You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@crail.apache.org by "Adrian Schuepbach (Jira)" <ji...@apache.org> on 2020/09/01 12:19:00 UTC

[jira] [Assigned] (CRAIL-111) Namenode mechanism to transition datanode state from running to leaving

     [ https://issues.apache.org/jira/browse/CRAIL-111?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Adrian Schuepbach reassigned CRAIL-111:
---------------------------------------

    Assignee: Adrian Schuepbach

> Namenode mechanism to transition datanode state from running to leaving
> -----------------------------------------------------------------------
>
>                 Key: CRAIL-111
>                 URL: https://issues.apache.org/jira/browse/CRAIL-111
>             Project: Apache Crail
>          Issue Type: New Feature
>            Reporter: Adrian Schuepbach
>            Assignee: Adrian Schuepbach
>            Priority: Major
>
> The namenode manages membership of datanodes in a running Crail instance. As such, it is also responsible for datanode joining and leaving.
>  
> Currently, Crail supports that datanodes join, wihle already running. However no mechanism is implemented to allow datanodes leaving, while the rest of the Crail instance remains running.
>  
> We need a mechanism to allow datanodes to leave the Crail cluster at runtime.
> A simple first implementation could consist of a mechanism, which marks datanodes that should leave and only removes them, when none of its blocks are allocated anymore. This does not involve data movement.
> We can borrow the idea from our Pocket implementation.
>  



--
This message was sent by Atlassian Jira
(v8.3.4#803005)