[
https://issues.apache.org/jira/browse/ATLAS-4076?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17248148#comment-17248148
]
Deep Singh commented on ATLAS-4076:
-----------------------------------
Analysis:
=========
After detecting relationship and creating relationship edge, the propagated
enteties(classifications) are notified to entityChangeListner through
entityChangeNotifier. However details of the propagated enteties are not passed
directly to notifier, but through request context (buffered into
addedPropagation list).
After processing every edge, AtlasRelationshipStore manager sends notification
to entityChangeListner, which simply gets all the items in request context
buffer list.
In this issue, Hive sends and event which has multiple relationships, and only
one relationship has propagated entities, but due to multiple
notifications(which is correct) same buffer list is processed multipletimes
(which is wrong).
> Creating an asset from existing asset adds duplicate Atlas audit entries
> -------------------------------------------------------------------------
>
> Key: ATLAS-4076
> URL: https://issues.apache.org/jira/browse/ATLAS-4076
> Project: Atlas
> Issue Type: Bug
> Components: atlas-core
> Affects Versions: 2.1.0
> Reporter: Deep Singh
> Assignee: Deep Singh
> Priority: Critical
>
> Steps :
> Go to beeline in hive.
> Create a new asset which is created from an existing asset, using the query :
> create table test_web_site as select * from web_site_five;
> Observe that web_site_five contains two managed tags which should be
> propagated to the newly created table , i.e : test_web_site
> Observe the audit tab for 'test_web_site' table in atlas contains multiple
> entries
--
This message was sent by Atlassian Jira
(v8.3.4#803005)