hadoop-yarn-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Daniel Templeton (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (YARN-6307) Refactor FairShareComparator#compare
Date Mon, 24 Jul 2017 21:15:00 GMT

    [ https://issues.apache.org/jira/browse/YARN-6307?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16099134#comment-16099134

Daniel Templeton commented on YARN-6307:

Nice patch, [~yufeigu].  Here are my comments:

# Since we're unnesting all the _if_ blocks, let's take do that here, too: {code}      if
(res == 0) {
        // Apps are tied in fairness ratio. Break the tie by submit time and job
        // name to get a deterministic ordering, which is useful for unit tests.
        res = (int) Math.signum(s1.getStartTime() - s2.getStartTime());
        if (res == 0) {
          res = s1.getName().compareTo(s2.getName());
# Let's not stack the declarations: {code}      double useToWeightRatio1, useToWeightRatio2;{code}

Otherwise, looks good.

> Refactor FairShareComparator#compare
> ------------------------------------
>                 Key: YARN-6307
>                 URL: https://issues.apache.org/jira/browse/YARN-6307
>             Project: Hadoop YARN
>          Issue Type: Bug
>          Components: fairscheduler
>            Reporter: Yufei Gu
>            Assignee: Yufei Gu
>         Attachments: YARN-6307.001.patch, YARN-6307.002.patch
> The method does three things: compare the min share usage, compare fair share usage by
checking weight ratio, break tied by submit time and name. They are mixed with each other
which is not easy to read and maintenance, poor style. Additionally, there are potential performance
issues, like no need to check weight ratio if minShare usage comparison already indicate the
order. It is worth to improve considering huge amount invokings in scheduler.

This message was sent by Atlassian JIRA

To unsubscribe, e-mail: yarn-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: yarn-issues-help@hadoop.apache.org

View raw message