[
https://issues.apache.org/jira/browse/MAPREDUCE-3259?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13136642#comment-13136642
]
Kihwal Lee commented on MAPREDUCE-3259:
---------------------------------------
Sorry, my explanation is rather wrong and confusing. I shouldn't have said
app-level. That's totally wrong.
Apps are fine as you said. It's about yarn system level env. When a JVM is
directly launched by the LTC, neither of LD_LIBRARY_PATH or java.library.path
is set. This is the case for ContainerLocalizer. It will fail to load
libhadoop.so if it needs it. If the library is installed in one of the
directories ld looks at, it will work without the setting. Or the hadoop lib
path may be added to /etc/ld.so.conf or /etc/ld.so.conf.d. Is it still
confusing? Or am I totally wrong about this?
I can leave out the second part in the patch for the client since a proper env
can be placed in the container script using the facility you mentioned. Only
thing I don't like is the default, which assumes the old dir layout. I could
try changing the default so that it has a better chance of working right out of
the box, but that's not critical.
> ContainerLocalizer should get the proper java.library.path from
> LinuxContainerExecutor
> --------------------------------------------------------------------------------------
>
> Key: MAPREDUCE-3259
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-3259
> Project: Hadoop Map/Reduce
> Issue Type: Bug
> Components: mrv2, nodemanager
> Affects Versions: 0.23.0, 0.24.0
> Reporter: Kihwal Lee
> Assignee: Kihwal Lee
> Priority: Blocker
> Fix For: 0.23.0, 0.24.0
>
> Attachments: mapreduce-3259.patch, mapreduce-3259.patch.txt
>
>
> As seen in MAPREDUCE-2915, java.library.path is not being passed when the LCE
> spawns a JVM for ContainerLocalizer.
> However, unlike branch-0.20-security, the task runtime in 0.23 is unaffected
> by this. This is because tasks' run-time environment is specified in the
> launch script by client. Setting LD_LIBRARY_PATH is the primary way of
> specifying the locations of required native library in this case. The config
> property, mapreduce.admin.user.env is always set in the job environment and
> the default value is to add the path to the hadoop native library to
> LD_LABRARY_PATH.
> For JVM's being launched by the hadoop system scripts, java.library.path is
> set.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
