[
https://issues.apache.org/jira/browse/TEZ-1734?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14199406#comment-14199406
]
Bikas Saha commented on TEZ-1734:
---------------------------------
Not sure. Here is what I saw.
testRecovery_VertexManagerErrorOnRecovery would fail if > 0 tasks were
specified in
{code}vertex3.restoreFromEvent(new VertexInitializedEvent(vertex3
.getVertexId(), "vertex3", initRequestedTime, initedTime, 0, "",
null));{code} It would fail with some unsupported transition
message.
In testRecovery_RecoveringFromNew would hit the numTasks assert in
RecoverTransition when I would send
{code}vertex3.handle(new VertexEventManagerUserCodeError(vertex3.getVertexId(),
new AMUserCodeException(Source.VertexManager, new
TezUncheckedException("test"))));{code}
before this
{code} VertexImpl vertex2 = (VertexImpl) dag.getVertex("vertex2");
vertex2.handle(new VertexEventRecoverVertex(vertex2.getVertexId(),
VertexState.RUNNING));{code}
I was expecting the VertexEventManagerUserCodeError to change the
recoveredState to FAILED and then vertex2 recovering would change vertex3 state
to FAILED.
Maybe we could add these testcases to this patch and verify that they work now
(they were not working when TEZ-1547 got committed).
> Vertex's taskNum may be -1 when recovered from NEW to FAILED/KILLED
> -------------------------------------------------------------------
>
> Key: TEZ-1734
> URL: https://issues.apache.org/jira/browse/TEZ-1734
> Project: Apache Tez
> Issue Type: Bug
> Affects Versions: 0.5.1
> Reporter: Jeff Zhang
> Assignee: Jeff Zhang
> Attachments: TEZ-1734.patch
>
>
> When vertex recovered from NEW to FAILED/KILLED, the taskNum may be -1, in
> this case, we don't need to recover its tasks
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
