[
https://issues.apache.org/jira/browse/HIVE-20853?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16677561#comment-16677561
]
Jaume M commented on HIVE-20853:
--------------------------------
That's correct that this has only been witnessed at recovery. However I don't
see a reason why it couldn't happen when the AM has just launched:
* The nodes are added (no credentials are registered because we don't have a
dagId yet)
* We schedule some fetchers that try to setup a connection to a node that
hasn't had any task scheduled yet and the setupConnection fails
However in the first point we have the information to add the credentials in a
the nodes that are being added since we only need the app id.
> Expose ShuffleHandler.registerDag in the llap daemon API
> --------------------------------------------------------
>
> Key: HIVE-20853
> URL: https://issues.apache.org/jira/browse/HIVE-20853
> Project: Hive
> Issue Type: Improvement
> Components: llap
> Affects Versions: 3.1.0
> Reporter: Jaume M
> Assignee: Jaume M
> Priority: Critical
> Attachments: HIVE-20853.1.patch, HIVE-20853.2.patch,
> HIVE-20853.3.patch, HIVE-20853.4.patch
>
>
> Currently DAGs are only registered when a submitWork is called for that DAG.
> At this point the crendentials are added to the ShuffleHandler and it can
> start serving.
> However Tez might (and will) schedule tasks to fetch from the ShuffleHandler
> before anything of this happens and all this tasks will fail which may
> results in the query failing.
> This happens in the scenario in which a LlapDaemon just comes up and tez
> fetchers try to open a connection before a DAG has been registered.
> Adding this API will allow to register the DAG against the Daemon when the AM
> notices that a new Daemon is up.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
