Mailing-List: contact dev-help@ambari.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@ambari.apache.org
Date: Sat, 26 Sep 2015 21:05:04 +0000 (UTC)
From: "Alejandro Fernandez (JIRA)" <jira@apache.org>
To: dev@ambari.apache.org
Message-ID: <JIRA.12896621.1443301054000.83350.1443301504097@Atlassian.JIRA>
In-Reply-To: <JIRA.12896621.1443301054000@Atlassian.JIRA>
References: <JIRA.12896621.1443301054000@Atlassian.JIRA>
 <JIRA.12896621.1443301054926@arcas>
Subject: [jira] [Updated] (AMBARI-13251) RU - HDFS_Client restart and
 hdp-select causes dfs_data_dir_mount.hist to be lost
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


     [ https://issues.apache.org/jira/browse/AMBARI-13251?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Alejandro Fernandez updated AMBARI-13251:
-----------------------------------------
    Attachment: AMBARI-13251.patch

> RU - HDFS_Client restart and hdp-select causes dfs_data_dir_mount.hist to be lost
> ---------------------------------------------------------------------------------
>
>                 Key: AMBARI-13251
>                 URL: https://issues.apache.org/jira/browse/AMBARI-13251
>             Project: Ambari
>          Issue Type: Bug
>          Components: ambari-server
>    Affects Versions: 2.1.2
>            Reporter: Alejandro Fernandez
>            Assignee: Alejandro Fernandez
>            Priority: Blocker
>             Fix For: 2.1.2
>
>         Attachments: AMBARI-13251.patch
>
>
> During the Clients group of RU, restarting hdfs_client changes changes the symlink /usr/hdp/current/hadoop-client/ from the old version to the new version.
> Because the Data Dir Mount History file is stored at /etc/hadoop/conf/dfs_data_dir_mount.hist and
> /etc/hadoop/conf symlinks to /usr/hdp/current/hadoop-client/conf 
> it means that the file must be copied to the new location.
> Multiple clients (HDFS, HBase, Hive, YARN, MapReduce, Pig, Slider, Tez) actually end up calling
> {code}
> hdp_select.select("hadoop-client", params.version)
> {code}
> that changes the symlink, so the first one to be ran during RU must copy the file. If DataNode is present, it means that HDFS Client is guaranteed to be there, and from all of those clients it is the first one in the Upgrade Pack.
> During a Rolling Downgrade, the file should already exist in the previous version, and the file is only written to when DataNode is restarted, so we are ok.


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