[
https://issues.apache.org/jira/browse/HIVE-12368?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Eric Lin updated HIVE-12368:
Description:
If we want to setup one cluster to support multiple environments, namely DEV,
QA, PRE-PROD etc, this is done in the way that data from different environments
can be generated into different locations in the HDFS and Hive Databases.
This works fine, however, when need to deploy UDF classes for different
environments, it becomes tricky, as each class has the same namespace, even
though we have created udf-dev.jar, udf-qa.jar etc.
Creating each HS2 per environment is another option, however, with LB setup, it
becomes harder.
The request is to have HS2 support loading UDFs in such environment, the
implementation is open to discussion.
I know that this setup is no ideal, as the better approach is to have one
cluster per environment, however, in the case that you have limited number of
nodes in the setup, this might be the only option and I believe many people can
benefit from it.
Thanks
was:
If we want to setup one cluster to support multiple environments, namely DEV,
QA, PRE-PROD etc, this is done in the way that data from different environment
will be generated into different locations in the HDFS and Hive Databases.
This works fine, however, when need to deploy UDF classes for different
environments, it becomes tricky, as each class has the same namespace, even
though we have created udf-dev.jar, udf-qa.jar etc.
Creating each HS2 per environment is another option, however, with LB setup, it
becomes harder.
The request is to have HS2 support loading UDFs in such environment, the
implementation is open to discussion.
I know that this setup is no ideal, as the better approach is to have one
cluster per environment, however, in the case that you have limited number of
nodes in the setup, this might be the only option and I believe many people can
benefit from it.
Thanks
> Provide support for different versions of same JAR files for loading UDFs
> -
>
> Key: HIVE-12368
> URL: https://issues.apache.org/jira/browse/HIVE-12368
> Project: Hive
> Issue Type: New Feature
> Components: HiveServer2
>Reporter: Eric Lin
>Assignee: Vaibhav Gumashta
>Priority: Minor
>
> If we want to setup one cluster to support multiple environments, namely DEV,
> QA, PRE-PROD etc, this is done in the way that data from different
> environments can be generated into different locations in the HDFS and Hive
> Databases.
> This works fine, however, when need to deploy UDF classes for different
> environments, it becomes tricky, as each class has the same namespace, even
> though we have created udf-dev.jar, udf-qa.jar etc.
> Creating each HS2 per environment is another option, however, with LB setup,
> it becomes harder.
> The request is to have HS2 support loading UDFs in such environment, the
> implementation is open to discussion.
> I know that this setup is no ideal, as the better approach is to have one
> cluster per environment, however, in the case that you have limited number of
> nodes in the setup, this might be the only option and I believe many people
> can benefit from it.
> Thanks
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
