[
https://issues.apache.org/jira/browse/YARN-4731?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Varun Vasudev updated YARN-4731:
--------------------------------
Attachment: YARN-4731.001.patch
Root cause here is that we are using fstat on an open fd. The open call follows
the symlink and we stat the directory pointed to by the symlink instead of the
actual symlink. As a result rmdir fails because it doesn't delete symlinks.
The other problem with following symlinks in our case is that we end up
deleting public resources because we use symlinks in the container work dir to
point to the actual resources.
I've attached a patch to not follow symlinks and just call unlink on the
symlink itself.
[~jlowe] - can you please take a look?
> Linux container executor fails to delete nmlocal folders
> --------------------------------------------------------
>
> Key: YARN-4731
> URL: https://issues.apache.org/jira/browse/YARN-4731
> Project: Hadoop YARN
> Issue Type: Bug
> Reporter: Bibin A Chundatt
> Assignee: Varun Vasudev
> Priority: Critical
> Attachments: YARN-4731.001.patch
>
>
> Enable LCE and CGroups
> Submit a mapreduce job
> {noformat}
> 2016-02-24 18:56:46,889 INFO
> org.apache.hadoop.yarn.server.nodemanager.LinuxContainerExecutor: Deleting
> absolute path :
> /opt/bibin/dsperf/HAINSTALL/nmlocal/usercache/dsperf/appcache/application_1456319010019_0003/container_e02_1456319010019_0003_01_000001
> 2016-02-24 18:56:46,894 WARN
> org.apache.hadoop.yarn.server.nodemanager.containermanager.linux.privileged.PrivilegedOperationExecutor:
> Shell execution returned exit code: 255. Privileged Execution Operation
> Output:
> main : command provided 3
> main : run as user is dsperf
> main : requested yarn user is dsperf
> failed to rmdir job.jar: Not a directory
> Error while deleting
> /opt/bibin/dsperf/HAINSTALL/nmlocal/usercache/dsperf/appcache/application_1456319010019_0003/container_e02_1456319010019_0003_01_000001:
> 20 (Not a directory)
> Full command array for failed execution:
> [/opt/bibin/dsperf/HAINSTALL/install/hadoop/nodemanager/bin/container-executor,
> dsperf, dsperf, 3,
> /opt/bibin/dsperf/HAINSTALL/nmlocal/usercache/dsperf/appcache/application_1456319010019_0003/container_e02_1456319010019_0003_01_000001]
> 2016-02-24 18:56:46,894 ERROR
> org.apache.hadoop.yarn.server.nodemanager.LinuxContainerExecutor:
> DeleteAsUser for
> /opt/bibin/dsperf/HAINSTALL/nmlocal/usercache/dsperf/appcache/application_1456319010019_0003/container_e02_1456319010019_0003_01_000001
> returned with exit code: 255
> org.apache.hadoop.yarn.server.nodemanager.containermanager.linux.privileged.PrivilegedOperationException:
> ExitCodeException exitCode=255:
> at
> org.apache.hadoop.yarn.server.nodemanager.containermanager.linux.privileged.PrivilegedOperationExecutor.executePrivilegedOperation(PrivilegedOperationExecutor.java:173)
> at
> org.apache.hadoop.yarn.server.nodemanager.containermanager.linux.privileged.PrivilegedOperationExecutor.executePrivilegedOperation(PrivilegedOperationExecutor.java:199)
> at
> org.apache.hadoop.yarn.server.nodemanager.LinuxContainerExecutor.deleteAsUser(LinuxContainerExecutor.java:569)
> at
> org.apache.hadoop.yarn.server.nodemanager.DeletionService$FileDeletionTask.run(DeletionService.java:265)
> at
> java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
> at java.util.concurrent.FutureTask.run(FutureTask.java:266)
> at
> java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$201(ScheduledThreadPoolExecutor.java:180)
> at
> java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:293)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:745)
> Caused by: ExitCodeException exitCode=255:
> at org.apache.hadoop.util.Shell.runCommand(Shell.java:927)
> at org.apache.hadoop.util.Shell.run(Shell.java:838)
> at
> org.apache.hadoop.util.Shell$ShellCommandExecutor.execute(Shell.java:1117)
> at
> org.apache.hadoop.yarn.server.nodemanager.containermanager.linux.privileged.PrivilegedOperationExecutor.executePrivilegedOperation(PrivilegedOperationExecutor.java:150)
> ... 10 more
> {noformat}
> As a result nodemanager-local directory are not getting deleted for each
> application
> {noformat}
> total 36
> drwxr-s--- 4 hdfs hadoop 4096 Feb 25 08:25 ./
> drwxr-s--- 7 hdfs hadoop 4096 Feb 25 08:25 ../
> -rw------- 1 hdfs hadoop 340 Feb 25 08:25 container_tokens
> lrwxrwxrwx 1 hdfs hadoop 111 Feb 25 08:25 job.jar ->
> /opt/bibin/dsperf/HAINSTALL/nmlocal/usercache/hdfs/appcache/application_1456364845478_0004/filecache/11/job.jar/
> lrwxrwxrwx 1 hdfs hadoop 111 Feb 25 08:25 job.xml ->
> /opt/bibin/dsperf/HAINSTALL/nmlocal/usercache/hdfs/appcache/application_1456364845478_0004/filecache/13/job.xml*
> drwxr-s--- 2 hdfs hadoop 4096 Feb 25 08:25 jobSubmitDir/
> -rwx------ 1 hdfs hadoop 5348 Feb 25 08:25 launch_container.sh*
> drwxr-s--- 2 hdfs hadoop 4096 Feb 25 08:25 tmp/
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
