You are viewing a plain text version of this content. The canonical link for it is here.
Posted to yarn-issues@hadoop.apache.org by "Arun Suresh (JIRA)" <ji...@apache.org> on 2016/05/26 04:52:12 UTC
[jira] [Comment Edited] (YARN-5035) FairScheduler: Adjust maxAssign
dynamically when assignMultiple is turned on
[ https://issues.apache.org/jira/browse/YARN-5035?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15301525#comment-15301525 ]
Arun Suresh edited comment on YARN-5035 at 5/26/16 4:52 AM:
------------------------------------------------------------
[~kasha], the patch looks pretty straightforward. Minor comments:
# In {{shouldContinueAssignment()}} :
{noformat}
if (maxAssignDynamic) {
return Resources.lessThan(DOMINANT_RESOURCE_CALCULATOR,
getClusterResource(), assignedResource, maxResourcesToAssign);
....
{noformat}
Dont you think you should use a {{Resources.fitsIn()}} check rather than the DRF less than ?
# I think we should make the *0.5f* configurable ?
was (Author: asuresh):
[~kasha], the patch looks pretty straightforward. Minor comments:
# In {{shouldContinueAssignment()}} :
{noformat}
if (maxAssignDynamic) {
return Resources.lessThan(DOMINANT_RESOURCE_CALCULATOR,
getClusterResource(), assignedResource, maxResourcesToAssign);
....
{noformat}
Don't you feel, dont you think you should use a {{Resources.fitsIn()}} check rather than the DRF less than ?
# I think we should make the *0.5f* configurable ?
> FairScheduler: Adjust maxAssign dynamically when assignMultiple is turned on
> ----------------------------------------------------------------------------
>
> Key: YARN-5035
> URL: https://issues.apache.org/jira/browse/YARN-5035
> Project: Hadoop YARN
> Issue Type: Improvement
> Components: fairscheduler
> Affects Versions: 2.8.0
> Reporter: Karthik Kambatla
> Assignee: Karthik Kambatla
> Attachments: yarn-5035-1.patch, yarn-5035-2.patch
>
>
> When assignMultiple is turned on, today the number of containers allocated to node within a single heartbeat is determined by maxAssign. Picking a higher value for this prevents the spreading of the workload, and a smaller number leads to slow ramp up.
> The proposal is to make it dynamic - half of the unallocated resources on the node.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: yarn-issues-help@hadoop.apache.org