hive-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Prasanth Jayachandran (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HIVE-16103) LLAP: Scheduler timeout monitor never stops with slot nodes
Date Fri, 03 Mar 2017 22:30:45 GMT

    [ https://issues.apache.org/jira/browse/HIVE-16103?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15895157#comment-15895157
] 

Prasanth Jayachandran commented on HIVE-16103:
----------------------------------------------

[~sseth] could you please take a look?

> LLAP: Scheduler timeout monitor never stops with slot nodes
> -----------------------------------------------------------
>
>                 Key: HIVE-16103
>                 URL: https://issues.apache.org/jira/browse/HIVE-16103
>             Project: Hive
>          Issue Type: Bug
>          Components: llap
>    Affects Versions: 2.2.0
>            Reporter: Prasanth Jayachandran
>            Assignee: Prasanth Jayachandran
>         Attachments: HIVE-16103.1.patch
>
>
> The scheduler timeout monitor is started when node count becomes 0 and stopped when node
count becomes 1. For node count, we were relying on the paths under llap namespace. With addition
of slot znodes, every node creates 2 paths (worker and slot). As a result, the size of the
instances cache will never be 1 (always multiple of 2) which leads to condition where timeout
monitor is not stopped. 



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

Mime
View raw message