[
https://issues.apache.org/jira/browse/HIVE-12229?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14991060#comment-14991060
]
Rui Li commented on HIVE-12229:
-------------------------------
I'll use some examples to explain my point. Let's consider the following case:
1. User adds A.jar from hive and runs some query.
2. User adds(updates) a different A.jar and run some query but the query fails
because spark.files.overwrite is false.
3. User sets spark.files.overwrite to true and run that failed query again.
But at this point, we'll create a new RSC as well as new executors. And these
new executors have no jars added by now. Then we add the two A.jar to RSC and
the second added A.jar will overwrite the first, and eventually get distributed
to the executors. But IMO the order of the two A.jar is not deterministic
because added jars are retrieved from {{Utilities.getResourceFiles}} and added
by an async RPC job.
Now let's consider another case:
1. User sets spark.files.overwrite to true, adds A.jar and runs some queries.
2. User adds another A.jar and runs another query.
At this point, the former A.jar is already added so we won't add it again and
just add the new A.jar to RSC. RSC ships the new A.jar to executors and
successfully overwrites the existing A.jar. Then the query will use the updated
jar, which is the expected behavior.
> Custom script in query cannot be executed in yarn-cluster mode [Spark Branch].
> ------------------------------------------------------------------------------
>
> Key: HIVE-12229
> URL: https://issues.apache.org/jira/browse/HIVE-12229
> Project: Hive
> Issue Type: Bug
> Components: Spark
> Affects Versions: 1.1.0
> Reporter: Lifeng Wang
> Assignee: Rui Li
> Attachments: HIVE-12229.1-spark.patch, HIVE-12229.2-spark.patch,
> HIVE-12229.3-spark.patch, HIVE-12229.3-spark.patch
>
>
> Added one python script in the query and the python script cannot be found
> during execution in yarn-cluster mode.
> {noformat}
> 15/10/21 21:10:55 INFO exec.ScriptOperator: Executing [/usr/bin/python,
> q2-sessionize.py, 3600]
> 15/10/21 21:10:55 INFO exec.ScriptOperator: tablename=null
> 15/10/21 21:10:55 INFO exec.ScriptOperator: partname=null
> 15/10/21 21:10:55 INFO exec.ScriptOperator: alias=null
> 15/10/21 21:10:55 INFO spark.SparkRecordHandler: processing 10 rows: used
> memory = 324896224
> 15/10/21 21:10:55 INFO exec.ScriptOperator: ErrorStreamProcessor calling
> reporter.progress()
> /usr/bin/python: can't open file 'q2-sessionize.py': [Errno 2] No such file
> or directory
> 15/10/21 21:10:55 INFO exec.ScriptOperator: StreamThread OutputProcessor done
> 15/10/21 21:10:55 INFO exec.ScriptOperator: StreamThread ErrorProcessor done
> 15/10/21 21:10:55 INFO spark.SparkRecordHandler: processing 100 rows: used
> memory = 325619920
> 15/10/21 21:10:55 ERROR exec.ScriptOperator: Error in writing to script:
> Stream closed
> 15/10/21 21:10:55 INFO exec.ScriptOperator: The script did not consume all
> input data. This is considered as an error.
> 15/10/21 21:10:55 INFO exec.ScriptOperator: set
> hive.exec.script.allow.partial.consumption=true; to ignore it.
> 15/10/21 21:10:55 ERROR spark.SparkReduceRecordHandler: Fatal error:
> org.apache.hadoop.hive.ql.metadata.HiveException: Error while processing row
> (tag=0)
> {"key":{"reducesinkkey0":2,"reducesinkkey1":3316240655},"value":{"_col0":5529}}
> org.apache.hadoop.hive.ql.metadata.HiveException: Error while processing row
> (tag=0)
> {"key":{"reducesinkkey0":2,"reducesinkkey1":3316240655},"value":{"_col0":5529}}
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkReduceRecordHandler.processKeyValues(SparkReduceRecordHandler.java:340)
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkReduceRecordHandler.processRow(SparkReduceRecordHandler.java:289)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveReduceFunctionResultList.processNextRecord(HiveReduceFunctionResultList.java:49)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveReduceFunctionResultList.processNextRecord(HiveReduceFunctionResultList.java:28)
> at
> org.apache.hadoop.hive.ql.exec.spark.HiveBaseFunctionResultList$ResultIterator.hasNext(HiveBaseFunctionResultList.java:95)
> at
> scala.collection.convert.Wrappers$JIteratorWrapper.hasNext(Wrappers.scala:41)
> at
> org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.insertAll(BypassMergeSortShuffleWriter.java:99)
> at
> org.apache.spark.shuffle.sort.SortShuffleWriter.write(SortShuffleWriter.scala:73)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:73)
> at
> org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:41)
> at org.apache.spark.scheduler.Task.run(Task.scala:88)
> at
> org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:214)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615)
> at java.lang.Thread.run(Thread.java:745)
> Caused by: org.apache.hadoop.hive.ql.metadata.HiveException: [Error 20001]:
> An error occurred while reading or writing to your custom script. It may have
> crashed with an error.
> at
> org.apache.hadoop.hive.ql.exec.ScriptOperator.processOp(ScriptOperator.java:453)
> at org.apache.hadoop.hive.ql.exec.Operator.forward(Operator.java:815)
> at
> org.apache.hadoop.hive.ql.exec.SelectOperator.processOp(SelectOperator.java:84)
> at
> org.apache.hadoop.hive.ql.exec.spark.SparkReduceRecordHandler.processKeyValues(SparkReduceRecordHandler.java:331)
> ... 14 more
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
