[
https://issues.apache.org/jira/browse/AMBARI-15790?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15253787#comment-15253787
]
Alex Bush commented on AMBARI-15790:
------------------------------------
Hi Dymtro,
I'm not sure the haadmin case is covered. I believe all the hdfs haadmin
commands now need a -ns flag:
https://github.com/apache/ambari/blob/5772ceb05a1804b1c16cccc7968cd17a25de8ccd/ambari-server/src/main/resources/common-services/HDFS/2.1.0.2.0/package/scripts/utils.py#L101
failover_command = format("hdfs haadmin -failover {namenode_id}
{other_namenode_id}")
should be
failover_command = format("hdfs haadmin -ns {!nameservice!} -failover
{namenode_id} {other_namenode_id}")
As per the discussion linked above. I could be wrong but it looks like this
tool alone doesn't respect dfs.internal.nameservices.
Otherwise you get:
/usr/hdp/current/hadoop-hdfs-namenode/bin/hdfs --config
/usr/hdp/current/hadoop-client/conf haadmin -getServiceState nn1 | grep
active''] {}
2016-04-18 09:51:39,868 - call returned (1, 'Illegal argument: Unable to
determine the name service ID. This is an HA configuration with multiple name
services configured. dfs.nameservices is set to [CLUSTER2, CLUSTER1]. Please
re-run with the -ns option.')
What do you think? Should another bug be opened to cover this?
> Clean up stack scripts that refer to dfs.nameservices to use
> dfs.internal.nameservices as first option
> ------------------------------------------------------------------------------------------------------
>
> Key: AMBARI-15790
> URL: https://issues.apache.org/jira/browse/AMBARI-15790
> Project: Ambari
> Issue Type: Bug
> Components: stacks
> Affects Versions: 2.2.2
> Reporter: Sumit Mohanty
> Assignee: Dmytro Sen
> Fix For: 2.4.0
>
> Attachments: AMBARI-15790-trunk_5.patch
>
>
> Several stack scripts refer to dfs.nameservices and they should be modified
> to use dfs.internal.nameservices
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
