[ 
https://issues.apache.org/jira/browse/HADOOP-15662?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16739942#comment-16739942
 ] 

Thomas Marquardt commented on HADOOP-15662:
-------------------------------------------

Some comments on the 002 patch:

AbfsRestOperation.java:

  L173: You have {{LOG.warn(String.format("{color:#FF0000}Can not{color} reach 
host: %s, retrying to {color:#FF0000}solve{color} the host name...", 
ex.getMessage()));}}

  It is important to include the host name in this message, so that the user 
can potentially identify the typo in the host name. Also, note that the word 
{color:#FF0000}cannot{color} is written as a single word, and we are retrying 
to {color:#FF0000}resolve{color} the host name, not solve the host name.  I 
recommend replacing L173 with:

   {{LOG.warn(String.format("Unknown host name '%s'.  Retrying to resolve the 
host name...", httpOperation.getUrl().getHost()));}}

 L174: We should not use else if here.  Instead it should be if.  I mean, we 
want to log the warning when there is a host name exception, but we should not 
change the existing debug logging.  The existing debug logging will trace the 
exception.  They are separate things.

> ABFS: Better exception handling of DNS errors
> ---------------------------------------------
>
>                 Key: HADOOP-15662
>                 URL: https://issues.apache.org/jira/browse/HADOOP-15662
>             Project: Hadoop Common
>          Issue Type: Sub-task
>          Components: fs/azure
>            Reporter: Thomas Marquardt
>            Assignee: Da Zhou
>            Priority: Major
>             Fix For: 3.2.1
>
>         Attachments: HADOOP-15662-001.patch, HADOOP-15662-002.patch
>
>
> DNS errors are common during testing due to typos or misconfiguration.  They 
> can also occur in production, as some transient DNS issues occur from time to 
> time. 
> 1) Let's investigate if we can distinguish between the two and fail fast for 
> the test issues, but continue to have retry logic for the transient DNS 
> issues in production.
> 2) Let's improve the error handling of DNS failures, so the user has an 
> actionable error message.



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

---------------------------------------------------------------------
To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: common-issues-h...@hadoop.apache.org

Reply via email to