[ 
https://issues.apache.org/jira/browse/OOZIE-2687?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16126555#comment-16126555
 ] 

Attila Sasvari commented on OOZIE-2687:
---------------------------------------

If we extract launcher / common properties to as separate xsd with 
{{targetNamespace="uri:oozie:common:0.1}} then we will need to use a workflow 
xml like:

{code}
<workflow-app xmlns="uri:oozie:workflow:0.6" 
xmlns:common="uri:oozie:common:0.1" name="java-main-wf">
    <global>
        <common:launcher>
            <common:memory>1024</common:memory>
            <common:vcores>2</common:vcores>
            <common:java-opts>dummyJavaOpts</common:java-opts>
            <common:env>env1=value1:env2=value2</common:env>
            <common:queue>dummyQueue</common:queue>
            <common:sharelib>a,b,c</common:sharelib>
        </common:launcher>
    </global>
    <start to="java-node"/>
    <action name="java-node">
...
{code}
In other action definition, where we intend to include {{launcher}} properties, 
 the {{common}} prefix will be needed too. It is similar to the {{sla}} schema.

I can update LiteWorkflowAppParser in the next patch so that it uses the 
namespace with the {{common}} prefix when dealing with launcher configuration 
or we can create a new JIRA and implement common xsd extraction there after a 
first working version that allows to specify queue, memory, vcores, java-opts, 
etc for the launcher job. 

> Create XML schema for launcher configurations
> ---------------------------------------------
>
>                 Key: OOZIE-2687
>                 URL: https://issues.apache.org/jira/browse/OOZIE-2687
>             Project: Oozie
>          Issue Type: Sub-task
>    Affects Versions: 5.0.0
>            Reporter: Peter Cseh
>            Assignee: Attila Sasvari
>            Priority: Critical
>         Attachments: OOZIE-2687-001.patch
>
>
> It was possible to manipulate the MapperLauncher's environment through 
> properties like:
> # mapreduce.map.memory.mb     
> # mapreduce.map.cpu.vcores
> # mapred.child.env
> # mapred.child.java.opts      
> # mapred.job.queue.name  - ability to set launcher queue
> E.g. We were using mapred.child.env to pass SPARK_HOME to the LauncherMapper 
> and make PySpark work. 
> Fixing OOZIE-2596 added a hack. We should decide how we support or break 
> compatibility and how we allow the manipulation of the Launcher environment.
> Verify if the new launcher section in global applies to actions in 
> sub-workflows as well. It did not use to work before and was only fixed in 
> OOZIE-2030. It would be good to have that testcase 
> (TestSubWorkflowActionExecutor. testParentGlobalConf) updated with the new 
> launcher section as well.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

Reply via email to