[
https://issues.apache.org/jira/browse/HIVE-21034?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16721689#comment-16721689
]
Alan Gates commented on HIVE-21034:
-----------------------------------
Sorry, but I still don't understand the use case. For an ephemeral cluster I
can see two things that make sense:
# The metastore instance is long running, but the cluster and the data it
generated are ephemeral. In that case it makes sense to have a command that
drops all data generated by a cluster. Since we don't track that the easier
solution would be to have the ephemeral cluster use a particular database in
Hive and then do a drop database cascade as part of the cleanup. This leaves
other metadata in the long running metastore.
# The metastore instance is ephemeral, so we want to clean it up. An option
to remove the entire metastore RDBMS schema like -cleanMetastore makes sense
for this. This one is dangerous and would need some kind of "are you really
super sure" check.
But --dropAllBases doesn't match either of these. So I'm still confused. It
seems to me it brings lots of danger and no benefit.
> Add option to schematool to drop Hive databases
> -----------------------------------------------
>
> Key: HIVE-21034
> URL: https://issues.apache.org/jira/browse/HIVE-21034
> Project: Hive
> Issue Type: Improvement
> Reporter: Daniel Voros
> Assignee: Daniel Voros
> Priority: Major
>
> An option to remove all Hive managed data could be a useful addition to
> {{schematool}}.
> I propose to introduce a new flag {{-dropAllDatabases}} that would *drop all
> databases with CASCADE* to remove all data of managed tables.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
