atlas-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Suma Shivaprasad (JIRA)" <j...@apache.org>
Subject [jira] [Comment Edited] (ATLAS-597) Handle Cluster renaming
Date Mon, 12 Sep 2016 20:17:20 GMT

    [ https://issues.apache.org/jira/browse/ATLAS-597?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15438633#comment-15438633
] 

Suma Shivaprasad edited comment on ATLAS-597 at 9/12/16 8:16 PM:
-----------------------------------------------------------------

Changes required in DSL search :

clusterName is also being used as a separate searchable attribute for asset search. This also
needs to be replaced with "clusterId". 

All searches on clusterName will have to be replaced with cluster.name
for eg: 

clusterName currently is part of hive_db type which will now be replaced with "cluster.clusterName".


Search like 

hive_table where db.name='reporting' and db.clusterName='default' will now be replaced with
hive_table where db.name='reporting' and db.cluster.name='default'









was (Author: suma.shivaprasad):
Changes required in DSL search :

clusterName is also being used as a separate searchable attribute for asset search. This also
needs to be replaced with "clusterId". 
One question to consider is 
Should we continue to support search by clusterName . If this is required, then we will need
to treat cluster as a special entity, map clusterName to clusterId during gremlin translation.
This can be implemented generically as a translation rule for DSL to Gremlin translation



> Handle Cluster renaming
> -----------------------
>
>                 Key: ATLAS-597
>                 URL: https://issues.apache.org/jira/browse/ATLAS-597
>             Project: Atlas
>          Issue Type: Sub-task
>    Affects Versions: 0.7-incubating
>            Reporter: Suma Shivaprasad
>
> Cluster name could easily be changed by users which is used in qualified names to dedup
entities. For eg: hive_table has a qualifiedName <dbName.tableName@clusterName> . Hence
all entities will be recreated if they are renamed unless a migration on repository is done
with new cluster name.
> Hence cluster  should be represented by an id which is constant and tied to clusterName
which could change



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

Mime
View raw message