ambari-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Keta Patel (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (AMBARI-17626) Blueprint registration step uses wrong format for property-attributes in Configuration
Date Fri, 08 Jul 2016 22:14:11 GMT

    [ https://issues.apache.org/jira/browse/AMBARI-17626?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15368586#comment-15368586
] 

Keta Patel commented on AMBARI-17626:
-------------------------------------

Updated the patch "AMBARI-17626-July08.patch" to include fix for the tests failing in the
last Hadoop QA run.

> Blueprint registration step uses wrong format for property-attributes in Configuration
> --------------------------------------------------------------------------------------
>
>                 Key: AMBARI-17626
>                 URL: https://issues.apache.org/jira/browse/AMBARI-17626
>             Project: Ambari
>          Issue Type: Bug
>          Components: blueprints
>    Affects Versions: trunk
>            Reporter: Keta Patel
>            Assignee: Keta Patel
>         Attachments: AMBARI-17626-July08.patch, AMBARI-17626.patch, cluster_with_defect_installed_from_blueprint.tiff,
cluster_with_fix_insatlled_from_blueprint.tiff, original_cluster_used_to_create_blueprint.tiff
>
>
> Blueprints make use of a population strategy in the registration step to create JSON
objects for properties and property-attributes. These properties and property-attributes get
stored in the database (DB) in the "clusterconfig" table under "config_data" and "config_attributes"
respectively. Format of these JSON objects is critical as the UI parses these objects assuming
them to be in a certain format.
> At present property-attributes like "final" get stored in "clusterconfig" table in the
format shown in attachment "original_cluster_used_to_create_blueprint.tiff".
> i.e. "final": {"attr1":"val1", "attr2":"val2"}
> When a new cluster is to be installed using a blueprint, after the blueprint is registered
and the new cluster is installed using the registered blueprint, the format of "property-attributes"
is as shown in attachment 
> "cluster_with_defect_installed_from_blueprint.tiff".
> i.e. "attr1":{"final":"val1"}, "attr2":{"final":"val2"}, "final":{"attr1":"val1", "attr2":"val2"}
> The population strategy is responsible for the "attr1":{"final":"val1"} and "attr2":{"final":"val2"}.
> The reason for "final":{"attr1":"val1", "attr2":"val2"} is because of a merge with the
Parent Configuration during blueprint registration. This Parent Configuration comes from Stack
using the XML files of the service properties. Because of this "final" attribute, the UI still
shows the properties correctly and hence it went undetected. 
> Proposed fix involves correcting the population strategy so that the attributes are placed
in the correct format. After the fix, the new cluster installed via blueprint shows "property-attributes"
as shown in attachment
> "cluster_with_fix_insatlled_from_blueprint.tiff"
> i.e. "final":{"attr1":"val1", "attr2":"val2"}



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

Mime
View raw message