[
https://issues.apache.org/jira/browse/HADOOP-11764?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14389755#comment-14389755
]
Zhijie Shen commented on HADOOP-11764:
--------------------------------------
bq. There are lots and lots of options here...
Yeah, among the lots of options, isn't it the most decent way to fix the env
issue in the scope where we work on evn?
bq. it becomes a real problem at extremely long lengths.
In most cases, we don't need to specify JAVA_LIBRARY_PATH. leveldb works fine
with the default path. In the case that is described above and that we need to
specify JAVA_LIBRARY_PATH to point to leveldb jni libary dir, the reasonable
path length should usually not exceed the arg max length. In addition, AFAIK,
ARG_MAX is not a small number (e.g., 2621440). JAVA_LIBRARY_PATH alone is not
that easy to exceed it, but it's true that there will be the extreme cases.
However, the point in my previous comment is that whether JAVA_LIBRARY_PATH is
set by users manually or automatically by the script, it has to be set to fix
leveldb JNI issue. That said, automatic setting JAVA_LIBRARY_PATH is *NOT*
generating longer args than users' manual setting. Your argument sounds more
like agains using JAVA_LIBRARY_PATH. If you have concern that JAVA_LIBRARY_PATH
will break ARG_MAX, please feel free to start a separate thread to discuss it.
This jira is focusing on fixing leveldb jni.
> Hadoop should have the option to use directory other than tmp for extracting
> and loading leveldbjni
> ---------------------------------------------------------------------------------------------------
>
> Key: HADOOP-11764
> URL: https://issues.apache.org/jira/browse/HADOOP-11764
> Project: Hadoop Common
> Issue Type: Bug
> Reporter: Anubhav Dhoot
> Assignee: Anubhav Dhoot
> Attachments: YARN-3331.001.patch, YARN-3331.002.patch
>
>
> /tmp can be required to be noexec in many environments. This causes a
> problem when nodemanager tries to load the leveldbjni library which can get
> unpacked and executed from /tmp.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
