-
Notifications
You must be signed in to change notification settings - Fork 129
Description
Describe the issue
When using is_single_node: true with kind: CLASSIC_PREVIEW in a job cluster configuration, the CLI still emits a warning about single-node cluster misconfiguration. This warning is unnecessary because is_single_node: true automatically sets the required spark_conf, custom_tags, and num_workers values.
The validation logic appears to check for manual single-node configuration (num_workers: 0 + spark_conf + custom_tags) without considering that is_single_node: true handles this automatically.
Configuration
resources:
jobs:
my_job:
name: "My Job"
job_clusters:
- job_cluster_key: "single_node_cluster"
new_cluster:
kind: CLASSIC_PREVIEW
is_single_node: true
spark_version: "15.4.x-scala2.12"
node_type_id: "m5.xlarge"
num_workers: 0
tasks:
- task_key: "my_task"
job_cluster_key: "single_node_cluster"
# ... task configuration
Steps to reproduce the behavior
- Create a bundle with a job cluster using is_single_node: true and kind: CLASSIC_PREVIEW
- Run databricks bundle validate or databricks bundle deploy
- See warning:
Warning: Single node cluster is not correctly configured
at resources.jobs.<job_name>.job_clusters[0].new_cluster
in resources/init.py:49:1
num_workers should be 0 only for single-node clusters. To create a
valid single node cluster please ensure that the following properties
are correctly set in the cluster specification:
spark_conf:
spark.databricks.cluster.profile: singleNode
spark.master: local[*]
custom_tags:
ResourceClass: SingleNode
Expected Behavior
When is_single_node: true is set with kind: CLASSIC_PREVIEW, the CLI should recognize this as a valid single-node cluster configuration and not emit the warning. The is_single_node option was specifically introduced to simplify single-node cluster configuration by automatically setting the required spark_conf, custom_tags, and num_workers values.
Actual Behavior
The CLI emits a warning suggesting manual configuration of spark_conf and custom_tags, even though is_single_node: true already handles this automatically. This creates confusion and unnecessary noise in the output.
OS and CLI version
- CLI version: v0.281.0
- OS: macOS 15.5
Related issues:
- DAB deployment fails with
Error: cannot create job: NumWorkers could be 0 only for SingleNode clusters#1546 - Original single-node cluster deployment issue