[ https://issues.apache.org/jira/browse/SPARK-24491?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Marcelo Vanzin resolved SPARK-24491. ------------------------------------ Resolution: Duplicate Consensus seems to be that pod templates (SPARK-24434) already provide this feature. > Support for requesting GPU resources on K8S > ------------------------------------------- > > Key: SPARK-24491 > URL: https://issues.apache.org/jira/browse/SPARK-24491 > Project: Spark > Issue Type: New Feature > Components: Kubernetes > Affects Versions: 2.3.0 > Reporter: Alex Milowski > Priority: Minor > > If GPU resources are required for executor PODS in a multi-tentan k8s > cluster, the resource limits section needs to specify that GPUs are required > [1]. This is a simple request: > {{resources:}} > {{ limits:}} > {{ nvidia.com/gpu: 1 # requesting 1 GPU}} > A simple configuration limit could be specified (just like executor memory): > {{ --conf spark.kubernetes.executor.limit.gpus=1}} > [1] [https://kubernetes.io/docs/tasks/manage-gpus/scheduling-gpus/] > I have an implementation I will submit via a pull request. -- This message was sent by Atlassian JIRA (v7.6.3#76005) --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org For additional commands, e-mail: issues-h...@spark.apache.org