Closed ranli-warnermedia closed 2 months ago
Indeed, in 0.228.0 there was a behaviour change regarding how variables are being merged which resulted in the change of behaviour. In your particular case as a workaround until we prepare a fix you can remove type: complex
line from target override for your variable and variables will be merged correctly
Describe the issue
clusters.yml
databricks.yml
In cli version
v0.227.1
and previous versions,databricks bundle validate -t dev --output json
passes without warnings, and the variables get substituted in the jobs.In cli version
v0.228.0
,databricks bundle validate -t dev --output json
returns 2 warnings:and the variables do not get substituted in the jobs.
Configuration
Install
v0.228.0
curl -fsSL https://raw.githubusercontent.com/databricks/setup-cli/main/install.sh | sh
Installv0.227.1
curl -fsSL https://raw.githubusercontent.com/databricks/setup-cli/v0.227.1/install.sh | sh
Steps to reproduce the behavior
databricks bundle validate -t dev --output json
Expected Behavior
type attributes in complex variables should not be considered as a field
Actual Behavior
type attributes in complex variables is considered as a field, and all the fields within the variable are unable to parse.
OS and CLI version
v0.228.0
Is this a regression?
Yes. work for all previous version.
Debug Logs
Output logs if you run the command with debug logs enabled. Example: databricks bundle deploy --log-level=debug. Redact if needed