[
https://issues.apache.org/jira/browse/ATLAS-271?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14984907#comment-14984907
]
Shwetha G S commented on ATLAS-271:
-----------------------------------
The way to address it is by using namespaces for types. It can't be fixed
without namespaces
> Hive data model import fails if there exists any type of name "hive_process"
> ----------------------------------------------------------------------------
>
> Key: ATLAS-271
> URL: https://issues.apache.org/jira/browse/ATLAS-271
> Project: Atlas
> Issue Type: Bug
> Affects Versions: 0.6-incubating
> Reporter: Ayub Khan
>
> Hive data model import fails if there exists a type of name "hive_process"
> 1. Create a trait with typename as hive_process.
> For example:
> {noformat}
> curl 'http://localhost:21000/api/atlas/types' -H 'Origin:
> http://localhost:21000' -H 'Content-Type: application/json;charset=UTF-8'
> --data-binary
> '{"enumTypes":[],"traitTypes":[{"superTypes":[],"typeName":"hive_process","attributeDefinitions":[],"hierarchicalMetaTypeName":"org.apache.atlas.typesystem.types.TraitType"}],"structTypes":[],"classTypes":[]}'
> --compressed
> {noformat}
> 2. Now run import_hive.sh
> Import hive data model fails because atlas thinks that hive data model is
> already registered.
> This issue is quite possible when the user/admin(mostly unaware of hive data
> model and its types) is creating tags while bringing atlas.
> hive data model is just an example, this issue can happen with any data model.
> One solution might be to restrict creating types with reserved keywords.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
