[
https://issues.apache.org/jira/browse/MAPREDUCE-6384?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14575010#comment-14575010
]
zhihai xu commented on MAPREDUCE-6384:
--------------------------------------
Hi [~lichangleo], thanks for working on this issue. this looks like a good
improvement to show more diagnostic information about fetch failure.
In your patch, I think you want to get the host name for the reduce task.
But it looks like you use the host name for Application Master as
reduceHostname in the following code.
{code}
hostname = InetAddress.getLocalHost().getHostName();
{code}
I think the local host is the AM host not reduce task host, it will be correct
only when both AM and reduce tasks are running on the same host.
I think we should use {{this.container.getNodeId().getHost()}} as
reduceHostname, which is the same as containerHostName.
Also it looks like the following line is more than 80 characters .
{code}
Arrays.asList(new TaskAttemptId[] {mapAttempt.getID()}),
"sudoHost"));
{code}
> add the last reporting reducer host info and attempt id on the map error
> message due to too many fetch failure
> --------------------------------------------------------------------------------------------------------------
>
> Key: MAPREDUCE-6384
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-6384
> Project: Hadoop Map/Reduce
> Issue Type: Bug
> Reporter: Chang Li
> Assignee: Chang Li
> Attachments: MAPREDUCE-6384.2.patch, MAPREDUCE-6384.3.patch,
> MAPREDUCE-6384.4.2.patch, MAPREDUCE-6384.4.patch
>
>
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
