[
https://issues.apache.org/jira/browse/YARN-6496?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15985238#comment-15985238
]
Varun Saxena commented on YARN-6496:
------------------------------------
[~rohithsharma], these are still effectively kV-pairs.
Can't we flatten out the JSON representation to send key-value pairs. What I
mean is KVs' in example above can be
{code}
properties.rohith.test.properties: "inside-properties"
env.NUM_SHARDS: "insde-env"
files.type: "HADOOP_XML_TEMPLATE",
files.src_file: "hdfs://yclouddev/tmp/conf/core-site.xml",
files.dest_file: "/etc/hadoop/conf/core-site.xml"
{code}
Do you have to reconstruct the JSON representation after reading it back from
ATSv2 Reader?
> Rethink on configurations published into ATSv2
> ----------------------------------------------
>
> Key: YARN-6496
> URL: https://issues.apache.org/jira/browse/YARN-6496
> Project: Hadoop YARN
> Issue Type: Sub-task
> Reporter: Rohith Sharma K S
> Assignee: Rohith Sharma K S
>
> Slider configuration model is bit different from Hadoop. Slider has
> configuration object with several fields differentiations such as properties,
> environment and files that has more functionality such as fileName, fileType,
> srcFile, destFile.
> {code}
> "configuration" : {
> "properties": {
> "rohith.test.properties": "inside-properties"
> },
> "env": {
> "NUM_SHARDS": "insde-env"
> }
> "files" : [
> {
> "type": "HADOOP_XML_TEMPLATE",
> "src_file": "hdfs://yclouddev/tmp/conf/core-site.xml",
> "dest_file": "/etc/hadoop/conf/core-site.xml"
> }]
> {code}
> Timeline entity config is modeled as a flattened map of String to String,
> that is not flexible enough to satisfy above use-case. This need to rethink
> how can it be modeled.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
