[BUG] spark-operator-spark serviceaccount lacks necessary permissions
peter-mcclonski opened this issue · 2 comments
Description
Execution of example jobs fails due to insufficient privileges for the spark-operator-spark service account in the default namespace.
- ✋ I have searched the open/closed issues and my issue is not listed.
Reproduction Code [Required]
Steps to reproduce the behavior:
- kubectl apply -f https://raw.githubusercontent.com/peter-mcclonski/spark-on-k8s-operator/fix-examples/examples/spark-pi.yaml
- Wait for the job to fail
Expected behavior
Documentation for the examples should explain the need to specify --set sparkJobNamespaces=default, OR the default value of sparkJobNamespaces should be [default]
Actual behavior
The job fails due to the configured service account lacking access to the default namespace.
Terminal Output Screenshot(s)
Environment & Versions
- Spark Operator App version: 1.4.5
- Helm Chart Version: 1.2.14
- Kubernetes Version: 1.29.3
- Apache Spark version: 3.5.0
Additional context
The core issue here seems to be reflected in this section of the docs: https://github.com/kubeflow/spark-operator/blob/master/docs/quick-start-guide.md#about-the-spark-job-namespace
The bug is effectively an incorrect default in the values.yaml. We supply an empty list, rather than [""]
I believe this was introduced by #1988