[
https://issues.apache.org/jira/browse/YARN-5079?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15298805#comment-15298805
]
Gour Saha commented on YARN-5079:
---------------------------------
[~kasha], thank you for your comments. Let me try to address few of the
Slider-related questions.
bq. 1. For folks who don't know Slider all that well, it would help to
enumerate the components of Slider and a super brief description of what they
do.
In one short sentence, Slider is a Universal Application Master. To add a
little more, Slider is a collection of tools and technologies to package,
deploy, and manage services or long running applications on Apache Hadoop YARN
clusters. It is the fastest way to deploy any non-YARN-enabled applications in
a YARN cluster. It is comprised of 3 primary components - Slider AM, Slider
Client and Slider Agent. It also contains sample application packages, few
written by the Slider committers and others donated by the wider community.
For detailed documentation, one can start here:
http://slider.incubator.apache.org
http://slider.incubator.apache.org/design/architecture.html
bq. 2. What parts among these are being proposed for a merge?
The following is a proposal:
Components to be merged: Slider Core – comprising of AM and Client code, and
the corresponding tests that back them
Components that will not be merged: Slider Agent
Components that need more deliberation: Slider App-Packages
One way to do this, is by:
a. Create a branch in Hadoop and move Slider AM, Client and tests into it
b. Create a branch in Slider which will have only Agent and App-Packages and
will point to Slider core (AM and Client) as Hadoop modules
bq. 3. What happens to the rest of the parts? Continue to live in Slider? Is
the slider community comfortable with that? From the thread Gour pointed to,
app packages seem like something that need more discussion.
There will be a sufficiently large window, when the current Slider project &
its releases and the newly migrated code in YARN, will coexist. During this
time, we will create v2 versions of App-package definitions and find their
final home in the future.
bq. 4. What release would we target for a Yarn-ified release of said Slider
components? Hadoop 3?
It is likely that we start, by contributing this code into a branch in Hadoop
and then work on an Agent-less or thin-Agent architecture. Subsequently, we can
include this in one of the early Hadoop-3 releases.
bq. 6. With respect to CLI and UIs, how flexible is Slider and the Slider
community with homogenizing with Yarn where applicable. I doubt if we have a
lot of this, but would be good to discuss.
Slider has an AM UI, which uses the YARN UI framework code and is already
navigable from the RM UI.
Currently, Slider has a separate CLI, but we can work towards homogenizing it.
> [Umbrella] Native YARN framework layer for services and beyond
> --------------------------------------------------------------
>
> Key: YARN-5079
> URL: https://issues.apache.org/jira/browse/YARN-5079
> Project: Hadoop YARN
> Issue Type: New Feature
> Reporter: Vinod Kumar Vavilapalli
> Assignee: Vinod Kumar Vavilapalli
>
> (See overview doc at YARN-4692, modifying and copy-pasting some of the
> relevant pieces and sub-section 3.3.1 to track the specific sub-item.)
> (This is a companion to YARN-4793 in our effort to simplify the entire story,
> but focusing on APIs)
> So far, YARN by design has restricted itself to having a very low-level API
> that can support any type of application. Frameworks like Apache Hadoop
> MapReduce, Apache Tez, Apache Spark, Apache REEF, Apache Twill, Apache Helix
> and others ended up exposing higher level APIs that end-users can directly
> leverage to build their applications on top of YARN. On the services side,
> Apache Slider has done something similar.
> With our current attention on making services first-class and simplified,
> it's time to take a fresh look at how we can make Apache Hadoop YARN support
> services well out of the box. Beyond the functionality that I outlined in the
> previous sections in the doc on how NodeManagers can be enhanced to help
> services, the biggest missing piece is the framework itself. There is a lot
> of very important functionality that a services' framework can own together
> with YARN in executing services end-to-end.
> In this JIRA I propose we look at having a native Apache Hadoop framework for
> running services natively on YARN.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
