[ 
https://issues.apache.org/jira/browse/HIVE-15242?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Sergey Shelukhin updated HIVE-15242:
------------------------------------
    Attachment: HIVE-15242.04.patch

Again...

> LLAP: Act on Node update notifications from registry, fix isAlive checks
> ------------------------------------------------------------------------
>
>                 Key: HIVE-15242
>                 URL: https://issues.apache.org/jira/browse/HIVE-15242
>             Project: Hive
>          Issue Type: Task
>            Reporter: Siddharth Seth
>            Assignee: Sergey Shelukhin
>         Attachments: HIVE-15242.01.patch, HIVE-15242.02.patch, 
> HIVE-15242.03.patch, HIVE-15242.04.patch
>
>
> isAlive checks are currently completely broken, since the ZK registry does 
> not update existing ServiceIsntances. Instead it creates new instances each 
> time.
> This causes non-existant nodes to be used for scheduling in case of node 
> failures.
> Also, act on the notifications sent by the registry about nodes going down, 
> or new nodes being added. (new nodes already handled)



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to