[
https://issues.apache.org/jira/browse/AMBARI-24981?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
ASF GitHub Bot updated AMBARI-24981:
------------------------------------
Labels: pull-request-available (was: )
> JournalNode may fail to start due to unreadable config file
> -----------------------------------------------------------
>
> Key: AMBARI-24981
> URL: https://issues.apache.org/jira/browse/AMBARI-24981
> Project: Ambari
> Issue Type: Bug
> Components: ambari-agent
> Reporter: Doroszlai, Attila
> Assignee: Doroszlai, Attila
> Priority: Major
> Labels: pull-request-available
> Fix For: 2.8.0
>
>
> During startup, JournalNode reads config files that belong to YARN, not HDFS
> (eg. {{yarn-site.xml}}), if they are present in the Hadoop config dir.
> Ambari's {{File()}} provider creates files at the final location, then sets
> metadata (ownership and permissions). This makes the file visible to other
> processes with possibly the wrong metadata.
> If agent-side parallel execution is enabled, JournalNode may be started
> concurrently with some YARN component. JournalNode may encounter a config
> file that it cannot read, because the file's permissions and owner are still
> 440 and root, respectively, which causes it to shutdown:
> {noformat}
> ERROR conf.Configuration (Configuration.java:loadResource(2999)) - error
> parsing conf yarn-site.xml
> java.io.FileNotFoundException: /etc/hadoop/conf/yarn-site.xml (Permission
> denied)
> {noformat}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
