Choosing what happens after a failed step

Context — a non-critical notification must not block the load

The task is currently manual and its assumptions are not recorded. Hydra turns it into files that can be reviewed and rerun.

Where things stand

• The current result is fragile. Its assumptions are split between tools, clicks and memory.
• Reruns are uncertain. The write mode or orchestration rule is not visible beside the data.
• Evidence is missing. A colleague cannot compare a declared rule with a concrete before and after state.

The question

How do you make on_failure explicit in workflow.yaml and verify the resulting execution states?

• Keep jobs independently runnable
• Give every workflow step a unique name
• Validate the graph before execution
• Read every step status in the summary

The solution in one line

A small workflow.yaml, one validation command and one run with observable step states.

Steps

1. prepare the jobs and boundary

1. Name the work the workflow will coordinate.
2. Compare the file with the explanation in the workbench.
3. Record the shown check before moving to the next step.

workflow.yaml
jobs/
  sales/
  inventory/
  publish/

2. declare on_failure

1. Write the trigger, steps and policy.
2. Compare the file with the explanation in the workbench.
3. Record the shown check before moving to the next step.

workflow:
  version: "1.0"
  name: on_failure_workshop
  description: "Choosing what happens after a failed step"
  trigger:
    type: manual
  steps:
    - name: notify
      type: action
      action: webhook
      params:
        url: ${ENV:OPS_WEBHOOK_URL}
      on_failure: continue
    - name: load_sales
      type: job
      job: ./jobs/sales
      depends_on: [notify]

3. validate the graph

1. Catch missing jobs, duplicate names and invalid dependencies.
2. Compare the file with the explanation in the workbench.
3. Record the shown check before moving to the next step.

$ hdrctl workflow validate workflow.yaml

✅ Workflow valid: on_failure_workshop
Steps: 1

4. run the workflow

1. Execute and follow step states.
2. Compare the file with the explanation in the workbench.
3. Record the shown check before moving to the next step.

$ hdrctl workflow run workflow.yaml

notify      failed · continued
load_sales  succeeded

5. read the orchestration result

1. Distinguish success, skip, retry and continuation.
2. Compare the file with the explanation in the workbench.
3. Record the shown check before moving to the next step.

notify      failed · continued
load_sales  succeeded

Expected result

• The workflow validates before execution.
• notify failed · continued
• The run summary explains every terminal state.

Reading the results

Read the summary as a graph, not a flat log. Use `continue` only when downstream work remains valid without the failed step.

What the counters do—and do not—prove

They prove how many records entered and left this run. They do not replace checking the target schema, the business meaning of values or the reason a workflow step was skipped.

The rule to carry forward

Use `continue` only when downstream work remains valid without the failed step.

Did we answer the question?

Before and after

• Before — a non-critical notification must not block the load requires a person to remember the order, options and checks.
• After — one reviewed manifest and one command produce the same observable result.
• What is really gained — The durable gain is the contract: a colleague can read the configuration, reproduce the run and challenge the assumptions.

Close by

• sources.yaml
• transformations.yaml
• destinations.yaml
• workflow.yaml
• Tutorial — your first job
• All Guide workshops