Skip to main content

Parameterization Using Job Options

The following tutorial walks you through how to parameterize a JobConfig using Job Options.

Example: Refunding customer payments

In this tutorial, we will use a simple example scenario to run a Job to execute some payment refunds to our customers.

Since this is a toy example, the JobConfig doesn't actually perform any financial transactions, but it will serve as a good example to illustrate how Job Options can be used.

Create the JobConfig

Install the example JobConfig from GitHub in a single one-liner:

kubectl create -f

This will install a JobConfig named adhoc-refund-customer-payments to the cluster.

Understanding the option configuration

If you installed the above example JobConfig, you can inspect the JobConfig to see that there are 3 options defined:

- name: customer_id
type: String
required: true

- name: refund_type
type: Select
required: true
default: "immediate"
- immediate
- best-effort

- name: dry_run
type: Bool
default: false
format: Custom
trueVal: "y"
falseVal: "n"

Each of these options have their own unique name, which corresponds to a unique key for substitution in the task template.

In our above example scenario, this corresponds to three different parameters:

  1. Customer ID: Specifies the unique ID of the customer that we want to execute a refund for.
  2. Refund Type: Specifies whether this is an immediate or best-effort refund. An immediate refund will be blocking, and may take several seconds or minutes to complete. On the other hand, a best-effort refund will be asynchronous, and we will simply add the refund request to a queue to be processed separately.
  3. Dry Run: Specifies whether this is a dry-run, which allows the user to simulate running the job without actually performing any financial transactions.

As such, we can see that a single Job will process a single customer at a time, and the refund type will determine whether this Job will be blocking and take several minutes, or terminate almost immediately.


In the case of sensitive transactions (such as this example), it may be wise to prevent multiple Jobs from executing at the same time for this JobConfig. As such, the Forbid concurrency policy is used to ensure that payments do not have race conditions resulting in unnecessary financial losses.

Understanding the taskTemplate

When a Job is scheduled or executed on an ad-hoc basis, the context variables in the taskTemplate will be substituted with the final rendered option values.

If you installed the above example JobConfig, you can inspect the JobConfig's taskTemplate field:

- name: container
image: bash
- bash
- "-c"
- |
# Display what we are about to do.
echo "[*] Will refund for customer '${option.customer_id}' using ${option.refund_type} refund method."

# Exit if only dry run.
if [[ "${option.dry_run}" == "y" ]]; then
echo "[!] Dry run mode specified, exiting."
exit 0

# Simulates some actual work to be done.
case "${option.refund_type}" in
echo "[*] Processing immediate refund for customer '${option.customer_id}' now..."
sleep 10
echo "[!] Succesfully processed refund."

sleep 1
echo "[!] Successfully enqueued best-effort refund for customer '${option.customer_id}', it will be processed within 3 business days."

echo "[*] Refund process complete."

The above taskTemplate is just an example to simulate the sequence of events in a real job that processes payment refunds. From this, we can see that:

  1. All job options can be substituted safely, even inside of a shell script.
  2. Job option substitutions can take place even in other fields, and is not limited to command like above.

Executing an ad-hoc job

To execute the ad-hoc job, run the following:

furiko run adhoc-refund-customer-payments

The job options will be prompted via an interactive shell, auto-filling any default values as necessary: