Mailing-List: contact yarn-issues-help@hadoop.apache.org; run by ezmlm
Precedence: bulk
Date: Mon, 28 Aug 2017 22:35:01 +0000 (UTC)
From: "ASF GitHub Bot (JIRA)" <jira@apache.org>
To: yarn-issues@hadoop.apache.org
Message-ID: <JIRA.12721318.1402858766000.141458.1503959701801@Atlassian.JIRA>
In-Reply-To: <JIRA.12721318.1402858766000@Atlassian.JIRA>
References: <JIRA.12721318.1402858766000@Atlassian.JIRA> <JIRA.12721318.1402858766413@jira-lw-us.apache.org>
Subject: [jira] [Commented] (YARN-2162) add ability in Fair Scheduler to
 optionally configure maxResources in terms of percentage
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit
archived-at: Mon, 28 Aug 2017 22:35:09 -0000


    [ https://issues.apache.org/jira/browse/YARN-2162?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16144472#comment-16144472 ] 

ASF GitHub Bot commented on YARN-2162:
--------------------------------------

Github user kambatla commented on a diff in the pull request:

    https://github.com/apache/hadoop/pull/261#discussion_r135650575
  
    --- Diff: hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/scheduler/fair/FairSchedulerConfiguration.java ---
    @@ -287,27 +289,69 @@ public float getReservableNodes() {
        * 
        * @throws AllocationConfigurationException
        */
    -  public static Resource parseResourceConfigValue(String val)
    +  public static ConfigurableResource parseResourceConfigValue(String val)
           throws AllocationConfigurationException {
    +    ConfigurableResource configurableResource;
         try {
           val = StringUtils.toLowerCase(val);
    -      int memory = findResource(val, "mb");
    -      int vcores = findResource(val, "vcores");
    -      return BuilderUtils.newResource(memory, vcores);
    +      if (val.contains("%")) {
    +        configurableResource = new ConfigurableResource(
    +            getResourcePercentage(val));
    +      } else {
    +        int memory = findResource(val, "mb");
    +        int vcores = findResource(val, "vcores");
    +        configurableResource = new ConfigurableResource(
    +            BuilderUtils.newResource(memory, vcores));
    +      }
         } catch (AllocationConfigurationException ex) {
           throw ex;
         } catch (Exception ex) {
           throw new AllocationConfigurationException(
               "Error reading resource config", ex);
         }
    +    return configurableResource;
    +  }
    +
    +  private static double[] getResourcePercentage(
    +      String val) throws AllocationConfigurationException {
    +    double[] resourcePercentage = new double[ResourceType.values().length];
    +    String[] strings = val.split(",");
    +    if (strings.length == 1) {
    +      double percentage = findPercentage(strings[0], "");
    +      for (int i = 0 ; i < ResourceType.values().length ; i++) {
    +        resourcePercentage[i] = percentage/100;
    +      }
    +    } else {
    +      double memPercentage = findPercentage(val, "memory");
    +      double vcorePercentage = findPercentage(val, "cpu");
    +      resourcePercentage[0] = memPercentage/100;
    --- End diff --
    
    With ResourceTypes coming in, I feel this has to be handled better.  Loop over and use that index instead? 


> add ability in Fair Scheduler to optionally configure maxResources in terms of percentage
> -----------------------------------------------------------------------------------------
>
>                 Key: YARN-2162
>                 URL: https://issues.apache.org/jira/browse/YARN-2162
>             Project: Hadoop YARN
>          Issue Type: Improvement
>          Components: fairscheduler, scheduler
>            Reporter: Ashwin Shankar
>            Assignee: Yufei Gu
>              Labels: scheduler
>         Attachments: YARN-2162.001.patch, YARN-2162.002.patch, YARN-2162.003.patch
>
>
> minResources and maxResources in fair scheduler configs are expressed in terms of absolute numbers X mb, Y vcores. 
> As a result, when we expand or shrink our hadoop cluster, we need to recalculate and change minResources/maxResources accordingly, which is pretty inconvenient.
> We can circumvent this problem if we can optionally configure these properties in terms of percentage of cluster capacity. 


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

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