[
https://issues.apache.org/jira/browse/HIVE-17374?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Federico De Giuli updated HIVE-17374:
-------------------------------------
Tags: hcatalog, hive, truncate (was: hcatalog)
Description:
It looks like truncating a table in hive does not remove partition directories
in the filesystem. This causes conflicts when HCatalog tries to write to the
same partition:
{{---------------------
Commit failed for output: outputName:scope-705 of vertex/vertexGroup:scope-708
isVertexGroupOutput:false, java.io.IOException:
java.lang.reflect.InvocationTargetException
at
org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.PigOutputCommitter.commitJob(PigOutputCommitter.java:281)
at
org.apache.pig.backend.hadoop.executionengine.tez.runtime.PigOutputFormatTez$PigOutputCommitterTez.commitJob(PigOutputFormatTez.java:98)
at
org.apache.tez.mapreduce.committer.MROutputCommitter.commitOutput(MROutputCommitter.java:99)
at org.apache.tez.dag.app.dag.impl.DAGImpl$1.run(DAGImpl.java:1018)
at org.apache.tez.dag.app.dag.impl.DAGImpl$1.run(DAGImpl.java:1015)
at java.security.AccessController.doPrivileged(Native Method)
at javax.security.auth.Subject.doAs(Subject.java:422)
at
org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1936)
at
org.apache.tez.dag.app.dag.impl.DAGImpl.commitOutput(DAGImpl.java:1015)
at org.apache.tez.dag.app.dag.impl.DAGImpl.access$2000(DAGImpl.java:149)
at org.apache.tez.dag.app.dag.impl.DAGImpl$3.call(DAGImpl.java:1094)
at org.apache.tez.dag.app.dag.impl.DAGImpl$3.call(DAGImpl.java:1089)
at java.util.concurrent.FutureTask.run(FutureTask.java:266)
at
java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
at
java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
at java.lang.Thread.run(Thread.java:745)
Caused by: java.lang.reflect.InvocationTargetException
at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
at
sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
at
sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
at java.lang.reflect.Method.invoke(Method.java:498)
at
org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.PigOutputCommitter.commitJob(PigOutputCommitter.java:279)
... 15 more
Caused by: org.apache.hive.hcatalog.common.HCatException : 2012 : Moving of
data failed during commit : Failed to move file:
hdfs://HOSTNAME/user/fdegiuli/db/video_sessions/_DYN0.6589663085145253/dt=2017080500
to hdfs://HOSTNAME/user/fdegiuli/db/video_sessions
at
org.apache.hive.hcatalog.mapreduce.FileOutputCommitterContainer.moveTaskOutputs(FileOutputCommitterContainer.java:825)
at
org.apache.hive.hcatalog.mapreduce.FileOutputCommitterContainer.moveTaskOutputs(FileOutputCommitterContainer.java:782)
at
org.apache.hive.hcatalog.mapreduce.FileOutputCommitterContainer.registerPartitions(FileOutputCommitterContainer.java:1176)
at
org.apache.hive.hcatalog.mapreduce.FileOutputCommitterContainer.commitJob(FileOutputCommitterContainer.java:272)
... 20 more
DAG did not succeed due to COMMIT_FAILURE. failedVertices:0 killedVertices:0}}
Here's a look at the directory. As you can see, {{dt=2017080500}} is empty.
{{[fdegiuli ~]$ hdfs dfs -ls $NN/user/fdegiuli/db/video_sessions/
Found 4 items
-rw------- 3 fdegiuli users 0 2017-08-16 15:48
NAMENODE/user/fdegiuli/db/video_sessions/_placeholder_0.016147276492671336
-rw------- 3 fdegiuli users 0 2017-08-15 21:19
NAMENODE/user/fdegiuli/db/video_sessions/_placeholder_0.15239090174209513
-rw------- 3 fdegiuli users 0 2017-08-16 14:49
NAMENODE/user/fdegiuli/db/video_sessions/_placeholder_0.7007380672193727
drwx------ - fdegiuli users 0 2017-08-16 15:44
NAMENODE/user/fdegiuli/db/video_sessions/dt=2017080500
[fdegiuli ~]$ hdfs dfs -ls $NN/user/fdegiuli/db/video_sessions/dt=2017080500
[fdegiuli ~]$}}
I'm using a pig script to repopulate the data in the table. I don't know if
this error happens when using Spark.
The way I see it, there's two possible bugs here:
1. TRUNCATE TABLE should delete the partition directories
2. HCatalog should handle existing partition directories correctly.
I'm happy to provide any more information you may want, and help with an
eventual patch.
Copied from
http://mail-archives.apache.org/mod_mbox/hive-user/201708.mbox/%3CCAFPScVjA8PYDXOajgJR8LgvypFd00eJDZuSrFVFCci1odt42_g%40mail.gmail.com%3E
> HCatalog storage failures after hive TRUNCATE
> ---------------------------------------------
>
> Key: HIVE-17374
> URL: https://issues.apache.org/jira/browse/HIVE-17374
> Project: Hive
> Issue Type: Bug
> Components: HCatalog, Hive
> Affects Versions: 1.2.2
> Environment: RHEL 6.8
> Hive 1.2.2.19.1707130004
> Reporter: Federico De Giuli
> Priority: Minor
>
> It looks like truncating a table in hive does not remove partition
> directories in the filesystem. This causes conflicts when HCatalog tries to
> write to the same partition:
> {{---------------------
> Commit failed for output: outputName:scope-705 of
> vertex/vertexGroup:scope-708 isVertexGroupOutput:false, java.io.IOException:
> java.lang.reflect.InvocationTargetException
> at
> org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.PigOutputCommitter.commitJob(PigOutputCommitter.java:281)
> at
> org.apache.pig.backend.hadoop.executionengine.tez.runtime.PigOutputFormatTez$PigOutputCommitterTez.commitJob(PigOutputFormatTez.java:98)
> at
> org.apache.tez.mapreduce.committer.MROutputCommitter.commitOutput(MROutputCommitter.java:99)
> at org.apache.tez.dag.app.dag.impl.DAGImpl$1.run(DAGImpl.java:1018)
> at org.apache.tez.dag.app.dag.impl.DAGImpl$1.run(DAGImpl.java:1015)
> at java.security.AccessController.doPrivileged(Native Method)
> at javax.security.auth.Subject.doAs(Subject.java:422)
> at
> org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1936)
> at
> org.apache.tez.dag.app.dag.impl.DAGImpl.commitOutput(DAGImpl.java:1015)
> at
> org.apache.tez.dag.app.dag.impl.DAGImpl.access$2000(DAGImpl.java:149)
> at org.apache.tez.dag.app.dag.impl.DAGImpl$3.call(DAGImpl.java:1094)
> at org.apache.tez.dag.app.dag.impl.DAGImpl$3.call(DAGImpl.java:1089)
> at java.util.concurrent.FutureTask.run(FutureTask.java:266)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:745)
> Caused by: java.lang.reflect.InvocationTargetException
> at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
> at
> sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
> at
> sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
> at java.lang.reflect.Method.invoke(Method.java:498)
> at
> org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.PigOutputCommitter.commitJob(PigOutputCommitter.java:279)
> ... 15 more
> Caused by: org.apache.hive.hcatalog.common.HCatException : 2012 : Moving of
> data failed during commit : Failed to move file:
> hdfs://HOSTNAME/user/fdegiuli/db/video_sessions/_DYN0.6589663085145253/dt=2017080500
> to hdfs://HOSTNAME/user/fdegiuli/db/video_sessions
> at
> org.apache.hive.hcatalog.mapreduce.FileOutputCommitterContainer.moveTaskOutputs(FileOutputCommitterContainer.java:825)
> at
> org.apache.hive.hcatalog.mapreduce.FileOutputCommitterContainer.moveTaskOutputs(FileOutputCommitterContainer.java:782)
> at
> org.apache.hive.hcatalog.mapreduce.FileOutputCommitterContainer.registerPartitions(FileOutputCommitterContainer.java:1176)
> at
> org.apache.hive.hcatalog.mapreduce.FileOutputCommitterContainer.commitJob(FileOutputCommitterContainer.java:272)
> ... 20 more
> DAG did not succeed due to COMMIT_FAILURE. failedVertices:0 killedVertices:0}}
> Here's a look at the directory. As you can see, {{dt=2017080500}} is empty.
> {{[fdegiuli ~]$ hdfs dfs -ls $NN/user/fdegiuli/db/video_sessions/
> Found 4 items
> -rw------- 3 fdegiuli users 0 2017-08-16 15:48
> NAMENODE/user/fdegiuli/db/video_sessions/_placeholder_0.016147276492671336
> -rw------- 3 fdegiuli users 0 2017-08-15 21:19
> NAMENODE/user/fdegiuli/db/video_sessions/_placeholder_0.15239090174209513
> -rw------- 3 fdegiuli users 0 2017-08-16 14:49
> NAMENODE/user/fdegiuli/db/video_sessions/_placeholder_0.7007380672193727
> drwx------ - fdegiuli users 0 2017-08-16 15:44
> NAMENODE/user/fdegiuli/db/video_sessions/dt=2017080500
> [fdegiuli ~]$ hdfs dfs -ls $NN/user/fdegiuli/db/video_sessions/dt=2017080500
> [fdegiuli ~]$}}
>
> I'm using a pig script to repopulate the data in the table. I don't know if
> this error happens when using Spark.
> The way I see it, there's two possible bugs here:
> 1. TRUNCATE TABLE should delete the partition directories
> 2. HCatalog should handle existing partition directories correctly.
> I'm happy to provide any more information you may want, and help with an
> eventual patch.
> Copied from
> http://mail-archives.apache.org/mod_mbox/hive-user/201708.mbox/%3CCAFPScVjA8PYDXOajgJR8LgvypFd00eJDZuSrFVFCci1odt42_g%40mail.gmail.com%3E
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
