[
https://issues.apache.org/jira/browse/TEZ-2190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14356119#comment-14356119
]
Jeff Zhang commented on TEZ-2190:
---------------------------------
If check input-initializer not null enough ? I notice MRInput can create a
DataSourceDescriptor with both input initializer and shard num.
{code}
DataSourceDescriptor ds = DataSourceDescriptor.create(
InputDescriptor.create(inputClassName).setUserPayload(payload),
InputInitializerDescriptor.create(MRInputSplitDistributor.class.getName()),
inputSplitInfo.getNumTasks(), credentials,
VertexLocationHint.create(inputSplitInfo.getTaskLocationHints()),
null);
{code}
> TestOrderedWordCount fails when generateSplitsInClient set to true
> ------------------------------------------------------------------
>
> Key: TEZ-2190
> URL: https://issues.apache.org/jira/browse/TEZ-2190
> Project: Apache Tez
> Issue Type: Bug
> Reporter: Hitesh Shah
> Assignee: Hitesh Shah
> Attachments: TEZ-2190.1.patch, TEZ-2190.2.patch, TEZ-2190.3.patch
>
>
> numTasks for the map stage is unset in this scenario leading to the following
> error:
> {code}
> java.lang.IllegalStateException: initialmap has -1 tasks but does not have
> input initializers, 1-1 uninited sources or custom vertex manager to set it
> at runtime
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
