You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ambari.apache.org by "Attila Magyar (JIRA)" <ji...@apache.org> on 2018/12/04 13:27:00 UTC

[jira] [Commented] (AMBARI-24893) Add relevant NameNode start logic to OneFS mpack

    [ https://issues.apache.org/jira/browse/AMBARI-24893?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16708698#comment-16708698 ] 

Attila Magyar commented on AMBARI-24893:
----------------------------------------

hi [~dmtucker], [~kbhusanurmath],

The problem is that in the mpack there is only one client component which is installed on multiple nodes. So if we put an initialization like this into the onefs_client.py, it will be executed multiple times, so on a large cluster it'll generate unnecessary overhead.  But I don't see any other options. Let me think about this.

> Add relevant NameNode start logic to OneFS mpack
> ------------------------------------------------
>
>                 Key: AMBARI-24893
>                 URL: https://issues.apache.org/jira/browse/AMBARI-24893
>             Project: Ambari
>          Issue Type: Bug
>    Affects Versions: 2.7.0
>         Environment: HDP 3.0.1 + OneFS 8.1.2
>            Reporter: David Tucker
>            Priority: Major
>
> Deploying with OneFS using the mpack seems to omit setup steps. I'm not sure exactly where the steps are defined in 2.7, but in 2.6 they seem to be here:
> https://github.com/apache/ambari/blob/ac651d9d88d0c65d7ec1141ac5cd2af659781d79/ambari-server/src/main/resources/common-services/HDFS/2.1.0.2.0/package/scripts/hdfs_namenode.py#L102-L238
> It's possible not all the steps are relevant, but at the very least these seem to be:
> https://github.com/apache/ambari/blob/ac651d9d88d0c65d7ec1141ac5cd2af659781d79/ambari-server/src/main/resources/common-services/HDFS/2.1.0.2.0/package/scripts/hdfs_namenode.py#L234-L235
> According to [Kiran|https://issues.apache.org/jira/secure/ViewProfile.jspa?name=kbhusanurmath], "this is causing issue starting some of the services and importantly timeline server V2 is unable to create necessary znode and start ats-hbase app."
> It also appears to cause the YARN service check to fail with an ACE (since {{/user/ambari-qa}} was not set up correctly).



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)