You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@spark.apache.org by "guifeng (JIRA)" <ji...@apache.org> on 2017/02/17 09:51:42 UTC
[jira] [Comment Edited] (SPARK-19645) structured streaming job
restart bug
[ https://issues.apache.org/jira/browse/SPARK-19645?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15871552#comment-15871552 ]
guifeng edited comment on SPARK-19645 at 2/17/17 9:50 AM:
----------------------------------------------------------
But, spark mater don't set rename options that support overwrite, so I think rename will failed also.
was (Author: guifengleaf@gmail.com):
But, spark mater don't set rename options that support overwrite.
> structured streaming job restart bug
> ------------------------------------
>
> Key: SPARK-19645
> URL: https://issues.apache.org/jira/browse/SPARK-19645
> Project: Spark
> Issue Type: Bug
> Components: Structured Streaming
> Affects Versions: 2.1.0
> Reporter: guifeng
> Priority: Critical
>
> We are trying to use Structured Streaming in product, however currently there exists a bug refer to the process of streaming job restart.
> The following is the concrete error message:
> {quote}
> Caused by: java.lang.IllegalStateException: Error committing version 2 into HDFSStateStore[id = (op=0, part=136), dir = /tmp/Pipeline_112346-continueagg-bxaxs/state/0/136]
> at org.apache.spark.sql.execution.streaming.state.HDFSBackedStateStoreProvider$HDFSBackedStateStore.commit(HDFSBackedStateStoreProvider.scala:162)
> at org.apache.spark.sql.execution.streaming.StateStoreSaveExec$$anonfun$doExecute$3.apply(StatefulAggregate.scala:173)
> at org.apache.spark.sql.execution.streaming.StateStoreSaveExec$$anonfun$doExecute$3.apply(StatefulAggregate.scala:123)
> at org.apache.spark.sql.execution.streaming.state.StateStoreRDD.compute(StateStoreRDD.scala:64)
> at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:323)
> at org.apache.spark.rdd.RDD.iterator(RDD.scala:287)
> at org.apache.spark.rdd.MapPartitionsRDD.compute(MapPartitionsRDD.scala:38)
> at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:323)
> at org.apache.spark.rdd.RDD.iterator(RDD.scala:287)
> at org.apache.spark.scheduler.ResultTask.runTask(ResultTask.scala:87)
> at org.apache.spark.scheduler.Task.run(Task.scala:99)
> at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:282)
> at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:745)
> Caused by: java.io.IOException: Failed to rename /tmp/Pipeline_112346-continueagg-bxaxs/state/0/136/temp--5345709896617324284 to /tmp/Pipeline_112346-continueagg-bxaxs/state/0/136/2.delta
> at org.apache.spark.sql.execution.streaming.state.HDFSBackedStateStoreProvider.org$apache$spark$sql$execution$streaming$state$HDFSBackedStateStoreProvider$$commitUpdates(HDFSBackedStateStoreProvider.scala:259)
> at org.apache.spark.sql.execution.streaming.state.HDFSBackedStateStoreProvider$HDFSBackedStateStore.commit(HDFSBackedStateStoreProvider.scala:156)
> ... 14 more
> {quote}
> The bug can be easily reproduce when restart previous streaming job, and the main reason is that when restart streaming job spark will recompute WAL offsets and generate the same hdfs delta file(latest delta file generated before restart and named "currentBatchId.delta") . In my opinion, this is a bug. If you guy consider that this is a bug also, I can fix it.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org