[
https://issues.apache.org/jira/browse/HADOOP-11656?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14345489#comment-14345489
]
Vinod Kumar Vavilapalli commented on HADOOP-11656:
--------------------------------------------------
Back to the classpath issue. Here's what we have today
- With YARN, the platform classpath (YARN server) is already decoupled from
apps.
- With Hadoop 2.6, MapReduce framework is no longer needed to be installed on
each machine (MAPREDUCE-4421). The recommended way now is via
distributed-cache. This is a big help for rolling-upgrades too.
- Other frameworks like Apache Tez had always resorted to deployment via
distributed-cache.
- Given the above, and coupled with HDFS-6200, apps are now only exposed to
Hadoop client artifacts. The servers can evolved completely independently in
terms of what their dependencies are.
What we are left with is the isolation of the dependencies of Hadoop client
artifacts. This can be done by one of the following
- Shading hadoop-client artifacts: If this works, I don't have arguments
against it. This works both inside containers as well as inside client
processes on gateways.
- Use the ApplicationClassLoader: This already exists, but only works for
containers running inside the cluster.
The above should already solve our classpath isolation problem?
> Classpath isolation for downstream clients
> ------------------------------------------
>
> Key: HADOOP-11656
> URL: https://issues.apache.org/jira/browse/HADOOP-11656
> Project: Hadoop Common
> Issue Type: New Feature
> Reporter: Sean Busbey
> Assignee: Sean Busbey
> Labels: classloading, classpath, dependencies
>
> Currently, Hadoop exposes downstream clients to a variety of third party
> libraries. As our code base grows and matures we increase the set of
> libraries we rely on. At the same time, as our user base grows we increase
> the likelihood that some downstream project will run into a conflict while
> attempting to use a different version of some library we depend on. This has
> already happened with i.e. Guava several times for HBase, Accumulo, and Spark
> (and I'm sure others).
> While YARN-286 and MAPREDUCE-1700 provided an initial effort, they default to
> off and they don't do anything to help dependency conflicts on the driver
> side or for folks talking to HDFS directly. This should serve as an umbrella
> for changes needed to do things thoroughly on the next major version.
> We should ensure that downstream clients
> 1) can depend on a client artifact for each of HDFS, YARN, and MapReduce that
> doesn't pull in any third party dependencies
> 2) only see our public API classes (or as close to this as feasible) when
> executing user provided code, whether client side in a launcher/driver or on
> the cluster in a container or within MR.
> This provides us with a double benefit: users get less grief when they want
> to run substantially ahead or behind the versions we need and the project is
> freer to change our own dependency versions because they'll no longer be in
> our compatibility promises.
> Project specific task jiras to follow after I get some justifying use cases
> written in the comments.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
