[ 
https://issues.apache.org/jira/browse/YARN-2681?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14613804#comment-14613804
 ] 

Nam H. Do commented on YARN-2681:
---------------------------------

+ The release audit and findbugs warning are incorrectly reported due to the 
compile of trunk hadoop-mapreduce-client-app was broken in the pre-patch phase. 
In fact, hs_err_pid3335.log is generated during this compile (see 
trunkFindBugsOutputhadoop-mapreduce-client-app.txt), and as it is broken, it 
wasn't able to detect the related findbugs warning (this warning belongs to 
NodeLabelExpression feature!).

+ The failed tests are not related to the patch. I verified that all these 
tests (after merged with the latest trunk) are passed in our local environment 
(Java 1.7.0_80 HotSpot(TM) 64-Bit Server VM (build 24.80-b11, mixed mode)/Linux 
xcp116 3.13.0-55-generic #94-Ubuntu SMP Thu Jun 18 00:27:10 UTC 2015 x86_64 
x86_64 x86_64 GNU/Linux)

> Support bandwidth enforcement for containers while reading from HDFS
> --------------------------------------------------------------------
>
>                 Key: YARN-2681
>                 URL: https://issues.apache.org/jira/browse/YARN-2681
>             Project: Hadoop YARN
>          Issue Type: New Feature
>          Components: nodemanager
>    Affects Versions: 2.5.1
>         Environment: Linux
>            Reporter: Nam H. Do
>              Labels: BB2015-05-TBR
>             Fix For: 2.7.0
>
>         Attachments: Traffic Control Design.png, YARN-2681.001.patch, 
> YARN-2681.002.patch, YARN-2681.patch
>
>
> To read/write data from HDFS on data node, applications establise TCP/IP 
> connections with the datanode. The HDFS read can be controled by setting 
> Linux Traffic Control  (TC) subsystem on the data node to make filters on 
> appropriate connections.
> The current cgroups net_cls concept can not be applied on the node where the 
> container is launched, netheir on data node since:
> -   TC hanldes outgoing bandwidth only, so it can be set on container node 
> (HDFS read = incoming data for the container)
> -   Since HDFS data node is handled by only one process,  it is not possible 
> to use net_cls to separate connections from different containers to the 
> datanode.
> Tasks:
> 1) Extend Resource model to define bandwidth enforcement rate
> 2) Monitor TCP/IP connection estabilised by container handling process and 
> its child processes
> 3) Set Linux Traffic Control rules on data node base on address:port pairs in 
> order to enforce bandwidth of outgoing data
> Concept: http://www.hit.bme.hu/~do/papers/EnforcementDesign.pdf
> Implementation: 
> http://www.hit.bme.hu/~dohoai/documents/HdfsTrafficControl.pdf
> http://www.hit.bme.hu/~dohoai/documents/HdfsTrafficControl_UML_diagram.png



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to