[
https://issues.apache.org/jira/browse/YARN-1040?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15167674#comment-15167674
]
Arun Suresh commented on YARN-1040:
-----------------------------------
Thanks for clarifying [~bikassaha]
I propose we break this down into sub-jiras :
# New APIs specific to delinking container life-cycle from the process : This
will include 4 of new APIs specified above (excluding localization) but will
assume single process (and this the startProcess does not need to return a
processId)
# Add support for clubbing APIs into a single RPC
** Might have to think a bit about validating the order and multiplicity of the
API calls in each command (which I expect might be different for single process
/ multiple processes)
# Add support for localize API
# Add support for Multiple processes
** A processId will be returned for a startProcess. Might have to think thru
this further. for eg. how does this integrate with YARN-4725
For the purpose of Application Upgrades (for which this JIRA is marked as a
sub-task of... also why im calling it out specifically) : Add support for
Container Upgrades
# Expose a canned NMCommand that has the list of APIs to upgrade based on some
policy
If folks are fine with this, I will ahead and open JIRAs and link this issue to
each of the above JIRAs (Since I don't think I can create subtasks for this
JIRA) so that we can start work on the same..
> De-link container life cycle from the process and add ability to execute
> multiple processes in the same long-lived container
> ----------------------------------------------------------------------------------------------------------------------------
>
> Key: YARN-1040
> URL: https://issues.apache.org/jira/browse/YARN-1040
> Project: Hadoop YARN
> Issue Type: Sub-task
> Components: nodemanager
> Affects Versions: 3.0.0
> Reporter: Steve Loughran
>
> The AM should be able to exec >1 process in a container, rather than have the
> NM automatically release the container when the single process exits.
> This would let an AM restart a process on the same container repeatedly,
> which for HBase would offer locality on a restarted region server.
> We may also want the ability to exec multiple processes in parallel, so that
> something could be run in the container while a long-lived process was
> already running. This can be useful in monitoring and reconfiguring the
> long-lived process, as well as shutting it down.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
