You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@ambari.apache.org by "Dmitry Lysnichenko (JIRA)" <ji...@apache.org> on 2015/02/03 19:51:34 UTC

[jira] [Created] (AMBARI-9461) Issue with namenode format

Dmitry Lysnichenko created AMBARI-9461:
------------------------------------------

             Summary: Issue with namenode format
                 Key: AMBARI-9461
                 URL: https://issues.apache.org/jira/browse/AMBARI-9461
             Project: Ambari
          Issue Type: Bug
          Components: ambari-server
    Affects Versions: 2.0.0
            Reporter: Dmitry Lysnichenko
            Assignee: Dmitry Lysnichenko
             Fix For: 2.0.0


Appears to be a bug:
checkForFormat.sh looks for a "${mark_dir}" to determine if the namenode dir is formatted.
But "${mark_dir}" seems to be hardcoded to /var/lib/hdfs/namenode/formatted
when it should instead use {dfs.namenode.data.dir}/formatted/.
That results in:
…/formatted/ never being created (it tries but the parent dir /var/lib/hdfs/namenode doesn’t exist so it can’t)
…/formatted/ not being checked on cluster build
I’ve temporarily hacked around the issue by applying a symbolic link before Ambari automation starts:
mkdir /hadoop/hdfs; ln -s /hadoop/hdfs /var/lib/
(where /hadoop/hdfs is the base of all hdfs dirs. Result in /hadoop/hdfs/namenode/formatted being available at /var/lib/hdfs/namenode/formatted)




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