DAB cannot create mlflow model: Missing required field: tag value when targets is development mode #922
Description
Databricks CLI version
$ databricks -v
Databricks CLI v0.208.2Bug reproduction
When deploying a model resource to a target with development mode, the automatic tagging mechanism added a tag that does not comply to Databricks create model API and caused the following error
$ databricks bundle deploy
Starting upload of bundle files
Uploaded bundle files at /Users/<redacted>/.bundle/test-dab/dev/files!
Starting resource deployment
Error: terraform apply: exit status 1
Error: cannot create mlflow model: Missing required field: tag value
with databricks_mlflow_model.test_model,
on bundle.tf.json line 22, in resource.databricks_mlflow_model.test_model:
22: } File databricks.yml
bundle:
name: test-dab
targets:
dev:
mode: development
default: true
resources:
models:
test_model:
name: test-modelFile bundle.tf.json output from databricks bundle validate
{
"terraform": {
"required_providers": {
"databricks": {
"source": "databricks/databricks",
"version": "1.28.0"
}
}
},
"provider": {
"databricks": {}
},
"resource": {
"databricks_mlflow_model": {
"test_model": {
"name": "[dev <redacted>] test-model",
"tags": [
{
"key": "dev"
}
]
}
}
}
}From the file we can observe that the tags "key": "dev" does not have a value that caused the error. For production mode there is no automatic tag added so no error is raised.