Referring to this documention : here.

It mentions, variable substitution can be done for databricks bundle validate OR databricks bundle deploy OR databricks bundle run. The important word here is: OR.

I have a job, which has named params, where I pass on variables, which have a default value. Now, I would like to trigger a run of that job, with a different variable value and not change the value for all runs.

This is part of my task definition:

tasks:
  - task_key: ci_start
    job_cluster_key: job_cluster
    libraries:
      - whl: ../../dist/*.whl
        python_wheel_task:
        package_name: ci_workflows
        entry_point: ci_start
        named_parameters:
            postfix: ${var.postfix}

So, I am trying to substitute the value of postfix for just one particular run.

My variable declartion looks like this:

variables:
   postfix:
      description: Suffix used for integration tests during CI.
      default: "init"

So, if I run

databricks bundle deploy

databricks bundle run --var="postfix=abc" my_job

I still see the triggered run with the var postfix as init, which is the default value in the variable declaration, and not what was passed in the databricks bundle run command.

If however, I substitute the var value in databricks bundle deploy and then trigger databricks bundle run (even without specifying var, it then gets the new var value)

i.e.

databricks bundle deploy --var="postfix=abc"

followed by

databricks bundle run my_job

the value of postfix becomes abc.

But the problem is- this changes the default variable value in the job and not of a particular run of a job. I would like to retain the original default value of postfix (init) for all other runs (where value of postfix is not supplied), and only like to override this in a particular run

Is it not possible to trigger the run with a different variable value without changing the job definition? The documentation says **OR**, so it should be possible to do so.

what am I missing?