[
https://issues.apache.org/jira/browse/TEZ-2292?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14503506#comment-14503506
]
Bikas Saha commented on TEZ-2292:
---------------------------------
After looking at this, it seems this is not needed. The code calls VM methods
on a threadpool and that already takes care of catching all exceptions and
reporting them properly. In fact, given that this exists, IMO, we should remove
the TezException from the reconfigureVertex() method so that users dont have to
explicitly catch these exceptions. Further, if users catch and swallow these
exceptions then it may lead to the state machines being left behind in an
invalid state and weird errors down the line. This is probably why originally
the exceptions were all unchecked.
[~hitesh] Given the above, I am proposing to remove TezException from the
method signature and adding an e2e test case that validates this. There are
already some tests that cover these scenarios which broke when I tried to "fix"
this by catching and throwing exceptions.
> Add API that allows Vertex Manager to fail a vertex
> ---------------------------------------------------
>
> Key: TEZ-2292
> URL: https://issues.apache.org/jira/browse/TEZ-2292
> Project: Apache Tez
> Issue Type: Task
> Reporter: Bikas Saha
> Assignee: Bikas Saha
>
> If the Vertex Manager has an error or cannot apply a required reconfiguration
> then it should be allowed to fail the vertex.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
