apache-spark-on-k8s/spark

share hadoop-config configmap with multiple apps

chenchun opened this issue · 0 comments

Each spark app creates a new hadoop-config configmap may waste lots of etcd storage and memory of k8s master components. Many of configmaps are duplicate assume lots of apps may share the same hdfs. I suggest adding an spark.kubernetes.configmap.hadoop.config config to make this possible.