You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@ambari.apache.org by "Jaimin D Jetly (JIRA)" <ji...@apache.org> on 2013/12/11 02:50:07 UTC
[jira] [Updated] (AMBARI-4039) HDFS service check failed after HA
enabled on upgraded (1.3.2->1.3.3->2.0.6) cluster.
[ https://issues.apache.org/jira/browse/AMBARI-4039?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Jaimin D Jetly updated AMBARI-4039:
-----------------------------------
Description:
We start journalNode before Reconfigure HDFS in NameNode HA wizard in branch-1.4.2 (fixed in trunk). JournalNode reads the property _dfs.journalnode.edits.dir_ from hdfs-site.xml before HA related properties are configured in the wizard.
* In normal scenario, the value of _dfs.journalnode.edits.dir_ is /grid/0/hdfs/journal even before the HA related properties are added as it is present in HDP-2.0.6 stack definition.
* In the upgrade scenario from HDP-1.X stack to HDP-2.X, we do not add the property _dfs.journalnode.edits.dir_ in hdfs-site.xml. So JournalNode While starting doesn't find the property and puts all namenode edits data in the default path _/tmp/hadoop/dfs/journalnode_. After that HA wizard adds the property _dfs.journalnode.edits.dir_ = /grid/0/hdfs/journal in hdfs-site.xml.
* Next time when journalNode is restarted it doesn't find the namenode edits data as the path stated by property _dfs.journalnode.edits.dir_ is reconfigured.
was:
We start journalNode before Reconfigure HDFS in NameNode HA wizard in branch-1.4.2 (fixed in trunk). JournalNode reads the property _dfs.journalnode.edits.dir_ from hdfs-site.xml before HA related properties are configured in the wizard.
* In normal scenario, the value of _dfs.journalnode.edits.dir_ is /grid/0/hdfs/journal even before the HA related properties are added as it is present in HDP-2.0.6 stack definition.
* In the upgrade scenario from HDP-1.X stack to HDP-2.X, we do not add the property _dfs.journalnode.edits.dir_ in hdfs-site.xml. So JournalNode While starting doesn't find the property and puts all namenode edits data in the default path _/tmp/hadoop/dfs/journalnode_. After that HA wizard adds the property in hdfs-site.xml.
* Next time When journalNode is restarted it doesn't find the namenode edits data as the path stated by property _dfs.journalnode.edits.dir_ is reconfigured.
> HDFS service check failed after HA enabled on upgraded (1.3.2->1.3.3->2.0.6) cluster.
> -------------------------------------------------------------------------------------
>
> Key: AMBARI-4039
> URL: https://issues.apache.org/jira/browse/AMBARI-4039
> Project: Ambari
> Issue Type: Bug
> Affects Versions: 1.4.2
> Reporter: Jaimin D Jetly
> Assignee: Jaimin D Jetly
> Fix For: 1.4.2
>
>
> We start journalNode before Reconfigure HDFS in NameNode HA wizard in branch-1.4.2 (fixed in trunk). JournalNode reads the property _dfs.journalnode.edits.dir_ from hdfs-site.xml before HA related properties are configured in the wizard.
> * In normal scenario, the value of _dfs.journalnode.edits.dir_ is /grid/0/hdfs/journal even before the HA related properties are added as it is present in HDP-2.0.6 stack definition.
> * In the upgrade scenario from HDP-1.X stack to HDP-2.X, we do not add the property _dfs.journalnode.edits.dir_ in hdfs-site.xml. So JournalNode While starting doesn't find the property and puts all namenode edits data in the default path _/tmp/hadoop/dfs/journalnode_. After that HA wizard adds the property _dfs.journalnode.edits.dir_ = /grid/0/hdfs/journal in hdfs-site.xml.
> * Next time when journalNode is restarted it doesn't find the namenode edits data as the path stated by property _dfs.journalnode.edits.dir_ is reconfigured.
--
This message was sent by Atlassian JIRA
(v6.1.4#6159)