[ https://issues.apache.org/jira/browse/YARN-5180?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Arun Suresh updated YARN-5180: ------------------------------ Attachment: YARN-5180.005.patch Updating patch.. thanks for the rev [~kasha].. bq. Can we get #2 of my previous comment fixed. And, by constructor, I think I was referring to the newInstance method. Oops.. missed that bq. Given there are only 4 possibilities of ExecutionTypeRequest, should we just hardcode them to avoid as many objects as container requests? Would this be premature optimization? Thought of that.. then decided against it, considering we might introduce more types later on, which I guess was the justification for having a wrapper type on the request side.. right ? > Add ensureExecutionType boolean flag in ResourceRequest > ------------------------------------------------------- > > Key: YARN-5180 > URL: https://issues.apache.org/jira/browse/YARN-5180 > Project: Hadoop YARN > Issue Type: Improvement > Reporter: Arun Suresh > Assignee: Arun Suresh > Attachments: YARN-5180.001.patch, YARN-5180.002.patch, > YARN-5180.003.patch, YARN-5180.004.patch, YARN-5180.005.patch > > > YARN-2882 introduced the concept of *ExecutionTypes*. > YARN-4335 allowed AMs to specify the ExecutionType in the ResourceRequest. > This JIRA proposes to add a boolean flag to the ResourceRequest to signal to > the Scheduler that the AM is fine receiving a Container with a different > Execution type than what is asked. -- This message was sent by Atlassian JIRA (v6.3.4#6332) --------------------------------------------------------------------- To unsubscribe, e-mail: yarn-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: yarn-issues-h...@hadoop.apache.org