind to split not to bind to each record. So
I changed to following idea.
{code}
inputRDD1inputRDD2
|CopyFunction| CopyFunction
CopyRDD1CopyRDD2
| |
MT_11 MT_12
| |
RT_1 RT_2
\ /
Union
{code}
in the CopyRDD1, I only store IOContext as the tuple_1() of result when the
IOContext#getInputPath is changed and store null in other situation. Thus it
will reduce the size of data size increment of this solution. In MT_12, it
initialize the IOContext when IOContext#getInputPath is null, once
IOContext#getInputPath has value, we need not initialize it again in the same
thread.
> Investigate to enable MapInput cache in Hive on Spark
> -
>
> Key: HIVE-18301
> URL: https://issues.apache.org/jira/browse/HIVE-18301
> Project: Hive
> Issue Type: Bug
>Reporter: liyunzhang
>Assignee: liyunzhang
>Priority: Major
> Attachments: HIVE-18301.1.patch, HIVE-18301.2.patch, HIVE-18301.patch
>
>
> Before IOContext problem is found in MapTran when spark rdd cache is enabled
> in HIVE-8920.
> so we disabled rdd cache in MapTran at
> [SparkPlanGenerator|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/spark/SparkPlanGenerator.java#L202].
> The problem is IOContext seems not initialized correctly in the spark yarn
> client/cluster mode and caused the exception like
> {code}
> Job aborted due to stage failure: Task 93 in stage 0.0 failed 4 times, most
> recent failure: Lost task 93.3 in stage 0.0 (TID 616, bdpe48):
> java.lang.RuntimeException: Error processing row:
> java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:165)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:48)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:27)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveBaseFunctionResultList.hasNext(HiveBaseFunctionResultList.java:85)
> at
> scala.collection.convert.Wrappers$JIteratorWrapper.hasNext(Wrappers.scala:42)
> at
> org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.write(BypassMergeSortShuffleWriter.java:125)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:79)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:47)
> at org.apache.spark.scheduler.Task.run(Task.scala:85)
> at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:274)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:745)
> Caused by: java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.AbstractMapOperator.getNominalPath(AbstractMapOperator.java:101)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.cleanUpInputFileChangedOp(MapOperator.java:516)
> at
> org.apache.hadoop.hive.ql.exec.Operator.cleanUpInputFileChanged(Operator.java:1187)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.process(MapOperator.java:546)
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:152)
> ... 12 more
> Driver stacktrace:
> {code}
> in yarn client/cluster mode, sometimes
> [ExecMapperContext#currentInputPath|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/mr/ExecMapperContext.java#L109]
> is null when rdd cach is enabled.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
[jira] [Commented] (HIVE-18301) Investigate to enable MapInput cache in Hive on Spark
[
https://issues.apache.org/jira/browse/HIVE-18301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16348429#comment-16348429
]
Hive QA commented on HIVE-18301:
Here are the results of testing the latest attachment:
https://issues.apache.org/jira/secure/attachment/12908669/HIVE-18301.2.patch
{color:red}ERROR:{color} -1 due to no test(s) being added or modified.
{color:red}ERROR:{color} -1 due to 21 failed/errored test(s), 12965 tests
executed
*Failed tests:*
{noformat}
org.apache.hadoop.hive.cli.TestAccumuloCliDriver.testCliDriver[accumulo_queries]
(batchId=240)
org.apache.hadoop.hive.cli.TestCliDriver.testCliDriver[auto_sortmerge_join_2]
(batchId=49)
org.apache.hadoop.hive.cli.TestCliDriver.testCliDriver[mapjoin_hook]
(batchId=13)
org.apache.hadoop.hive.cli.TestCliDriver.testCliDriver[ppd_join5] (batchId=36)
org.apache.hadoop.hive.cli.TestEncryptedHDFSCliDriver.testCliDriver[encryption_move_tbl]
(batchId=175)
org.apache.hadoop.hive.cli.TestMiniLlapCliDriver.testCliDriver[llap_smb]
(batchId=152)
org.apache.hadoop.hive.cli.TestMiniLlapLocalCliDriver.testCliDriver[bucket_map_join_tez1]
(batchId=172)
org.apache.hadoop.hive.cli.TestMiniLlapLocalCliDriver.testCliDriver[insert_values_orig_table_use_metadata]
(batchId=167)
org.apache.hadoop.hive.cli.TestMiniLlapLocalCliDriver.testCliDriver[llap_acid]
(batchId=171)
org.apache.hadoop.hive.cli.TestMiniLlapLocalCliDriver.testCliDriver[llap_acid_fast]
(batchId=161)
org.apache.hadoop.hive.cli.TestMiniLlapLocalCliDriver.testCliDriver[resourceplan]
(batchId=164)
org.apache.hadoop.hive.cli.TestMiniLlapLocalCliDriver.testCliDriver[sysdb]
(batchId=161)
org.apache.hadoop.hive.cli.TestMiniLlapLocalCliDriver.testCliDriver[vectorization_input_format_excludes]
(batchId=163)
org.apache.hadoop.hive.cli.TestSparkCliDriver.testCliDriver[ppd_join5]
(batchId=122)
org.apache.hadoop.hive.cli.control.TestDanglingQOuts.checkDanglingQOut
(batchId=221)
org.apache.hadoop.hive.ql.exec.TestOperators.testNoConditionalTaskSizeForLlap
(batchId=282)
org.apache.hadoop.hive.ql.io.TestDruidRecordWriter.testWrite (batchId=256)
org.apache.hive.beeline.cli.TestHiveCli.testNoErrorDB (batchId=188)
org.apache.hive.jdbc.TestSSL.testConnectionMismatch (batchId=234)
org.apache.hive.jdbc.TestSSL.testConnectionWrongCertCN (batchId=234)
org.apache.hive.jdbc.TestSSL.testMetastoreConnectionWrongCertCN (batchId=234)
{noformat}
Test results: https://builds.apache.org/job/PreCommit-HIVE-Build/8965/testReport
Console output: https://builds.apache.org/job/PreCommit-HIVE-Build/8965/console
Test logs: http://104.198.109.242/logs/PreCommit-HIVE-Build-8965/
Messages:
{noformat}
Executing org.apache.hive.ptest.execution.TestCheckPhase
Executing org.apache.hive.ptest.execution.PrepPhase
Executing org.apache.hive.ptest.execution.YetusPhase
Executing org.apache.hive.ptest.execution.ExecutionPhase
Executing org.apache.hive.ptest.execution.ReportingPhase
Tests exited with: TestsFailedException: 21 tests failed
{noformat}
This message is automatically generated.
ATTACHMENT ID: 12908669 - PreCommit-HIVE-Build
> Investigate to enable MapInput cache in Hive on Spark
> -
>
> Key: HIVE-18301
> URL: https://issues.apache.org/jira/browse/HIVE-18301
> Project: Hive
> Issue Type: Bug
>Reporter: liyunzhang
>Assignee: liyunzhang
>Priority: Major
> Attachments: HIVE-18301.1.patch, HIVE-18301.2.patch, HIVE-18301.patch
>
>
> Before IOContext problem is found in MapTran when spark rdd cache is enabled
> in HIVE-8920.
> so we disabled rdd cache in MapTran at
> [SparkPlanGenerator|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/spark/SparkPlanGenerator.java#L202].
> The problem is IOContext seems not initialized correctly in the spark yarn
> client/cluster mode and caused the exception like
> {code}
> Job aborted due to stage failure: Task 93 in stage 0.0 failed 4 times, most
> recent failure: Lost task 93.3 in stage 0.0 (TID 616, bdpe48):
> java.lang.RuntimeException: Error processing row:
> java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:165)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:48)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:27)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveBaseFunctionResultList.hasNext(HiveBaseFunctionResultList.java:85)
> at
> scala.collection.convert.Wrappers$JIteratorWrapper.hasNext(Wrappers.scala:42)
> at
> org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.write(BypassMergeSortShuffleWriter.java:125)
> at
>
[jira] [Commented] (HIVE-18301) Investigate to enable MapInput cache in Hive on Spark
[
https://issues.apache.org/jira/browse/HIVE-18301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16348349#comment-16348349
]
Hive QA commented on HIVE-18301:
| (x) *{color:red}-1 overall{color}* |
\\
\\
|| Vote || Subsystem || Runtime || Comment ||
|| || || || {color:brown} Prechecks {color} ||
| {color:blue}0{color} | {color:blue} findbugs {color} | {color:blue} 0m
0s{color} | {color:blue} Findbugs executables are not available. {color} |
| {color:green}+1{color} | {color:green} @author {color} | {color:green} 0m
1s{color} | {color:green} The patch does not contain any @author tags. {color} |
|| || || || {color:brown} master Compile Tests {color} ||
| {color:blue}0{color} | {color:blue} mvndep {color} | {color:blue} 1m
23s{color} | {color:blue} Maven dependency ordering for branch {color} |
| {color:green}+1{color} | {color:green} mvninstall {color} | {color:green} 5m
46s{color} | {color:green} master passed {color} |
| {color:green}+1{color} | {color:green} compile {color} | {color:green} 1m
15s{color} | {color:green} master passed {color} |
| {color:green}+1{color} | {color:green} checkstyle {color} | {color:green} 0m
54s{color} | {color:green} master passed {color} |
| {color:green}+1{color} | {color:green} javadoc {color} | {color:green} 1m
10s{color} | {color:green} master passed {color} |
|| || || || {color:brown} Patch Compile Tests {color} ||
| {color:blue}0{color} | {color:blue} mvndep {color} | {color:blue} 0m
24s{color} | {color:blue} Maven dependency ordering for patch {color} |
| {color:green}+1{color} | {color:green} mvninstall {color} | {color:green} 1m
48s{color} | {color:green} the patch passed {color} |
| {color:green}+1{color} | {color:green} compile {color} | {color:green} 1m
18s{color} | {color:green} the patch passed {color} |
| {color:green}+1{color} | {color:green} javac {color} | {color:green} 1m
18s{color} | {color:green} the patch passed {color} |
| {color:red}-1{color} | {color:red} checkstyle {color} | {color:red} 0m
40s{color} | {color:red} ql: The patch generated 51 new + 91 unchanged - 2
fixed = 142 total (was 93) {color} |
| {color:green}+1{color} | {color:green} whitespace {color} | {color:green} 0m
0s{color} | {color:green} The patch has no whitespace issues. {color} |
| {color:green}+1{color} | {color:green} javadoc {color} | {color:green} 1m
6s{color} | {color:green} the patch passed {color} |
|| || || || {color:brown} Other Tests {color} ||
| {color:red}-1{color} | {color:red} asflicense {color} | {color:red} 0m
13s{color} | {color:red} The patch generated 1 ASF License warnings. {color} |
| {color:black}{color} | {color:black} {color} | {color:black} 16m 40s{color} |
{color:black} {color} |
\\
\\
|| Subsystem || Report/Notes ||
| Optional Tests | asflicense javac javadoc findbugs checkstyle compile |
| uname | Linux hiveptest-server-upstream 3.16.0-4-amd64 #1 SMP Debian
3.16.36-1+deb8u1 (2016-09-03) x86_64 GNU/Linux |
| Build tool | maven |
| Personality | /data/hiveptest/working/yetus/dev-support/hive-personality.sh |
| git revision | master / 419593e |
| Default Java | 1.8.0_111 |
| checkstyle |
http://104.198.109.242/logs//PreCommit-HIVE-Build-8965/yetus/diff-checkstyle-ql.txt
|
| asflicense |
http://104.198.109.242/logs//PreCommit-HIVE-Build-8965/yetus/patch-asflicense-problems.txt
|
| modules | C: common ql itests U: . |
| Console output |
http://104.198.109.242/logs//PreCommit-HIVE-Build-8965/yetus.txt |
| Powered by | Apache Yetushttp://yetus.apache.org |
This message was automatically generated.
> Investigate to enable MapInput cache in Hive on Spark
> -
>
> Key: HIVE-18301
> URL: https://issues.apache.org/jira/browse/HIVE-18301
> Project: Hive
> Issue Type: Bug
>Reporter: liyunzhang
>Assignee: liyunzhang
>Priority: Major
> Attachments: HIVE-18301.1.patch, HIVE-18301.2.patch, HIVE-18301.patch
>
>
> Before IOContext problem is found in MapTran when spark rdd cache is enabled
> in HIVE-8920.
> so we disabled rdd cache in MapTran at
> [SparkPlanGenerator|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/spark/SparkPlanGenerator.java#L202].
> The problem is IOContext seems not initialized correctly in the spark yarn
> client/cluster mode and caused the exception like
> {code}
> Job aborted due to stage failure: Task 93 in stage 0.0 failed 4 times, most
> recent failure: Lost task 93.3 in stage 0.0 (TID 616, bdpe48):
> java.lang.RuntimeException: Error processing row:
> java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:165)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:48)
>
[jira] [Commented] (HIVE-18301) Investigate to enable MapInput cache in Hive on Spark
[
https://issues.apache.org/jira/browse/HIVE-18301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16347253#comment-16347253
]
Xuefu Zhang commented on HIVE-18301:
It seems that IOContext is used in many places and the logics complicated.
Instead of putting the input patch in each row, like what the patch is
proposing, could we send a serialized IOContext object as a special row
whenever the content of the object changes? I'm not sure how feasible it's, so
it's just a rough idea to be explored.
> Investigate to enable MapInput cache in Hive on Spark
> -
>
> Key: HIVE-18301
> URL: https://issues.apache.org/jira/browse/HIVE-18301
> Project: Hive
> Issue Type: Bug
>Reporter: liyunzhang
>Assignee: liyunzhang
>Priority: Major
> Attachments: HIVE-18301.1.patch, HIVE-18301.patch
>
>
> Before IOContext problem is found in MapTran when spark rdd cache is enabled
> in HIVE-8920.
> so we disabled rdd cache in MapTran at
> [SparkPlanGenerator|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/spark/SparkPlanGenerator.java#L202].
> The problem is IOContext seems not initialized correctly in the spark yarn
> client/cluster mode and caused the exception like
> {code}
> Job aborted due to stage failure: Task 93 in stage 0.0 failed 4 times, most
> recent failure: Lost task 93.3 in stage 0.0 (TID 616, bdpe48):
> java.lang.RuntimeException: Error processing row:
> java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:165)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:48)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:27)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveBaseFunctionResultList.hasNext(HiveBaseFunctionResultList.java:85)
> at
> scala.collection.convert.Wrappers$JIteratorWrapper.hasNext(Wrappers.scala:42)
> at
> org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.write(BypassMergeSortShuffleWriter.java:125)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:79)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:47)
> at org.apache.spark.scheduler.Task.run(Task.scala:85)
> at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:274)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:745)
> Caused by: java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.AbstractMapOperator.getNominalPath(AbstractMapOperator.java:101)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.cleanUpInputFileChangedOp(MapOperator.java:516)
> at
> org.apache.hadoop.hive.ql.exec.Operator.cleanUpInputFileChanged(Operator.java:1187)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.process(MapOperator.java:546)
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:152)
> ... 12 more
> Driver stacktrace:
> {code}
> in yarn client/cluster mode, sometimes
> [ExecMapperContext#currentInputPath|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/mr/ExecMapperContext.java#L109]
> is null when rdd cach is enabled.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
[jira] [Commented] (HIVE-18301) Investigate to enable MapInput cache in Hive on Spark
[
https://issues.apache.org/jira/browse/HIVE-18301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16346418#comment-16346418
]
liyunzhang commented on HIVE-18301:
---
[~lirui]: thanks for question.
{quote}
The IOContext has quite a few other fields. I wonder whether they are available
if the RDD is cached.q
{quote}
Yes, IOContext#other fields are called in HiveContextAwareRecordReader and
other places. We can skip fields which is only used in
HiveContextAwareRecordReader, like IOContext#nextBlockStart,
IOContext#isBlockPointer. Others like IOContext#isBinarySearching is also used
in FilterOperator#process. Can not skip to cache it.
> Investigate to enable MapInput cache in Hive on Spark
> -
>
> Key: HIVE-18301
> URL: https://issues.apache.org/jira/browse/HIVE-18301
> Project: Hive
> Issue Type: Bug
>Reporter: liyunzhang
>Assignee: liyunzhang
>Priority: Major
> Attachments: HIVE-18301.1.patch, HIVE-18301.patch
>
>
> Before IOContext problem is found in MapTran when spark rdd cache is enabled
> in HIVE-8920.
> so we disabled rdd cache in MapTran at
> [SparkPlanGenerator|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/spark/SparkPlanGenerator.java#L202].
> The problem is IOContext seems not initialized correctly in the spark yarn
> client/cluster mode and caused the exception like
> {code}
> Job aborted due to stage failure: Task 93 in stage 0.0 failed 4 times, most
> recent failure: Lost task 93.3 in stage 0.0 (TID 616, bdpe48):
> java.lang.RuntimeException: Error processing row:
> java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:165)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:48)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:27)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveBaseFunctionResultList.hasNext(HiveBaseFunctionResultList.java:85)
> at
> scala.collection.convert.Wrappers$JIteratorWrapper.hasNext(Wrappers.scala:42)
> at
> org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.write(BypassMergeSortShuffleWriter.java:125)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:79)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:47)
> at org.apache.spark.scheduler.Task.run(Task.scala:85)
> at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:274)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:745)
> Caused by: java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.AbstractMapOperator.getNominalPath(AbstractMapOperator.java:101)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.cleanUpInputFileChangedOp(MapOperator.java:516)
> at
> org.apache.hadoop.hive.ql.exec.Operator.cleanUpInputFileChanged(Operator.java:1187)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.process(MapOperator.java:546)
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:152)
> ... 12 more
> Driver stacktrace:
> {code}
> in yarn client/cluster mode, sometimes
> [ExecMapperContext#currentInputPath|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/mr/ExecMapperContext.java#L109]
> is null when rdd cach is enabled.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
[jira] [Commented] (HIVE-18301) Investigate to enable MapInput cache in Hive on Spark
[
https://issues.apache.org/jira/browse/HIVE-18301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16346196#comment-16346196
]
Rui Li commented on HIVE-18301:
---
Hi [~kellyzly], is the input path the only thing we need to store with cached
RDD? The IOContext has quite a few other fields. I wonder whether they are
available if the RDD is cached.
> Investigate to enable MapInput cache in Hive on Spark
> -
>
> Key: HIVE-18301
> URL: https://issues.apache.org/jira/browse/HIVE-18301
> Project: Hive
> Issue Type: Bug
>Reporter: liyunzhang
>Assignee: liyunzhang
>Priority: Major
> Attachments: HIVE-18301.1.patch, HIVE-18301.patch
>
>
> Before IOContext problem is found in MapTran when spark rdd cache is enabled
> in HIVE-8920.
> so we disabled rdd cache in MapTran at
> [SparkPlanGenerator|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/spark/SparkPlanGenerator.java#L202].
> The problem is IOContext seems not initialized correctly in the spark yarn
> client/cluster mode and caused the exception like
> {code}
> Job aborted due to stage failure: Task 93 in stage 0.0 failed 4 times, most
> recent failure: Lost task 93.3 in stage 0.0 (TID 616, bdpe48):
> java.lang.RuntimeException: Error processing row:
> java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:165)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:48)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:27)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveBaseFunctionResultList.hasNext(HiveBaseFunctionResultList.java:85)
> at
> scala.collection.convert.Wrappers$JIteratorWrapper.hasNext(Wrappers.scala:42)
> at
> org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.write(BypassMergeSortShuffleWriter.java:125)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:79)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:47)
> at org.apache.spark.scheduler.Task.run(Task.scala:85)
> at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:274)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:745)
> Caused by: java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.AbstractMapOperator.getNominalPath(AbstractMapOperator.java:101)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.cleanUpInputFileChangedOp(MapOperator.java:516)
> at
> org.apache.hadoop.hive.ql.exec.Operator.cleanUpInputFileChanged(Operator.java:1187)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.process(MapOperator.java:546)
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:152)
> ... 12 more
> Driver stacktrace:
> {code}
> in yarn client/cluster mode, sometimes
> [ExecMapperContext#currentInputPath|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/mr/ExecMapperContext.java#L109]
> is null when rdd cach is enabled.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
[jira] [Commented] (HIVE-18301) Investigate to enable MapInput cache in Hive on Spark
[
https://issues.apache.org/jira/browse/HIVE-18301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16342225#comment-16342225
]
Hive QA commented on HIVE-18301:
Here are the results of testing the latest attachment:
https://issues.apache.org/jira/secure/attachment/12907839/HIVE-18301.1.patch
{color:green}SUCCESS:{color} +1 due to 1 test(s) being added or modified.
{color:red}ERROR:{color} -1 due to 21 failed/errored test(s), 11839 tests
executed
*Failed tests:*
{noformat}
org.apache.hadoop.hive.cli.TestAccumuloCliDriver.testCliDriver[accumulo_queries]
(batchId=240)
org.apache.hadoop.hive.cli.TestCliDriver.testCliDriver[mapjoin_hook]
(batchId=13)
org.apache.hadoop.hive.cli.TestCliDriver.testCliDriver[ppd_join5] (batchId=36)
org.apache.hadoop.hive.cli.TestCliDriver.testCliDriver[row__id] (batchId=78)
org.apache.hadoop.hive.cli.TestEncryptedHDFSCliDriver.testCliDriver[encryption_move_tbl]
(batchId=175)
org.apache.hadoop.hive.cli.TestMiniLlapLocalCliDriver.testCliDriver[bucket_map_join_tez1]
(batchId=172)
org.apache.hadoop.hive.cli.TestMiniLlapLocalCliDriver.testCliDriver[insert_values_orig_table_use_metadata]
(batchId=167)
org.apache.hadoop.hive.cli.TestMiniLlapLocalCliDriver.testCliDriver[llap_acid]
(batchId=171)
org.apache.hadoop.hive.cli.TestMiniLlapLocalCliDriver.testCliDriver[llap_acid_fast]
(batchId=161)
org.apache.hadoop.hive.cli.TestMiniLlapLocalCliDriver.testCliDriver[mergejoin]
(batchId=166)
org.apache.hadoop.hive.cli.TestMiniLlapLocalCliDriver.testCliDriver[resourceplan]
(batchId=164)
org.apache.hadoop.hive.cli.TestMiniLlapLocalCliDriver.testCliDriver[sysdb]
(batchId=161)
org.apache.hadoop.hive.cli.TestMiniLlapLocalCliDriver.testCliDriver[vectorization_input_format_excludes]
(batchId=163)
org.apache.hadoop.hive.cli.TestSparkCliDriver.testCliDriver[ppd_join5]
(batchId=122)
org.apache.hadoop.hive.cli.TestSparkPerfCliDriver.testCliDriver[query39]
(batchId=250)
org.apache.hadoop.hive.ql.exec.TestOperators.testNoConditionalTaskSizeForLlap
(batchId=282)
org.apache.hadoop.hive.ql.io.TestDruidRecordWriter.testWrite (batchId=256)
org.apache.hive.beeline.cli.TestHiveCli.testNoErrorDB (batchId=188)
org.apache.hive.jdbc.TestSSL.testConnectionMismatch (batchId=234)
org.apache.hive.jdbc.TestSSL.testConnectionWrongCertCN (batchId=234)
org.apache.hive.jdbc.TestSSL.testMetastoreConnectionWrongCertCN (batchId=234)
{noformat}
Test results: https://builds.apache.org/job/PreCommit-HIVE-Build/8872/testReport
Console output: https://builds.apache.org/job/PreCommit-HIVE-Build/8872/console
Test logs: http://104.198.109.242/logs/PreCommit-HIVE-Build-8872/
Messages:
{noformat}
Executing org.apache.hive.ptest.execution.TestCheckPhase
Executing org.apache.hive.ptest.execution.PrepPhase
Executing org.apache.hive.ptest.execution.YetusPhase
Executing org.apache.hive.ptest.execution.ExecutionPhase
Executing org.apache.hive.ptest.execution.ReportingPhase
Tests exited with: TestsFailedException: 21 tests failed
{noformat}
This message is automatically generated.
ATTACHMENT ID: 12907839 - PreCommit-HIVE-Build
> Investigate to enable MapInput cache in Hive on Spark
> -
>
> Key: HIVE-18301
> URL: https://issues.apache.org/jira/browse/HIVE-18301
> Project: Hive
> Issue Type: Bug
>Reporter: liyunzhang
>Assignee: liyunzhang
>Priority: Major
> Attachments: HIVE-18301.1.patch, HIVE-18301.patch
>
>
> Before IOContext problem is found in MapTran when spark rdd cache is enabled
> in HIVE-8920.
> so we disabled rdd cache in MapTran at
> [SparkPlanGenerator|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/spark/SparkPlanGenerator.java#L202].
> The problem is IOContext seems not initialized correctly in the spark yarn
> client/cluster mode and caused the exception like
> {code}
> Job aborted due to stage failure: Task 93 in stage 0.0 failed 4 times, most
> recent failure: Lost task 93.3 in stage 0.0 (TID 616, bdpe48):
> java.lang.RuntimeException: Error processing row:
> java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:165)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:48)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:27)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveBaseFunctionResultList.hasNext(HiveBaseFunctionResultList.java:85)
> at
> scala.collection.convert.Wrappers$JIteratorWrapper.hasNext(Wrappers.scala:42)
> at
> org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.write(BypassMergeSortShuffleWriter.java:125)
> at
>
[jira] [Commented] (HIVE-18301) Investigate to enable MapInput cache in Hive on Spark
[
https://issues.apache.org/jira/browse/HIVE-18301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16342203#comment-16342203
]
Hive QA commented on HIVE-18301:
| (x) *{color:red}-1 overall{color}* |
\\
\\
|| Vote || Subsystem || Runtime || Comment ||
|| || || || {color:brown} Prechecks {color} ||
| {color:blue}0{color} | {color:blue} findbugs {color} | {color:blue} 0m
0s{color} | {color:blue} Findbugs executables are not available. {color} |
| {color:green}+1{color} | {color:green} @author {color} | {color:green} 0m
0s{color} | {color:green} The patch does not contain any @author tags. {color} |
|| || || || {color:brown} master Compile Tests {color} ||
| {color:blue}0{color} | {color:blue} mvndep {color} | {color:blue} 1m
15s{color} | {color:blue} Maven dependency ordering for branch {color} |
| {color:green}+1{color} | {color:green} mvninstall {color} | {color:green} 5m
15s{color} | {color:green} master passed {color} |
| {color:green}+1{color} | {color:green} compile {color} | {color:green} 1m
7s{color} | {color:green} master passed {color} |
| {color:green}+1{color} | {color:green} checkstyle {color} | {color:green} 0m
50s{color} | {color:green} master passed {color} |
| {color:green}+1{color} | {color:green} javadoc {color} | {color:green} 1m
3s{color} | {color:green} master passed {color} |
|| || || || {color:brown} Patch Compile Tests {color} ||
| {color:blue}0{color} | {color:blue} mvndep {color} | {color:blue} 0m
22s{color} | {color:blue} Maven dependency ordering for patch {color} |
| {color:green}+1{color} | {color:green} mvninstall {color} | {color:green} 1m
29s{color} | {color:green} the patch passed {color} |
| {color:green}+1{color} | {color:green} compile {color} | {color:green} 1m
7s{color} | {color:green} the patch passed {color} |
| {color:green}+1{color} | {color:green} javac {color} | {color:green} 1m
7s{color} | {color:green} the patch passed {color} |
| {color:red}-1{color} | {color:red} checkstyle {color} | {color:red} 0m
34s{color} | {color:red} ql: The patch generated 49 new + 76 unchanged - 2
fixed = 125 total (was 78) {color} |
| {color:green}+1{color} | {color:green} whitespace {color} | {color:green} 0m
0s{color} | {color:green} The patch has no whitespace issues. {color} |
| {color:green}+1{color} | {color:green} javadoc {color} | {color:green} 1m
1s{color} | {color:green} the patch passed {color} |
|| || || || {color:brown} Other Tests {color} ||
| {color:red}-1{color} | {color:red} asflicense {color} | {color:red} 0m
11s{color} | {color:red} The patch generated 7 ASF License warnings. {color} |
| {color:black}{color} | {color:black} {color} | {color:black} 14m 52s{color} |
{color:black} {color} |
\\
\\
|| Subsystem || Report/Notes ||
| Optional Tests | asflicense javac javadoc findbugs checkstyle compile |
| uname | Linux hiveptest-server-upstream 3.16.0-4-amd64 #1 SMP Debian
3.16.36-1+deb8u1 (2016-09-03) x86_64 GNU/Linux |
| Build tool | maven |
| Personality | /data/hiveptest/working/yetus/dev-support/hive-personality.sh |
| git revision | master / cebe012 |
| Default Java | 1.8.0_111 |
| checkstyle |
http://104.198.109.242/logs//PreCommit-HIVE-Build-8872/yetus/diff-checkstyle-ql.txt
|
| asflicense |
http://104.198.109.242/logs//PreCommit-HIVE-Build-8872/yetus/patch-asflicense-problems.txt
|
| modules | C: common ql itests U: . |
| Console output |
http://104.198.109.242/logs//PreCommit-HIVE-Build-8872/yetus.txt |
| Powered by | Apache Yetushttp://yetus.apache.org |
This message was automatically generated.
> Investigate to enable MapInput cache in Hive on Spark
> -
>
> Key: HIVE-18301
> URL: https://issues.apache.org/jira/browse/HIVE-18301
> Project: Hive
> Issue Type: Bug
>Reporter: liyunzhang
>Assignee: liyunzhang
>Priority: Major
> Attachments: HIVE-18301.1.patch, HIVE-18301.patch
>
>
> Before IOContext problem is found in MapTran when spark rdd cache is enabled
> in HIVE-8920.
> so we disabled rdd cache in MapTran at
> [SparkPlanGenerator|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/spark/SparkPlanGenerator.java#L202].
> The problem is IOContext seems not initialized correctly in the spark yarn
> client/cluster mode and caused the exception like
> {code}
> Job aborted due to stage failure: Task 93 in stage 0.0 failed 4 times, most
> recent failure: Lost task 93.3 in stage 0.0 (TID 616, bdpe48):
> java.lang.RuntimeException: Error processing row:
> java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:165)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:48)
> at
>
[jira] [Commented] (HIVE-18301) Investigate to enable MapInput cache in Hive on Spark
[
https://issues.apache.org/jira/browse/HIVE-18301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16342195#comment-16342195
]
Hive QA commented on HIVE-18301:
Here are the results of testing the latest attachment:
https://issues.apache.org/jira/secure/attachment/12907839/HIVE-18301.1.patch
{color:green}SUCCESS:{color} +1 due to 1 test(s) being added or modified.
{color:red}ERROR:{color} -1 due to 21 failed/errored test(s), 11839 tests
executed
*Failed tests:*
{noformat}
org.apache.hadoop.hive.cli.TestAccumuloCliDriver.testCliDriver[accumulo_queries]
(batchId=240)
org.apache.hadoop.hive.cli.TestCliDriver.testCliDriver[mapjoin_hook]
(batchId=13)
org.apache.hadoop.hive.cli.TestCliDriver.testCliDriver[ppd_join5] (batchId=36)
org.apache.hadoop.hive.cli.TestCliDriver.testCliDriver[row__id] (batchId=78)
org.apache.hadoop.hive.cli.TestEncryptedHDFSCliDriver.testCliDriver[encryption_move_tbl]
(batchId=175)
org.apache.hadoop.hive.cli.TestMiniLlapCliDriver.testCliDriver[llap_smb]
(batchId=152)
org.apache.hadoop.hive.cli.TestMiniLlapLocalCliDriver.testCliDriver[bucket_map_join_tez1]
(batchId=172)
org.apache.hadoop.hive.cli.TestMiniLlapLocalCliDriver.testCliDriver[insert_values_orig_table_use_metadata]
(batchId=167)
org.apache.hadoop.hive.cli.TestMiniLlapLocalCliDriver.testCliDriver[llap_acid]
(batchId=171)
org.apache.hadoop.hive.cli.TestMiniLlapLocalCliDriver.testCliDriver[llap_acid_fast]
(batchId=161)
org.apache.hadoop.hive.cli.TestMiniLlapLocalCliDriver.testCliDriver[resourceplan]
(batchId=164)
org.apache.hadoop.hive.cli.TestMiniLlapLocalCliDriver.testCliDriver[sysdb]
(batchId=161)
org.apache.hadoop.hive.cli.TestMiniLlapLocalCliDriver.testCliDriver[vectorization_input_format_excludes]
(batchId=163)
org.apache.hadoop.hive.cli.TestSparkCliDriver.testCliDriver[ppd_join5]
(batchId=122)
org.apache.hadoop.hive.ql.TestCreateUdfEntities.testUdfWithDfsResource
(batchId=228)
org.apache.hadoop.hive.ql.exec.TestOperators.testNoConditionalTaskSizeForLlap
(batchId=282)
org.apache.hadoop.hive.ql.io.TestDruidRecordWriter.testWrite (batchId=256)
org.apache.hive.beeline.cli.TestHiveCli.testNoErrorDB (batchId=188)
org.apache.hive.jdbc.TestSSL.testConnectionMismatch (batchId=234)
org.apache.hive.jdbc.TestSSL.testConnectionWrongCertCN (batchId=234)
org.apache.hive.jdbc.TestSSL.testMetastoreConnectionWrongCertCN (batchId=234)
{noformat}
Test results: https://builds.apache.org/job/PreCommit-HIVE-Build/8871/testReport
Console output: https://builds.apache.org/job/PreCommit-HIVE-Build/8871/console
Test logs: http://104.198.109.242/logs/PreCommit-HIVE-Build-8871/
Messages:
{noformat}
Executing org.apache.hive.ptest.execution.TestCheckPhase
Executing org.apache.hive.ptest.execution.PrepPhase
Executing org.apache.hive.ptest.execution.YetusPhase
Executing org.apache.hive.ptest.execution.ExecutionPhase
Executing org.apache.hive.ptest.execution.ReportingPhase
Tests exited with: TestsFailedException: 21 tests failed
{noformat}
This message is automatically generated.
ATTACHMENT ID: 12907839 - PreCommit-HIVE-Build
> Investigate to enable MapInput cache in Hive on Spark
> -
>
> Key: HIVE-18301
> URL: https://issues.apache.org/jira/browse/HIVE-18301
> Project: Hive
> Issue Type: Bug
>Reporter: liyunzhang
>Assignee: liyunzhang
>Priority: Major
> Attachments: HIVE-18301.1.patch, HIVE-18301.patch
>
>
> Before IOContext problem is found in MapTran when spark rdd cache is enabled
> in HIVE-8920.
> so we disabled rdd cache in MapTran at
> [SparkPlanGenerator|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/spark/SparkPlanGenerator.java#L202].
> The problem is IOContext seems not initialized correctly in the spark yarn
> client/cluster mode and caused the exception like
> {code}
> Job aborted due to stage failure: Task 93 in stage 0.0 failed 4 times, most
> recent failure: Lost task 93.3 in stage 0.0 (TID 616, bdpe48):
> java.lang.RuntimeException: Error processing row:
> java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:165)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:48)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:27)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveBaseFunctionResultList.hasNext(HiveBaseFunctionResultList.java:85)
> at
> scala.collection.convert.Wrappers$JIteratorWrapper.hasNext(Wrappers.scala:42)
> at
> org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.write(BypassMergeSortShuffleWriter.java:125)
> at
>
[jira] [Commented] (HIVE-18301) Investigate to enable MapInput cache in Hive on Spark
[
https://issues.apache.org/jira/browse/HIVE-18301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16342164#comment-16342164
]
Hive QA commented on HIVE-18301:
| (x) *{color:red}-1 overall{color}* |
\\
\\
|| Vote || Subsystem || Runtime || Comment ||
|| || || || {color:brown} Prechecks {color} ||
| {color:blue}0{color} | {color:blue} findbugs {color} | {color:blue} 0m
1s{color} | {color:blue} Findbugs executables are not available. {color} |
| {color:green}+1{color} | {color:green} @author {color} | {color:green} 0m
0s{color} | {color:green} The patch does not contain any @author tags. {color} |
|| || || || {color:brown} master Compile Tests {color} ||
| {color:blue}0{color} | {color:blue} mvndep {color} | {color:blue} 1m
15s{color} | {color:blue} Maven dependency ordering for branch {color} |
| {color:green}+1{color} | {color:green} mvninstall {color} | {color:green} 5m
6s{color} | {color:green} master passed {color} |
| {color:green}+1{color} | {color:green} compile {color} | {color:green} 1m
7s{color} | {color:green} master passed {color} |
| {color:green}+1{color} | {color:green} checkstyle {color} | {color:green} 0m
47s{color} | {color:green} master passed {color} |
| {color:green}+1{color} | {color:green} javadoc {color} | {color:green} 1m
0s{color} | {color:green} master passed {color} |
|| || || || {color:brown} Patch Compile Tests {color} ||
| {color:blue}0{color} | {color:blue} mvndep {color} | {color:blue} 0m
20s{color} | {color:blue} Maven dependency ordering for patch {color} |
| {color:green}+1{color} | {color:green} mvninstall {color} | {color:green} 1m
26s{color} | {color:green} the patch passed {color} |
| {color:green}+1{color} | {color:green} compile {color} | {color:green} 1m
12s{color} | {color:green} the patch passed {color} |
| {color:green}+1{color} | {color:green} javac {color} | {color:green} 1m
12s{color} | {color:green} the patch passed {color} |
| {color:red}-1{color} | {color:red} checkstyle {color} | {color:red} 0m
35s{color} | {color:red} ql: The patch generated 49 new + 76 unchanged - 2
fixed = 125 total (was 78) {color} |
| {color:green}+1{color} | {color:green} whitespace {color} | {color:green} 0m
0s{color} | {color:green} The patch has no whitespace issues. {color} |
| {color:green}+1{color} | {color:green} javadoc {color} | {color:green} 1m
0s{color} | {color:green} the patch passed {color} |
|| || || || {color:brown} Other Tests {color} ||
| {color:red}-1{color} | {color:red} asflicense {color} | {color:red} 0m
11s{color} | {color:red} The patch generated 7 ASF License warnings. {color} |
| {color:black}{color} | {color:black} {color} | {color:black} 14m 38s{color} |
{color:black} {color} |
\\
\\
|| Subsystem || Report/Notes ||
| Optional Tests | asflicense javac javadoc findbugs checkstyle compile |
| uname | Linux hiveptest-server-upstream 3.16.0-4-amd64 #1 SMP Debian
3.16.36-1+deb8u1 (2016-09-03) x86_64 GNU/Linux |
| Build tool | maven |
| Personality | /data/hiveptest/working/yetus/dev-support/hive-personality.sh |
| git revision | master / cebe012 |
| Default Java | 1.8.0_111 |
| checkstyle |
http://104.198.109.242/logs//PreCommit-HIVE-Build-8871/yetus/diff-checkstyle-ql.txt
|
| asflicense |
http://104.198.109.242/logs//PreCommit-HIVE-Build-8871/yetus/patch-asflicense-problems.txt
|
| modules | C: common ql itests U: . |
| Console output |
http://104.198.109.242/logs//PreCommit-HIVE-Build-8871/yetus.txt |
| Powered by | Apache Yetushttp://yetus.apache.org |
This message was automatically generated.
> Investigate to enable MapInput cache in Hive on Spark
> -
>
> Key: HIVE-18301
> URL: https://issues.apache.org/jira/browse/HIVE-18301
> Project: Hive
> Issue Type: Bug
>Reporter: liyunzhang
>Assignee: liyunzhang
>Priority: Major
> Attachments: HIVE-18301.1.patch, HIVE-18301.patch
>
>
> Before IOContext problem is found in MapTran when spark rdd cache is enabled
> in HIVE-8920.
> so we disabled rdd cache in MapTran at
> [SparkPlanGenerator|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/spark/SparkPlanGenerator.java#L202].
> The problem is IOContext seems not initialized correctly in the spark yarn
> client/cluster mode and caused the exception like
> {code}
> Job aborted due to stage failure: Task 93 in stage 0.0 failed 4 times, most
> recent failure: Lost task 93.3 in stage 0.0 (TID 616, bdpe48):
> java.lang.RuntimeException: Error processing row:
> java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:165)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:48)
> at
>
[jira] [Commented] (HIVE-18301) Investigate to enable MapInput cache in Hive on Spark
[
https://issues.apache.org/jira/browse/HIVE-18301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16340717#comment-16340717
]
liyunzhang commented on HIVE-18301:
---
for the problem it may make the RDD bigger:
original tuple._1() is
[CombineKey|https://github.com/apache/hive/blob/master/shims/common/src/main/java/org/apache/hadoop/hive/shims/CombineHiveKey.java]
which stores offset(LongWriteObject), and in currentSolution , tuple._1() is
inputPath
([org.apache.hadoop.io|https://github.com/kellyzly/hadoop/blob/trunk/hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/io/Text.java]
[.Text|https://github.com/kellyzly/hadoop/blob/trunk/hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/io/Text.java])
whose size is decided by actual input file path length.
> Investigate to enable MapInput cache in Hive on Spark
> -
>
> Key: HIVE-18301
> URL: https://issues.apache.org/jira/browse/HIVE-18301
> Project: Hive
> Issue Type: Bug
>Reporter: liyunzhang
>Assignee: liyunzhang
>Priority: Major
> Attachments: HIVE-18301.1.patch, HIVE-18301.patch
>
>
> Before IOContext problem is found in MapTran when spark rdd cache is enabled
> in HIVE-8920.
> so we disabled rdd cache in MapTran at
> [SparkPlanGenerator|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/spark/SparkPlanGenerator.java#L202].
> The problem is IOContext seems not initialized correctly in the spark yarn
> client/cluster mode and caused the exception like
> {code}
> Job aborted due to stage failure: Task 93 in stage 0.0 failed 4 times, most
> recent failure: Lost task 93.3 in stage 0.0 (TID 616, bdpe48):
> java.lang.RuntimeException: Error processing row:
> java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:165)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:48)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:27)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveBaseFunctionResultList.hasNext(HiveBaseFunctionResultList.java:85)
> at
> scala.collection.convert.Wrappers$JIteratorWrapper.hasNext(Wrappers.scala:42)
> at
> org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.write(BypassMergeSortShuffleWriter.java:125)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:79)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:47)
> at org.apache.spark.scheduler.Task.run(Task.scala:85)
> at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:274)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:745)
> Caused by: java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.AbstractMapOperator.getNominalPath(AbstractMapOperator.java:101)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.cleanUpInputFileChangedOp(MapOperator.java:516)
> at
> org.apache.hadoop.hive.ql.exec.Operator.cleanUpInputFileChanged(Operator.java:1187)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.process(MapOperator.java:546)
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:152)
> ... 12 more
> Driver stacktrace:
> {code}
> in yarn client/cluster mode, sometimes
> [ExecMapperContext#currentInputPath|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/mr/ExecMapperContext.java#L109]
> is null when rdd cach is enabled.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
[jira] [Commented] (HIVE-18301) Investigate to enable MapInput cache in Hive on Spark
[
https://issues.apache.org/jira/browse/HIVE-18301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16338586#comment-16338586
]
liyunzhang commented on HIVE-18301:
---
[~lirui]: {quote}
does the proposed solution mean we need to cache the input path for each record
of the table?
{quote}
yes.
{quote}
I wonder whether we can reuse the Text for same input paths. Besides, it's not
efficient to check the job conf each time we process a row. You can just check
it once and remember the value.
{quote}
OK, I will reuse to reduce some computation.
[~lirui], [~xuefuz],[~csun]:
The concern to this solution is
1. it may make the RDD bigger( I don't compare the original tuple._1() and
inputPath, will do it later)
2. I modify the {{HadoopRDD-CopyRDD-MT-11}}, here CopyRDD does not store the
original data anymore, CopyRDD store the inputPath and orignal data. It maybe
confusing.
Do you think idea is ok to commit, if not ok, maybe I need to try other
solution or can you provide other better solution? Thanks!
> Investigate to enable MapInput cache in Hive on Spark
> -
>
> Key: HIVE-18301
> URL: https://issues.apache.org/jira/browse/HIVE-18301
> Project: Hive
> Issue Type: Bug
>Reporter: liyunzhang
>Assignee: liyunzhang
>Priority: Major
> Attachments: HIVE-18301.patch
>
>
> Before IOContext problem is found in MapTran when spark rdd cache is enabled
> in HIVE-8920.
> so we disabled rdd cache in MapTran at
> [SparkPlanGenerator|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/spark/SparkPlanGenerator.java#L202].
> The problem is IOContext seems not initialized correctly in the spark yarn
> client/cluster mode and caused the exception like
> {code}
> Job aborted due to stage failure: Task 93 in stage 0.0 failed 4 times, most
> recent failure: Lost task 93.3 in stage 0.0 (TID 616, bdpe48):
> java.lang.RuntimeException: Error processing row:
> java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:165)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:48)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:27)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveBaseFunctionResultList.hasNext(HiveBaseFunctionResultList.java:85)
> at
> scala.collection.convert.Wrappers$JIteratorWrapper.hasNext(Wrappers.scala:42)
> at
> org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.write(BypassMergeSortShuffleWriter.java:125)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:79)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:47)
> at org.apache.spark.scheduler.Task.run(Task.scala:85)
> at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:274)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:745)
> Caused by: java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.AbstractMapOperator.getNominalPath(AbstractMapOperator.java:101)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.cleanUpInputFileChangedOp(MapOperator.java:516)
> at
> org.apache.hadoop.hive.ql.exec.Operator.cleanUpInputFileChanged(Operator.java:1187)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.process(MapOperator.java:546)
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:152)
> ... 12 more
> Driver stacktrace:
> {code}
> in yarn client/cluster mode, sometimes
> [ExecMapperContext#currentInputPath|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/mr/ExecMapperContext.java#L109]
> is null when rdd cach is enabled.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
[jira] [Commented] (HIVE-18301) Investigate to enable MapInput cache in Hive on Spark
[
https://issues.apache.org/jira/browse/HIVE-18301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16337632#comment-16337632
]
Rui Li commented on HIVE-18301:
---
Hi [~kellyzly], does the proposed solution mean we need to cache the input path
for each record of the table? I wonder whether we can reuse the Text for same
input paths. Besides, it's not efficient to check the job conf each time we
process a row. You can just check it once and remember the value. Anyway, it's
better to get some measurement of the overhead.
> Investigate to enable MapInput cache in Hive on Spark
> -
>
> Key: HIVE-18301
> URL: https://issues.apache.org/jira/browse/HIVE-18301
> Project: Hive
> Issue Type: Bug
>Reporter: liyunzhang
>Assignee: liyunzhang
>Priority: Major
> Attachments: HIVE-18301.patch
>
>
> Before IOContext problem is found in MapTran when spark rdd cache is enabled
> in HIVE-8920.
> so we disabled rdd cache in MapTran at
> [SparkPlanGenerator|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/spark/SparkPlanGenerator.java#L202].
> The problem is IOContext seems not initialized correctly in the spark yarn
> client/cluster mode and caused the exception like
> {code}
> Job aborted due to stage failure: Task 93 in stage 0.0 failed 4 times, most
> recent failure: Lost task 93.3 in stage 0.0 (TID 616, bdpe48):
> java.lang.RuntimeException: Error processing row:
> java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:165)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:48)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:27)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveBaseFunctionResultList.hasNext(HiveBaseFunctionResultList.java:85)
> at
> scala.collection.convert.Wrappers$JIteratorWrapper.hasNext(Wrappers.scala:42)
> at
> org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.write(BypassMergeSortShuffleWriter.java:125)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:79)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:47)
> at org.apache.spark.scheduler.Task.run(Task.scala:85)
> at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:274)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:745)
> Caused by: java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.AbstractMapOperator.getNominalPath(AbstractMapOperator.java:101)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.cleanUpInputFileChangedOp(MapOperator.java:516)
> at
> org.apache.hadoop.hive.ql.exec.Operator.cleanUpInputFileChanged(Operator.java:1187)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.process(MapOperator.java:546)
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:152)
> ... 12 more
> Driver stacktrace:
> {code}
> in yarn client/cluster mode, sometimes
> [ExecMapperContext#currentInputPath|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/mr/ExecMapperContext.java#L109]
> is null when rdd cach is enabled.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
[jira] [Commented] (HIVE-18301) Investigate to enable MapInput cache in Hive on Spark
[
https://issues.apache.org/jira/browse/HIVE-18301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16336946#comment-16336946
]
liyunzhang commented on HIVE-18301:
---
In HIVE-18301.patch, it provides one solution to transfer the
{{IOContext::inputPath}}
{code}
inputRDD1inputRDD2
|CopyFunction| CopyFunction
CopyRDD1CopyRDD2
||
MT_11 MT_12
| |
RT_1 RT_2
\ /
Union
{code}
MT_11 will call following stack to initialize IOContext::inputPath
{code}
CombineHiveRecordReader#init
->HiveContextAwareRecordReader.initIOContext
->IOContext.setInputPath
{code}
inputRDD1 and inputRDD2 are same table's rdd, so CopyRDD1 and CopyRDD2 are same
rdd if rdd cache is enabled. When MT_12 will not call
CombineHiveRecordReader#init to initialize {{IOContext::inputPath}} but
{{MapOperator#process(Writable value)}} still need this value. IOContext is
bound to single thread, so the value is different in different thread.
{{inputRDD1-CopyRDD1-MT_11-RT_1}} and {{inputRDD2-CopyRDD2-MT_12-RT_2}} is
called in different thread. So IOContext can not be shared between these two
threads.
For this issue, I gave following solution:
We save the inputPath in CopyRDD1 when {{inputRDD1-CopyRDD1-MT_11-RT_1}} is
executed. CopyRDD2 get the cached value and inputPath from CopyRDD1 which is
stored in spark cache manager. We reinitialized the {{IOContext::inputPath}}
in {{MapOperator#process(Writable value)}} in MT_12.
*where to setInputPath?*
MapInput#CopyFunction#call, save inputPath in the first element of returned
tuple
{code}
public MapInput(SparkPlan sparkPlan, JavaPairRDD
hadoopRDD) {
this(sparkPlan, hadoopRDD, false);
@@ -79,10 +83,19 @@ public void setToCache(boolean toCache) {
call(Tuple2 tuple) throws Exception {
if (conf == null) {
conf = new Configuration();
+conf.set("hive.execution.engine","spark");
}
-
- return new Tuple2(tuple._1(),
- WritableUtils.clone(tuple._2(), conf));
+ //CopyFunction MapFunction
+ // HADOOPRDD-> RDD1-> RDD2.
+ // these transformation are in one stage and will be executed by 1 spark
task(thread),
+ // IOContext.get(conf).getInputPath will not be null.
+ String inputPath = IOContextMap.get(conf).getInputPath().toString();
+ Text inputPathText = new Text(inputPath);
+ // save inputPath in the first element of returned tuple
+ return new Tuple2(inputPathText,
+WritableUtils.clone(tuple._2(), conf));
}
}
{code}
*where to getInputPath?*
{code}
SparkMapRecordHandler#getInputPath
public void processRow(Object key, Object value) throws IOException {
+if (HiveConf.getBoolVar(jc,
HiveConf.ConfVars.HIVE_SPARK_SHARED_WORK_OPTIMIZATION)) {
+ Path inputPath = IOContextMap.get(jc).getInputPath();
+ // when inputPath is null, it means the record is cached
+ if (inputPath == null) {
+Text pathText = (Text) key;
+IOContextMap.get(jc).setInputPath(new Path(pathText.toString()));
+ }
+}
{code}
[~lirui], [~xuefuz], [~stakiar],[~csun], please give me your suggesions, thanks!
> Investigate to enable MapInput cache in Hive on Spark
> -
>
> Key: HIVE-18301
> URL: https://issues.apache.org/jira/browse/HIVE-18301
> Project: Hive
> Issue Type: Bug
>Reporter: liyunzhang
>Assignee: liyunzhang
>Priority: Major
> Attachments: HIVE-18301.patch
>
>
> Before IOContext problem is found in MapTran when spark rdd cache is enabled
> in HIVE-8920.
> so we disabled rdd cache in MapTran at
> [SparkPlanGenerator|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/spark/SparkPlanGenerator.java#L202].
> The problem is IOContext seems not initialized correctly in the spark yarn
> client/cluster mode and caused the exception like
> {code}
> Job aborted due to stage failure: Task 93 in stage 0.0 failed 4 times, most
> recent failure: Lost task 93.3 in stage 0.0 (TID 616, bdpe48):
> java.lang.RuntimeException: Error processing row:
> java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:165)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:48)
> at
>
[jira] [Commented] (HIVE-18301) Investigate to enable MapInput cache in Hive on Spark
[
https://issues.apache.org/jira/browse/HIVE-18301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16336941#comment-16336941
]
liyunzhang commented on HIVE-18301:
---
[~lirui]: {quote}
And we won't be able to tell the file boundaries because they're cached.
{quote}
yes. when IOContext.getInputPath() = null only means the record is cached.
Can not know file boundaries.
So IOContext.getInputPath is necessary.
> Investigate to enable MapInput cache in Hive on Spark
> -
>
> Key: HIVE-18301
> URL: https://issues.apache.org/jira/browse/HIVE-18301
> Project: Hive
> Issue Type: Bug
>Reporter: liyunzhang
>Assignee: liyunzhang
>Priority: Major
>
> Before IOContext problem is found in MapTran when spark rdd cache is enabled
> in HIVE-8920.
> so we disabled rdd cache in MapTran at
> [SparkPlanGenerator|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/spark/SparkPlanGenerator.java#L202].
> The problem is IOContext seems not initialized correctly in the spark yarn
> client/cluster mode and caused the exception like
> {code}
> Job aborted due to stage failure: Task 93 in stage 0.0 failed 4 times, most
> recent failure: Lost task 93.3 in stage 0.0 (TID 616, bdpe48):
> java.lang.RuntimeException: Error processing row:
> java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:165)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:48)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:27)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveBaseFunctionResultList.hasNext(HiveBaseFunctionResultList.java:85)
> at
> scala.collection.convert.Wrappers$JIteratorWrapper.hasNext(Wrappers.scala:42)
> at
> org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.write(BypassMergeSortShuffleWriter.java:125)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:79)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:47)
> at org.apache.spark.scheduler.Task.run(Task.scala:85)
> at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:274)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:745)
> Caused by: java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.AbstractMapOperator.getNominalPath(AbstractMapOperator.java:101)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.cleanUpInputFileChangedOp(MapOperator.java:516)
> at
> org.apache.hadoop.hive.ql.exec.Operator.cleanUpInputFileChanged(Operator.java:1187)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.process(MapOperator.java:546)
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:152)
> ... 12 more
> Driver stacktrace:
> {code}
> in yarn client/cluster mode, sometimes
> [ExecMapperContext#currentInputPath|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/mr/ExecMapperContext.java#L109]
> is null when rdd cach is enabled.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
[jira] [Commented] (HIVE-18301) Investigate to enable MapInput cache in Hive on Spark
[
https://issues.apache.org/jira/browse/HIVE-18301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16336898#comment-16336898
]
Rui Li commented on HIVE-18301:
---
{quote}We need not to call MapOperator#cleanUpInputFileChanged because
MapOperator#cleanUpInputFileChanged is only designed for one Mapper scanning
multiple files
{quote}
When RDD is cached, mapper reads records from the cache. But I think those
records may come from multiple underlying files right? And we won't be able to
tell the file boundaries because they're cached.
> Investigate to enable MapInput cache in Hive on Spark
> -
>
> Key: HIVE-18301
> URL: https://issues.apache.org/jira/browse/HIVE-18301
> Project: Hive
> Issue Type: Bug
>Reporter: liyunzhang
>Assignee: liyunzhang
>Priority: Major
>
> Before IOContext problem is found in MapTran when spark rdd cache is enabled
> in HIVE-8920.
> so we disabled rdd cache in MapTran at
> [SparkPlanGenerator|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/spark/SparkPlanGenerator.java#L202].
> The problem is IOContext seems not initialized correctly in the spark yarn
> client/cluster mode and caused the exception like
> {code}
> Job aborted due to stage failure: Task 93 in stage 0.0 failed 4 times, most
> recent failure: Lost task 93.3 in stage 0.0 (TID 616, bdpe48):
> java.lang.RuntimeException: Error processing row:
> java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:165)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:48)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:27)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveBaseFunctionResultList.hasNext(HiveBaseFunctionResultList.java:85)
> at
> scala.collection.convert.Wrappers$JIteratorWrapper.hasNext(Wrappers.scala:42)
> at
> org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.write(BypassMergeSortShuffleWriter.java:125)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:79)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:47)
> at org.apache.spark.scheduler.Task.run(Task.scala:85)
> at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:274)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:745)
> Caused by: java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.AbstractMapOperator.getNominalPath(AbstractMapOperator.java:101)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.cleanUpInputFileChangedOp(MapOperator.java:516)
> at
> org.apache.hadoop.hive.ql.exec.Operator.cleanUpInputFileChanged(Operator.java:1187)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.process(MapOperator.java:546)
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:152)
> ... 12 more
> Driver stacktrace:
> {code}
> in yarn client/cluster mode, sometimes
> [ExecMapperContext#currentInputPath|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/mr/ExecMapperContext.java#L109]
> is null when rdd cach is enabled.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
[jira] [Commented] (HIVE-18301) Investigate to enable MapInput cache in Hive on Spark
[
https://issues.apache.org/jira/browse/HIVE-18301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16330190#comment-16330190
]
liyunzhang commented on HIVE-18301:
---
[~lirui]:
{quote}My understanding is if the HadoopRDD is cached, the records are not
produced by record reader and IOContext is not populated. Therefore the
information in IOContext will be unavailable, e.g. the input path. This may
cause problem because some operators need to take certain actions when input
file changes – {{Operator::cleanUpInputFileChanged}}.
So basically my point is we have to figure out the scenarios where IOContext
is necessary. Then decide whether we should disable caching in such cases.
{quote}
Yes, if HadoopRDD is cached, it will not call
{code:java}
CombineHiveRecordReader#init
->HiveContextAwareRecordReader.initIOContext
->IOContext.setInputPath
{code}
.
It will use the cached result to call MapOperator#process(Writable value), so
NPE is thrown because at that time IOContext.getInputPath return null. Now I
just modify the code of MapOperator#process(Writable value) like
[link|https://github.com/kellyzly/hive/commit/e81b7df572e2c543095f55dd160b428c355da2fb]
Here my question is
1. when {{context.getIoCxt().getInputPath() == null}}, I think in this
situation, this record is from cache not from CombineHiveRecordReader. We need
not to call MapOperator#cleanUpInputFileChanged because
MapOperator#cleanUpInputFileChanged is only designed for one Mapper scanning
multiple files(like CombineFileInputFormat) and multiple partitions and
inputPath will change in these situations and need to call
{{cleanUpInputFileChanged}} to reinitialize [some
variables|https://github.com/apache/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/MapOperator.java#L532]
but we need not consider reinitialization for a cached record. Is my
understanding right? If right, is there any other way to judge this is cached
record or not except by {{context.getIoCxt().getInputPath() == null}}
2. how to initiliaze IOContext#getInputPath in cache situation? we need this
variable to reinitialize MapOperator::currentCtxs in
MapOperator#initializeContexts
{code:java}
public void initializeContexts() {
Path fpath = getExecContext().getCurrentInputPath();
String nominalPath = getNominalPath(fpath);
Map contexts = opCtxMap.get(nominalPath);
currentCtxs = contexts.values().toArray(new MapOpCtx[contexts.size()]);
}
{code}
in the code, we store MapOpCtx for every MapOperator in opCtxMap(Map>). In table with partitions, there will be multiple
elements in opCtxMap( opCtxMap.keySet() is a set containing partition names).
Currently I test on a table without partitions and can directly use
opCtxMap.values().iterator().next() to initialize
[context|https://github.com/kellyzly/hive/blob/HIVE-17486.4/ql/src/java/org/apache/hadoop/hive/ql/exec/MapOperator.java#L713]
and runs successfully in yarn mode. But I guess this is not right with
partitioned table.
> Investigate to enable MapInput cache in Hive on Spark
> -
>
> Key: HIVE-18301
> URL: https://issues.apache.org/jira/browse/HIVE-18301
> Project: Hive
> Issue Type: Bug
>Reporter: liyunzhang
>Assignee: liyunzhang
>Priority: Major
>
> Before IOContext problem is found in MapTran when spark rdd cache is enabled
> in HIVE-8920.
> so we disabled rdd cache in MapTran at
> [SparkPlanGenerator|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/spark/SparkPlanGenerator.java#L202].
> The problem is IOContext seems not initialized correctly in the spark yarn
> client/cluster mode and caused the exception like
> {code}
> Job aborted due to stage failure: Task 93 in stage 0.0 failed 4 times, most
> recent failure: Lost task 93.3 in stage 0.0 (TID 616, bdpe48):
> java.lang.RuntimeException: Error processing row:
> java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:165)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:48)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:27)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveBaseFunctionResultList.hasNext(HiveBaseFunctionResultList.java:85)
> at
> scala.collection.convert.Wrappers$JIteratorWrapper.hasNext(Wrappers.scala:42)
> at
> org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.write(BypassMergeSortShuffleWriter.java:125)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:79)
> at
>
[jira] [Commented] (HIVE-18301) Investigate to enable MapInput cache in Hive on Spark
[
https://issues.apache.org/jira/browse/HIVE-18301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16309361#comment-16309361
]
liyunzhang commented on HIVE-18301:
---
Here some update about NPE:
the normal case when enable rdd cache
the stacktrace of initIOContext which will intialize
ExecMapperContext#setCurrentInputPath is
{code}
org.apache.hadoop.hive.ql.io.HiveContextAwareRecordReader.initIOContext(HiveContextAwareRecordReader.java:175)
org.apache.hadoop.hive.ql.io.HiveContextAwareRecordReader.initIOContext(HiveContextAwareRecordReader.java:211)
org.apache.hadoop.hive.ql.io.CombineHiveRecordReader.(CombineHiveRecordReader.java:101)
sun.reflect.NativeConstructorAccessorImpl.newInstance0(Native Method)
sun.reflect.NativeConstructorAccessorImpl.newInstance(NativeConstructorAccessorImpl.java:62)
sun.reflect.DelegatingConstructorAccessorImpl.newInstance(DelegatingConstructorAccessorImpl.java:45)
java.lang.reflect.Constructor.newInstance(Constructor.java:423)
org.apache.hadoop.hive.shims.HadoopShimsSecure$CombineFileRecordReader.initNextRecordReader(HadoopShimsSecure.java:257)
org.apache.hadoop.hive.shims.HadoopShimsSecure$CombineFileRecordReader.(HadoopShimsSecure.java:217)
org.apache.hadoop.hive.shims.HadoopShimsSecure$CombineFileInputFormatShim.getRecordReader(HadoopShimsSecure.java:346)
org.apache.hadoop.hive.ql.io.CombineHiveInputFormat.getRecordReader(CombineHiveInputFormat.java:712)
org.apache.spark.rdd.HadoopRDD$$anon$1.(HadoopRDD.scala:246)
org.apache.spark.rdd.HadoopRDD.compute(HadoopRDD.scala:209)
org.apache.spark.rdd.HadoopRDD.compute(HadoopRDD.scala:102)
org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:319)
org.apache.spark.rdd.RDD.iterator(RDD.scala:283)
org.apache.spark.rdd.MapPartitionsRDD.compute(MapPartitionsRDD.scala:38)
org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:319)
org.apache.spark.rdd.RDD$$anonfun$8.apply(RDD.scala:332)
org.apache.spark.rdd.RDD$$anonfun$8.apply(RDD.scala:330)
org.apache.spark.storage.BlockManager$$anonfun$doPutIterator$1.apply(BlockManager.scala:919)
org.apache.spark.storage.BlockManager$$anonfun$doPutIterator$1.apply(BlockManager.scala:910)
org.apache.spark.storage.BlockManager.doPut(BlockManager.scala:866)
org.apache.spark.storage.BlockManager.doPutIterator(BlockManager.scala:910)
org.apache.spark.storage.BlockManager.getOrElseUpdate(BlockManager.scala:668)
org.apache.spark.rdd.RDD.getOrCompute(RDD.scala:330)
org.apache.spark.rdd.RDD.iterator(RDD.scala:281)
org.apache.spark.rdd.MapPartitionsRDD.compute(MapPartitionsRDD.scala:38)
org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:319)
org.apache.spark.rdd.RDD$$anonfun$8.apply(RDD.scala:332)
org.apache.spark.rdd.RDD$$anonfun$8.apply(RDD.scala:330)
org.apache.spark.storage.BlockManager$$anonfun$doPutIterator$1.apply(BlockManager.scala:919)
org.apache.spark.storage.BlockManager$$anonfun$doPutIterator$1.apply(BlockManager.scala:910)
org.apache.spark.storage.BlockManager.doPut(BlockManager.scala:866)
org.apache.spark.storage.BlockManager.doPutIterator(BlockManager.scala:910)
org.apache.spark.storage.BlockManager.getOrElseUpdate(BlockManager.scala:668)
org.apache.spark.rdd.RDD.getOrCompute(RDD.scala:330)
org.apache.spark.rdd.RDD.iterator(RDD.scala:281)
org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:79)
org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:47)
org.apache.spark.scheduler.Task.run(Task.scala:85)
org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:274)
java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
java.lang.Thread.run(Thread.java:745)
{code}
the stacktrace of ExecMapperContext#getCurrentInputPath
{code}
org.apache.hadoop.hive.ql.exec.mr.ExecMapperContext.getCurrentInputPath(ExecMapperContext.java:113)
org.apache.hadoop.hive.ql.exec.MapOperator.cleanUpInputFileChangedOp(MapOperator.java:512)
org.apache.hadoop.hive.ql.exec.Operator.cleanUpInputFileChanged(Operator.java:1187)
org.apache.hadoop.hive.ql.exec.MapOperator.process(MapOperator.java:543)
org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:136)
org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:48)
org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:27)
org.apache.hadoop.hive.ql.exec.spark.HiveBaseFunctionResultList.hasNext(HiveBaseFunctionResultList.java:85)
scala.collection.convert.Wrappers$JIteratorWrapper.hasNext(Wrappers.scala:42)
org.apache.spark.storage.memory.MemoryStore.putIteratorAsValues(MemoryStore.scala:213)
org.apache.spark.storage.BlockManager$$anonfun$doPutIterator$1.apply(BlockManager.scala:919)
[jira] [Commented] (HIVE-18301) Investigate to enable MapInput cache in Hive on Spark
[
https://issues.apache.org/jira/browse/HIVE-18301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16307619#comment-16307619
]
liyunzhang commented on HIVE-18301:
---
thanks for Rui's comment, will spend some time to investigate the NPE problem
.Here I want to say in
[HIVE-8920|https://issues.apache.org/jira/browse/HIVE-8920], the exception is
{code}
Caused by: java.lang.IllegalStateException: Invalid input path
hdfs://localhost:8020/user/hive/warehouse/dec2/dec.txt
at
org.apache.hadoop.hive.ql.exec.MapOperator.getNominalPath(MapOperator.java:406)
at
org.apache.hadoop.hive.ql.exec.MapOperator.cleanUpInputFileChangedOp(MapOperator.java:442)
{code}
Not NPE, so there are two different problems.
> Investigate to enable MapInput cache in Hive on Spark
> -
>
> Key: HIVE-18301
> URL: https://issues.apache.org/jira/browse/HIVE-18301
> Project: Hive
> Issue Type: Bug
>Reporter: liyunzhang
>Assignee: liyunzhang
>
> Before IOContext problem is found in MapTran when spark rdd cache is enabled
> in HIVE-8920.
> so we disabled rdd cache in MapTran at
> [SparkPlanGenerator|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/spark/SparkPlanGenerator.java#L202].
> The problem is IOContext seems not initialized correctly in the spark yarn
> client/cluster mode and caused the exception like
> {code}
> Job aborted due to stage failure: Task 93 in stage 0.0 failed 4 times, most
> recent failure: Lost task 93.3 in stage 0.0 (TID 616, bdpe48):
> java.lang.RuntimeException: Error processing row:
> java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:165)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:48)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:27)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveBaseFunctionResultList.hasNext(HiveBaseFunctionResultList.java:85)
> at
> scala.collection.convert.Wrappers$JIteratorWrapper.hasNext(Wrappers.scala:42)
> at
> org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.write(BypassMergeSortShuffleWriter.java:125)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:79)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:47)
> at org.apache.spark.scheduler.Task.run(Task.scala:85)
> at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:274)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:745)
> Caused by: java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.AbstractMapOperator.getNominalPath(AbstractMapOperator.java:101)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.cleanUpInputFileChangedOp(MapOperator.java:516)
> at
> org.apache.hadoop.hive.ql.exec.Operator.cleanUpInputFileChanged(Operator.java:1187)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.process(MapOperator.java:546)
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:152)
> ... 12 more
> Driver stacktrace:
> {code}
> in yarn client/cluster mode, sometimes
> [ExecMapperContext#currentInputPath|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/mr/ExecMapperContext.java#L109]
> is null when rdd cach is enabled.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
[jira] [Commented] (HIVE-18301) Investigate to enable MapInput cache in Hive on Spark
[
https://issues.apache.org/jira/browse/HIVE-18301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16306218#comment-16306218
]
Rui Li commented on HIVE-18301:
---
I think SMB map join is one case where the "input file change" event is
necessary -- whenever the big table input file changes, SMBMapJoinOperator
needs to find corresponding input files for small tables in order to performa
bucketed join. Maybe we can identify all such cases and make sure MapInput
cache is disabled. For other cases, we can cache MapInput and just fix the NPE.
[~xuefuz], could you share your thoughts on this? Thanks.
> Investigate to enable MapInput cache in Hive on Spark
> -
>
> Key: HIVE-18301
> URL: https://issues.apache.org/jira/browse/HIVE-18301
> Project: Hive
> Issue Type: Bug
>Reporter: liyunzhang
>Assignee: liyunzhang
>
> Before IOContext problem is found in MapTran when spark rdd cache is enabled
> in HIVE-8920.
> so we disabled rdd cache in MapTran at
> [SparkPlanGenerator|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/spark/SparkPlanGenerator.java#L202].
> The problem is IOContext seems not initialized correctly in the spark yarn
> client/cluster mode and caused the exception like
> {code}
> Job aborted due to stage failure: Task 93 in stage 0.0 failed 4 times, most
> recent failure: Lost task 93.3 in stage 0.0 (TID 616, bdpe48):
> java.lang.RuntimeException: Error processing row:
> java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:165)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:48)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:27)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveBaseFunctionResultList.hasNext(HiveBaseFunctionResultList.java:85)
> at
> scala.collection.convert.Wrappers$JIteratorWrapper.hasNext(Wrappers.scala:42)
> at
> org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.write(BypassMergeSortShuffleWriter.java:125)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:79)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:47)
> at org.apache.spark.scheduler.Task.run(Task.scala:85)
> at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:274)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:745)
> Caused by: java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.AbstractMapOperator.getNominalPath(AbstractMapOperator.java:101)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.cleanUpInputFileChangedOp(MapOperator.java:516)
> at
> org.apache.hadoop.hive.ql.exec.Operator.cleanUpInputFileChanged(Operator.java:1187)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.process(MapOperator.java:546)
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:152)
> ... 12 more
> Driver stacktrace:
> {code}
> in yarn client/cluster mode, sometimes
> [ExecMapperContext#currentInputPath|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/mr/ExecMapperContext.java#L109]
> is null when rdd cach is enabled.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
[jira] [Commented] (HIVE-18301) Investigate to enable MapInput cache in Hive on Spark
[
https://issues.apache.org/jira/browse/HIVE-18301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16304383#comment-16304383
]
Rui Li commented on HIVE-18301:
---
My understanding is if the HadoopRDD is cached, the records are not produced by
record reader and IOContext is not populated. Therefore the information in
IOContext will be unavailable, e.g. the input path. This may cause problem
because some operators need to take certain actions when input file changes --
{{Operator::cleanUpInputFileChanged}}.
So basically my point is we have to figure out the scenarios where IOContext is
necessary. Then decide whether we should disable caching in such cases.
> Investigate to enable MapInput cache in Hive on Spark
> -
>
> Key: HIVE-18301
> URL: https://issues.apache.org/jira/browse/HIVE-18301
> Project: Hive
> Issue Type: Bug
>Reporter: liyunzhang
>Assignee: liyunzhang
>
> Before IOContext problem is found in MapTran when spark rdd cache is enabled
> in HIVE-8920.
> so we disabled rdd cache in MapTran at
> [SparkPlanGenerator|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/spark/SparkPlanGenerator.java#L202].
> The problem is IOContext seems not initialized correctly in the spark yarn
> client/cluster mode and caused the exception like
> {code}
> Job aborted due to stage failure: Task 93 in stage 0.0 failed 4 times, most
> recent failure: Lost task 93.3 in stage 0.0 (TID 616, bdpe48):
> java.lang.RuntimeException: Error processing row:
> java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:165)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:48)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:27)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveBaseFunctionResultList.hasNext(HiveBaseFunctionResultList.java:85)
> at
> scala.collection.convert.Wrappers$JIteratorWrapper.hasNext(Wrappers.scala:42)
> at
> org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.write(BypassMergeSortShuffleWriter.java:125)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:79)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:47)
> at org.apache.spark.scheduler.Task.run(Task.scala:85)
> at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:274)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:745)
> Caused by: java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.AbstractMapOperator.getNominalPath(AbstractMapOperator.java:101)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.cleanUpInputFileChangedOp(MapOperator.java:516)
> at
> org.apache.hadoop.hive.ql.exec.Operator.cleanUpInputFileChanged(Operator.java:1187)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.process(MapOperator.java:546)
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:152)
> ... 12 more
> Driver stacktrace:
> {code}
> in yarn client/cluster mode, sometimes
> [ExecMapperContext#currentInputPath|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/mr/ExecMapperContext.java#L109]
> is null when rdd cach is enabled.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
[jira] [Commented] (HIVE-18301) Investigate to enable MapInput cache in Hive on Spark
[
https://issues.apache.org/jira/browse/HIVE-18301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16304359#comment-16304359
]
liyunzhang commented on HIVE-18301:
---
[~lirui]:
{quote}
My understanding is these information will be lost if the HadoopRDD is cached.
{quote}
You mean that HadoopRDD will not store the spark plan? If yes, actually in
hive, it stores the spark plan on a file on hdfs and deserialize and serialize
from the file. See more code in
hive/ql/src/java/org/apache/hadoop/hive/ql/exec/Utilities.java#getBaseWork. If
not, please spend more time to explain detail.
Here my question is that is there any other reason to disable MapInput#cache
besides avoiding multi-insert cases which there is union operator after
{{from}}
{code}
from (select * from dec union all select * from dec2) s
insert overwrite table dec3 select s.name, sum(s.value) group by s.name
insert overwrite table dec4 select s.name, s.value order by s.value;
{code}
If there is no other reason to disable MapInput# cache, I guess for
HIVE-17486, we can enable MapInput cache because HIVE-17486 is merge same
single table. There is few case like above ( from (select A union B) ).
> Investigate to enable MapInput cache in Hive on Spark
> -
>
> Key: HIVE-18301
> URL: https://issues.apache.org/jira/browse/HIVE-18301
> Project: Hive
> Issue Type: Bug
>Reporter: liyunzhang
>Assignee: liyunzhang
>
> Before IOContext problem is found in MapTran when spark rdd cache is enabled
> in HIVE-8920.
> so we disabled rdd cache in MapTran at
> [SparkPlanGenerator|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/spark/SparkPlanGenerator.java#L202].
> The problem is IOContext seems not initialized correctly in the spark yarn
> client/cluster mode and caused the exception like
> {code}
> Job aborted due to stage failure: Task 93 in stage 0.0 failed 4 times, most
> recent failure: Lost task 93.3 in stage 0.0 (TID 616, bdpe48):
> java.lang.RuntimeException: Error processing row:
> java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:165)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:48)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:27)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveBaseFunctionResultList.hasNext(HiveBaseFunctionResultList.java:85)
> at
> scala.collection.convert.Wrappers$JIteratorWrapper.hasNext(Wrappers.scala:42)
> at
> org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.write(BypassMergeSortShuffleWriter.java:125)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:79)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:47)
> at org.apache.spark.scheduler.Task.run(Task.scala:85)
> at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:274)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:745)
> Caused by: java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.AbstractMapOperator.getNominalPath(AbstractMapOperator.java:101)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.cleanUpInputFileChangedOp(MapOperator.java:516)
> at
> org.apache.hadoop.hive.ql.exec.Operator.cleanUpInputFileChanged(Operator.java:1187)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.process(MapOperator.java:546)
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:152)
> ... 12 more
> Driver stacktrace:
> {code}
> in yarn client/cluster mode, sometimes
> [ExecMapperContext#currentInputPath|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/mr/ExecMapperContext.java#L109]
> is null when rdd cach is enabled.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
[jira] [Commented] (HIVE-18301) Investigate to enable MapInput cache in Hive on Spark
[
https://issues.apache.org/jira/browse/HIVE-18301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16303680#comment-16303680
]
Rui Li commented on HIVE-18301:
---
I think we need to investigate how the input path is used, and what the
operators need to do when input file changes, etc. My understanding is these
information will be lost if the HadoopRDD is cached.
> Investigate to enable MapInput cache in Hive on Spark
> -
>
> Key: HIVE-18301
> URL: https://issues.apache.org/jira/browse/HIVE-18301
> Project: Hive
> Issue Type: Bug
>Reporter: liyunzhang
>Assignee: liyunzhang
>
> Before IOContext problem is found in MapTran when spark rdd cache is enabled
> in HIVE-8920.
> so we disabled rdd cache in MapTran at
> [SparkPlanGenerator|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/spark/SparkPlanGenerator.java#L202].
> The problem is IOContext seems not initialized correctly in the spark yarn
> client/cluster mode and caused the exception like
> {code}
> Job aborted due to stage failure: Task 93 in stage 0.0 failed 4 times, most
> recent failure: Lost task 93.3 in stage 0.0 (TID 616, bdpe48):
> java.lang.RuntimeException: Error processing row:
> java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:165)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:48)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:27)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveBaseFunctionResultList.hasNext(HiveBaseFunctionResultList.java:85)
> at
> scala.collection.convert.Wrappers$JIteratorWrapper.hasNext(Wrappers.scala:42)
> at
> org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.write(BypassMergeSortShuffleWriter.java:125)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:79)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:47)
> at org.apache.spark.scheduler.Task.run(Task.scala:85)
> at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:274)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:745)
> Caused by: java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.AbstractMapOperator.getNominalPath(AbstractMapOperator.java:101)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.cleanUpInputFileChangedOp(MapOperator.java:516)
> at
> org.apache.hadoop.hive.ql.exec.Operator.cleanUpInputFileChanged(Operator.java:1187)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.process(MapOperator.java:546)
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:152)
> ... 12 more
> Driver stacktrace:
> {code}
> in yarn client/cluster mode, sometimes
> [ExecMapperContext#currentInputPath|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/mr/ExecMapperContext.java#L109]
> is null when rdd cach is enabled.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
[jira] [Commented] (HIVE-18301) Investigate to enable MapInput cache in Hive on Spark
[
https://issues.apache.org/jira/browse/HIVE-18301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16301713#comment-16301713
]
liyunzhang commented on HIVE-18301:
---
[~xuefuz],[~csun]: I read jiras about MapInput IOContext problem and enable
MapInput rdd cache. And found the problem only happens OContext problem with
multiple MapWorks cloned for multi-insert \[Spark Branch\] like HIVE-8920
mentioned.
In HIVE-8920, I found the failure case is like
{code}
from (select * from dec union all select * from dec2) s
insert overwrite table dec3 select s.name, sum(s.value) group by s.name
insert overwrite table dec4 select s.name, s.value order by s.value;
{code}
I indeed saw the exception in my hive.log like
{code}
Caused by: java.lang.IllegalStateException: Invalid input path
hdfs://localhost:8020/user/hive/warehouse/dec2/dec.txt
at
org.apache.hadoop.hive.ql.exec.MapOperator.getNominalPath(MapOperator.java:406)
at
org.apache.hadoop.hive.ql.exec.MapOperator.cleanUpInputFileChangedOp(MapOperator.java:442)
{code}
here the problem happens on the MapInput is the union result of dec and dec2
case. But when I modify case
{code}
from (select * from dec ) s
insert overwrite table dec3 select s.name, sum(s.value) group by s.name
insert overwrite table dec4 select s.name, s.value order by s.value;
{code}
No such exception whether in local or yarn mode.
Whether the problem only happens in such complicated case( the rdd cache is
the union result of two tables)? If only happen in such complicated case, why
not only disable MapInput rdd cache in such case? Is there any other reason to
disable MapInput#rdd cache? Please spend some time to view it as both of you
have experience on it, thanks!
> Investigate to enable MapInput cache in Hive on Spark
> -
>
> Key: HIVE-18301
> URL: https://issues.apache.org/jira/browse/HIVE-18301
> Project: Hive
> Issue Type: Bug
>Reporter: liyunzhang
>Assignee: liyunzhang
>
> Before IOContext problem is found in MapTran when spark rdd cache is enabled
> in HIVE-8920.
> so we disabled rdd cache in MapTran at
> [SparkPlanGenerator|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/spark/SparkPlanGenerator.java#L202].
> The problem is IOContext seems not initialized correctly in the spark yarn
> client/cluster mode and caused the exception like
> {code}
> Job aborted due to stage failure: Task 93 in stage 0.0 failed 4 times, most
> recent failure: Lost task 93.3 in stage 0.0 (TID 616, bdpe48):
> java.lang.RuntimeException: Error processing row:
> java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:165)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:48)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:27)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveBaseFunctionResultList.hasNext(HiveBaseFunctionResultList.java:85)
> at
> scala.collection.convert.Wrappers$JIteratorWrapper.hasNext(Wrappers.scala:42)
> at
> org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.write(BypassMergeSortShuffleWriter.java:125)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:79)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:47)
> at org.apache.spark.scheduler.Task.run(Task.scala:85)
> at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:274)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:745)
> Caused by: java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.AbstractMapOperator.getNominalPath(AbstractMapOperator.java:101)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.cleanUpInputFileChangedOp(MapOperator.java:516)
> at
> org.apache.hadoop.hive.ql.exec.Operator.cleanUpInputFileChanged(Operator.java:1187)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.process(MapOperator.java:546)
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:152)
> ... 12 more
> Driver stacktrace:
> {code}
> in yarn client/cluster mode, sometimes
> [ExecMapperContext#currentInputPath|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/mr/ExecMapperContext.java#L109]
> is null when rdd cach is enabled.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
[jira] [Commented] (HIVE-18301) Investigate to enable MapInput cache in Hive on Spark
[
https://issues.apache.org/jira/browse/HIVE-18301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16301027#comment-16301027
]
liyunzhang commented on HIVE-18301:
---
[~xuefuz],[~lirui]: Is there any way to store IOcontext in some place and then
reinitialize it in current code when rdd cache is enabled?
> Investigate to enable MapInput cache in Hive on Spark
> -
>
> Key: HIVE-18301
> URL: https://issues.apache.org/jira/browse/HIVE-18301
> Project: Hive
> Issue Type: Bug
>Reporter: liyunzhang
>Assignee: liyunzhang
>
> Before IOContext problem is found in MapTran when spark rdd cache is enabled
> in HIVE-8920.
> so we disabled rdd cache in MapTran at
> [SparkPlanGenerator|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/spark/SparkPlanGenerator.java#L202].
> The problem is IOContext seems not initialized correctly in the spark yarn
> client/cluster mode and caused the exception like
> {code}
> Job aborted due to stage failure: Task 93 in stage 0.0 failed 4 times, most
> recent failure: Lost task 93.3 in stage 0.0 (TID 616, bdpe48):
> java.lang.RuntimeException: Error processing row:
> java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:165)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:48)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:27)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveBaseFunctionResultList.hasNext(HiveBaseFunctionResultList.java:85)
> at
> scala.collection.convert.Wrappers$JIteratorWrapper.hasNext(Wrappers.scala:42)
> at
> org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.write(BypassMergeSortShuffleWriter.java:125)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:79)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:47)
> at org.apache.spark.scheduler.Task.run(Task.scala:85)
> at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:274)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:745)
> Caused by: java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.AbstractMapOperator.getNominalPath(AbstractMapOperator.java:101)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.cleanUpInputFileChangedOp(MapOperator.java:516)
> at
> org.apache.hadoop.hive.ql.exec.Operator.cleanUpInputFileChanged(Operator.java:1187)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.process(MapOperator.java:546)
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:152)
> ... 12 more
> Driver stacktrace:
> {code}
> in yarn client/cluster mode, sometimes
> [ExecMapperContext#currentInputPath|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/mr/ExecMapperContext.java#L109]
> is null when rdd cach is enabled.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
[jira] [Commented] (HIVE-18301) Investigate to enable MapInput cache in Hive on Spark
[
https://issues.apache.org/jira/browse/HIVE-18301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16300999#comment-16300999
]
liyunzhang commented on HIVE-18301:
---
The reason why have this NPE is because
[org.apache.hadoop.hive.ql.io.IOContextMap#sparkThreadLocal|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/io/IOContextMap.java#L54
] is ThreadLocal variable.The value of the same ThreadLocal variable maybe
different in different threads.
we set the InputPath by
{code}
CombineHiveRecordReader#init
->HiveContextAwareRecordReader.initIOContext
->IOContext.setInputPath
{code}
we get the InputPath by
{code}
SparkMapRecordHandler.processRow
->MapOperator.process
->MapOperator.cleanUpInputFileChangedOp
->ExecMapperContext.getCurrentInputPath
->IOContext#getInputPath
{code}
when cache is enabled, there is no setInputPath , so return null when call
IOContext#getInputPath
> Investigate to enable MapInput cache in Hive on Spark
> -
>
> Key: HIVE-18301
> URL: https://issues.apache.org/jira/browse/HIVE-18301
> Project: Hive
> Issue Type: Bug
>Reporter: liyunzhang
>Assignee: liyunzhang
>
> Before IOContext problem is found in MapTran when spark rdd cache is enabled
> in HIVE-8920.
> so we disabled rdd cache in MapTran at
> [SparkPlanGenerator|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/spark/SparkPlanGenerator.java#L202].
> The problem is IOContext seems not initialized correctly in the spark yarn
> client/cluster mode and caused the exception like
> {code}
> Job aborted due to stage failure: Task 93 in stage 0.0 failed 4 times, most
> recent failure: Lost task 93.3 in stage 0.0 (TID 616, bdpe48):
> java.lang.RuntimeException: Error processing row:
> java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:165)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:48)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:27)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveBaseFunctionResultList.hasNext(HiveBaseFunctionResultList.java:85)
> at
> scala.collection.convert.Wrappers$JIteratorWrapper.hasNext(Wrappers.scala:42)
> at
> org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.write(BypassMergeSortShuffleWriter.java:125)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:79)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:47)
> at org.apache.spark.scheduler.Task.run(Task.scala:85)
> at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:274)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:745)
> Caused by: java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.AbstractMapOperator.getNominalPath(AbstractMapOperator.java:101)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.cleanUpInputFileChangedOp(MapOperator.java:516)
> at
> org.apache.hadoop.hive.ql.exec.Operator.cleanUpInputFileChanged(Operator.java:1187)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.process(MapOperator.java:546)
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:152)
> ... 12 more
> Driver stacktrace:
> {code}
> in yarn client/cluster mode, sometimes
> [ExecMapperContext#currentInputPath|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/mr/ExecMapperContext.java#L109]
> is null when rdd cach is enabled.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
[jira] [Commented] (HIVE-18301) Investigate to enable MapInput cache in Hive on Spark
[
https://issues.apache.org/jira/browse/HIVE-18301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16297897#comment-16297897
]
liyunzhang commented on HIVE-18301:
---
yes. I can achieve the merge the same tables into 1 mapInput and run
successfully such case(DS/query28) in HIVE-17486 in spark local mode. This
exception only happens in yarn mode.
> Investigate to enable MapInput cache in Hive on Spark
> -
>
> Key: HIVE-18301
> URL: https://issues.apache.org/jira/browse/HIVE-18301
> Project: Hive
> Issue Type: Bug
>Reporter: liyunzhang
>Assignee: liyunzhang
>
> Before IOContext problem is found in MapTran when spark rdd cache is enabled
> in HIVE-8920.
> so we disabled rdd cache in MapTran at
> [SparkPlanGenerator|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/spark/SparkPlanGenerator.java#L202].
> The problem is IOContext seems not initialized correctly in the spark yarn
> client/cluster mode and caused the exception like
> {code}
> Job aborted due to stage failure: Task 93 in stage 0.0 failed 4 times, most
> recent failure: Lost task 93.3 in stage 0.0 (TID 616, bdpe48):
> java.lang.RuntimeException: Error processing row:
> java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:165)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:48)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:27)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveBaseFunctionResultList.hasNext(HiveBaseFunctionResultList.java:85)
> at
> scala.collection.convert.Wrappers$JIteratorWrapper.hasNext(Wrappers.scala:42)
> at
> org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.write(BypassMergeSortShuffleWriter.java:125)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:79)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:47)
> at org.apache.spark.scheduler.Task.run(Task.scala:85)
> at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:274)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:745)
> Caused by: java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.AbstractMapOperator.getNominalPath(AbstractMapOperator.java:101)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.cleanUpInputFileChangedOp(MapOperator.java:516)
> at
> org.apache.hadoop.hive.ql.exec.Operator.cleanUpInputFileChanged(Operator.java:1187)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.process(MapOperator.java:546)
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:152)
> ... 12 more
> Driver stacktrace:
> {code}
> in yarn client/cluster mode, sometimes
> [ExecMapperContext#currentInputPath|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/mr/ExecMapperContext.java#L109]
> is null when rdd cach is enabled.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
[jira] [Commented] (HIVE-18301) Investigate to enable MapInput cache in Hive on Spark
[
https://issues.apache.org/jira/browse/HIVE-18301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16297894#comment-16297894
]
Rui Li commented on HIVE-18301:
---
If we can cache MapInput, will it be simpler to dynamically identify same
MapInputs and cache them, in order to achieve the purpose of HIVE-17486?
> Investigate to enable MapInput cache in Hive on Spark
> -
>
> Key: HIVE-18301
> URL: https://issues.apache.org/jira/browse/HIVE-18301
> Project: Hive
> Issue Type: Bug
>Reporter: liyunzhang
>Assignee: liyunzhang
>
> Before IOContext problem is found in MapTran when spark rdd cache is enabled
> in HIVE-8920.
> so we disabled rdd cache in MapTran at
> [SparkPlanGenerator|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/spark/SparkPlanGenerator.java#L202].
> The problem is IOContext seems not initialized correctly in the spark yarn
> client/cluster mode and caused the exception like
> {code}
> Job aborted due to stage failure: Task 93 in stage 0.0 failed 4 times, most
> recent failure: Lost task 93.3 in stage 0.0 (TID 616, bdpe48):
> java.lang.RuntimeException: Error processing row:
> java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:165)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:48)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveMapFunctionResultList.processNextRecord(HiveMapFunctionResultList.java:27)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveBaseFunctionResultList.hasNext(HiveBaseFunctionResultList.java:85)
> at
> scala.collection.convert.Wrappers$JIteratorWrapper.hasNext(Wrappers.scala:42)
> at
> org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.write(BypassMergeSortShuffleWriter.java:125)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:79)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:47)
> at org.apache.spark.scheduler.Task.run(Task.scala:85)
> at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:274)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:745)
> Caused by: java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.exec.AbstractMapOperator.getNominalPath(AbstractMapOperator.java:101)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.cleanUpInputFileChangedOp(MapOperator.java:516)
> at
> org.apache.hadoop.hive.ql.exec.Operator.cleanUpInputFileChanged(Operator.java:1187)
> at
> org.apache.hadoop.hive.ql.exec.MapOperator.process(MapOperator.java:546)
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkMapRecordHandler.processRow(SparkMapRecordHandler.java:152)
> ... 12 more
> Driver stacktrace:
> {code}
> in yarn client/cluster mode, sometimes
> [ExecMapperContext#currentInputPath|https://github.com/kellyzly/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/mr/ExecMapperContext.java#L109]
> is null when rdd cach is enabled.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
