[ 
https://issues.apache.org/jira/browse/AMBARI-10067?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14361358#comment-14361358
 ] 

Hudson commented on AMBARI-10067:
---------------------------------

SUCCESS: Integrated in Ambari-trunk-Commit #2031 (See 
[https://builds.apache.org/job/Ambari-trunk-Commit/2031/])
AMBARI-10067: ZKFailoverController failed on restart (jluniya) (jluniya: 
http://git-wip-us.apache.org/repos/asf?p=ambari.git&a=commit&h=a5c571e973badd90da3cb34e6c8345a3d80c9e4c)
* 
ambari-server/src/main/resources/common-services/HDFS/2.1.0.2.0/package/scripts/hdfs_namenode.py
* ambari-server/src/test/python/stacks/2.0.6/HDFS/test_zkfc.py
* 
ambari-server/src/main/resources/common-services/HDFS/2.1.0.2.0/package/scripts/zkfc_slave.py


> ZKFailoverController failed on restart
> --------------------------------------
>
>                 Key: AMBARI-10067
>                 URL: https://issues.apache.org/jira/browse/AMBARI-10067
>             Project: Ambari
>          Issue Type: Bug
>          Components: ambari-agent
>    Affects Versions: 2.0.0
>            Reporter: Jayush Luniya
>            Assignee: Jayush Luniya
>             Fix For: 2.0.0
>
>         Attachments: AMBARI-10067.patch
>
>
> Issue Summary:
> 1. When Namenode HA is enabled and "automatic failover" is selected, the 
> "hdfs zkfc -formatZK" command is used to create a znode in ZooKeeper inside 
> of which the automatic failover system stores its data.
> 2. When Namenode HA is enabled through the wizard, the user is asked to 
> manually run this command. 
> 3. When Namenode HA cluster is deployed using HA blueprint, the agent script 
> has been executing "hdfs zkfc -formatZK -force -nonInteractive" to 
> auto-format the znode. This means that everytime a ZKFC starts, we end up 
> formatting the znode used to maintain the failover state.
> Fix:
> 1. Changing the command to "hdfs zkfc -formatZK -nonInteractive" so that the 
> failover state is not reformatted and handling the case when the command 
> returns a status=2 to indicate that the znode already exists and hence 
> skipping formatting it.



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

Reply via email to