This example shows:

• Binding a script implementation to a standard lifecycle operation
• Binding a script implementation to a custom operation
• Running scripts via Fabric (SSH)
• Implementing a workflow as a Python script
• Running a workflow
• Resuming a workflow

First, upload the blueprint:

cfy blueprints upload blueprint.yaml -b execute_workflow

Next, create a deployment (this command-line example is for OpenStack. For AWS, GCP or Azure, replace openstack with aws, gcp or azure, respectively):

cfy deployments create dep_1 -b execute_workflow -i infra_name=openstack

To install the topology:

cfy executions start install -d dep_1

The out-of-the-box install workflow runs the standard lifecycle operations, so you should see printouts from these operations (you can find the respective scripts inside the scripts directory).

Often, there is a need to execute a specific operation on the topology. This can be done by the execute_operation workflow. In this example, we will run the poll operation on the app node.

cfy executions start execute_operation -d dep_1 -p operation=maintenance.poll -p node_ids=[app]

This command will execute the poll operation of the maintenance interface, on all instances of the app node template.

You can also execute the operation on a specific instance of the node template, if the node template has multiple instances. For that, you will need to obtain the node instance ID. You can use the following command:

cfy node-instances list -d dep_1 -n app

And then (assuming the node instance ID is app_123456):

cfy executions start execute_operation -d dep_1 -p operation=maintenance.poll -p node_instance_ids=[app_123456]

The blueprint defines a custom workflow called rollout. The workflow is mapped to a script.

To execute the workflow, run the following command:

cfy executions start rollout -d dep_1

This workflow runs the maintenance.update and maintenance.commit operations in sequence (see the workflow implementation in scripts/rollout.py).

The maintenance.commit operation contains a condition: the operation will fail if the app node instance has a runtime property called fail_commit and its value is True.

By default, this runtime property doesn't exist on any node instance.

To demonstrate how to resume a workflow, we will set the runtime property's value from the command line.

First, obtain the node instance ID:

cfy node-instances list -d dep_1 -n app

(We will assume that the node instance ID is app_123456)

Next, set the runtime property value:

cfy node-instances update-runtime -p fail_commit=True app_123456

Now, let's run the workflow again:

cfy executions start rollout -d dep_1

The execution will run, however it will fail at the maintenance.commit operation.

Now, we will remove the runtime property and then resume the execution (we will assume that the execution ID is abcdefgh):

cfy node-instances delete-runtime -p fail_commit app_123456
cfy executions resume abcdefgh

This will trigger the execution's resumption. You can then use the cfy events list command (optionally with the --tail parameter) to look at the execution log:

cfy events list abcdefgh