[
https://issues.apache.org/jira/browse/HADOOP-1971?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Doug Cutting updated HADOOP-1971:
---------------------------------
Attachment: HADOOP-1971.patch
I think a warning is enough. There are cases where this may not be an error
(e.g., if someone has installed their mapred classes on the cluster before
startup) so the warning may sometimes be spurious, but those cases should be
rare.
I changed the warning message to specifically complain about the root cause (no
job jar specified) and point to potential solutions.
> Constructing a JobConf without a class leads to a very misleading error
> message.
> --------------------------------------------------------------------------------
>
> Key: HADOOP-1971
> URL: https://issues.apache.org/jira/browse/HADOOP-1971
> Project: Hadoop
> Issue Type: Improvement
> Components: mapred
> Affects Versions: 0.14.2
> Environment: Centos, java 1.5
> Reporter: Ted Dunning
> Assignee: Enis Soztutar
> Priority: Minor
> Fix For: 0.15.0
>
> Attachments: HADOOP-1971.patch, notSetJarError_v1.patch
>
>
> If this line from a typical map/reduce program
> JobConf conf = new JobConf(Flatten.class);
> has the argument deleted then I get a message that the input format that I
> specify cannot be found. That message leads a naive user on a rat chase. I
> only corrected this by bisecting against a working program.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
