[
https://issues.apache.org/jira/browse/LIVY-1059?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Nilesh Rathi updated LIVY-1059:
-------------------------------
Description:
Currently, Apache Livy relies entirely on the client request payload to
determine which YARN queue a session should be submitted to. If a client
submits a session creation request without explicitly defining the {{"queue"}}
parameter, Livy passes a null or empty configuration to Spark, causing the
application to fall back to the Hadoop cluster's global default YARN queue
(usually {{{}default{}}}).
This creates a limitation for multi-tenant cluster administrators.
Administrators often need to enforce a specific default YARN queue exclusively
for all Livy-generated workloads (e.g., a queue named {{{}livy{}}}), without
relying on end-users to remember to pass the parameter manually in every REST
API request payload.
> Add livy.spark.yarn.queue configuration to set default YARN queue
> -----------------------------------------------------------------
>
> Key: LIVY-1059
> URL: https://issues.apache.org/jira/browse/LIVY-1059
> Project: Livy
> Issue Type: Improvement
> Reporter: Nilesh Rathi
> Assignee: Nilesh Rathi
> Priority: Major
>
> Currently, Apache Livy relies entirely on the client request payload to
> determine which YARN queue a session should be submitted to. If a client
> submits a session creation request without explicitly defining the
> {{"queue"}} parameter, Livy passes a null or empty configuration to Spark,
> causing the application to fall back to the Hadoop cluster's global default
> YARN queue (usually {{{}default{}}}).
> This creates a limitation for multi-tenant cluster administrators.
> Administrators often need to enforce a specific default YARN queue
> exclusively for all Livy-generated workloads (e.g., a queue named
> {{{}livy{}}}), without relying on end-users to remember to pass the parameter
> manually in every REST API request payload.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
