[
https://issues.apache.org/jira/browse/TEZ-1928?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14272132#comment-14272132
]
Siddharth Seth commented on TEZ-1928:
-------------------------------------
Yes, you can do that.
This primarily happens because some vertices may start up early - before tasks
from dependent vertices are complete. Pre-emption is used to handle this when
there isn't enough capacity, but does not work in LocalMode.
This scheduler should become the default in a future release - once a few more
features are added to it. For now, it just guarantees ordering.
> Tez local mode hang in Pig tez local mode
> -----------------------------------------
>
> Key: TEZ-1928
> URL: https://issues.apache.org/jira/browse/TEZ-1928
> Project: Apache Tez
> Issue Type: Bug
> Reporter: Daniel Dai
> Attachments: TestMultiQuery.log, TestMultiQueryBasic.log,
> TestScalarAliasesLocal.log
>
>
> Pig tez local mode tests hang under some scenario. I attached several stack
> trace of hanging tests.
> By setting "tez.am.inline.task.execution.max-tasks", the test does not hang.
> However, we cannot make it general since Pig backend code is not designed to
> be multithread-safe.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
