[ 
https://issues.apache.org/jira/browse/HDFS-13959?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16663239#comment-16663239
 ] 

Surendra Singh Lilhore edited comment on HDFS-13959 at 10/25/18 4:57 AM:
-------------------------------------------------------------------------

Thanks [~ayushtkn] for patch. 

minor comments 

>>This comment is not more reuired.
{code:java}
/**
 * Use host names that can be resolved (
 * InetSocketAddress#isUnresolved == false). Otherwise,
 * CombinedHostFileManager won't allow those hosts.
 */
 static final String[] hosts =
 {"host1", "host2", "host3", "host4",
 "host5", "host6"};{code}
>>Pls add here comment why the IP address used instead of hostname.  May be 
>>same comment you can use here.
{code:java}
+      datanodes[i].setHostName(datanodeID.getIpAddr());{code}


was (Author: surendrasingh):
Thanks [~ayushtkn] for patch. 

minor comments 
 # This comment is not more reuired.
{code:java}
/**
 * Use host names that can be resolved (
 * InetSocketAddress#isUnresolved == false). Otherwise,
 * CombinedHostFileManager won't allow those hosts.
 */
 static final String[] hosts =
 {"host1", "host2", "host3", "host4",
 "host5", "host6"};{code}

 # Pls add here comment why the IP address used instead of hostname.  May be 
same comment you can use here.

{code:java}
+      datanodes[i].setHostName(datanodeID.getIpAddr());{code}

> TestUpgradeDomainBlockPlacementPolicy is flaky
> ----------------------------------------------
>
>                 Key: HDFS-13959
>                 URL: https://issues.apache.org/jira/browse/HDFS-13959
>             Project: Hadoop HDFS
>          Issue Type: Bug
>            Reporter: Ayush Saxena
>            Assignee: Ayush Saxena
>            Priority: Major
>         Attachments: HDFS-13959-01.patch
>
>
> The procedure followed for rack mapping is ambiguous.
> Due to same host name the mapping of nodes to rack was not as per our 
> requirement.
> In slower systems all nodes gets mapped to the latter rack2 and in little 
> fast system one node gets mapped to rack1. Thus leading to test failure since 
> now the rack fault tolerance comes into existence.Which can not be satisfied 
> by this ambiguous mapping. 



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to