[jira] [Created] (AMBARI-23434) Fix stack issues in HDFS to support Namenode Federation setup

["Andrew Onischuk (JIRA)" <ji...@apache.org>]

Andrew Onischuk created AMBARI-23434:
----------------------------------------

             Summary: Fix stack issues in HDFS to support Namenode Federation setup
                 Key: AMBARI-23434
                 URL: https://issues.apache.org/jira/browse/AMBARI-23434
             Project: Ambari
          Issue Type: Bug
            Reporter: Andrew Onischuk
            Assignee: Andrew Onischuk
             Fix For: 2.7.0
         Attachments: AMBARI-23434.patch

For example, here are 2 things I found to be probably wrong.

1\. Journal node restart failed because we cannot find hdfs-site :
dfs.journalnode.edits.dir. We delete that property in the wizard. We may have
to change that to dfs.journalnode.edits.dir.<nameservice>

2\. The following snippet in params_linux.py on HDFS 3.0 stack seems wrong. It
has been designed to work with only 1 nameservice.

    
    
    
    dfs_ha_enabled = False
    dfs_ha_nameservices = default('/configurations/hdfs-site/dfs.internal.nameservices', None)
    if dfs_ha_nameservices is None:
      dfs_ha_nameservices = default('/configurations/hdfs-site/dfs.nameservices', None)
    dfs_ha_namenode_ids = default(format("/configurations/hdfs-site/dfs.ha.namenodes.{dfs_ha_nameservices}"), None)
    

3\. After setting up NN Fed, when I restart namenodes, I see the following
error.

    
    
    
        main_resource.resource.security_enabled, main_resource.resource.logoutput)
      File "/usr/lib/ambari-agent/lib/resource_management/libraries/providers/hdfs_resource.py", line 154, in __init__
        security_enabled, run_user)
      File "/usr/lib/ambari-agent/lib/resource_management/libraries/functions/namenode_ha_utils.py", line 204, in get_property_for_active_namenode
        if INADDR_ANY in value and rpc_key in hdfs_site:
      File "/usr/lib/ambari-agent/lib/resource_management/libraries/script/config_dictionary.py", line 73, in __getattr__
        raise Fail("Configuration parameter '" + self.name + "' was not found in configurations dictionary!")
    resource_management.core.exceptions.Fail: Configuration parameter 'dfs.namenode.https-address.ns2.nn1' was not found in configurations dictionary!
    

This is probably because the namenode_ha_utils is not equipped to handle
multiple nameservices.

We may have to create to fix such stack errors when the wizard is done.





--
This message was sent by Atlassian JIRA
(v7.6.3#76005)