Lifecycle ping recipe for StackStorm Orquesta workflows

[bartekrutkowski]

Hi,

I'm Bartek Rutkowski, founder of Watchgoose.com, a platform for monitoring cron/kubernetes/backup jobs and other scheduled tasks.

I have checked StackStorm v3.9.0, including the built-in core.http action and Orquesta transition engine. A workflow can report start, success, and failure to Watchgoose entirely in YAML:

version: 1.0

description: Report one workflow lifecycle to Watchgoose.

tasks:
watchgoose_start:
action: core.http
input:
method: POST
url: https://watchgoose.com/ping/<uuid>/start
timeout: 10
next:
- when: <% succeeded() %>
do: actual_work
- when: <% failed() %>
do: actual_work

actual_work:
action: core.noop
next:
- when: <% succeeded() %>
do: watchgoose_success
- when: <% failed() %>
do: watchgoose_failure

watchgoose_success:
action: core.http
input:
method: POST
url: https://watchgoose.com/ping/<uuid>
timeout: 10
next:
- when: <% failed() %>
do: noop

watchgoose_failure:
action: core.http
input:
method: POST
url: https://watchgoose.com/ping/<uuid>/fail
timeout: 10
next:
- when: <% succeeded() %>
do: fail
- when: <% failed() %>
do: fail

There is one important Orquesta detail in the complete example: failed monitoring on the success path transitions to noop while both outcomes of the failure ping task transition to the fail engine command. Without that last command the notification task can look like remediation and accidentally turn the workflow green.

With that wiring Watchgoose can detect a missing expected completion, an explicit failure, or a workflow still open beyond grace. Would a small Watchgoose recipe be useful in the Orquesta error handling or core.http documentation? I can provide PR-ready wording around this workflow.

Also, Watchgoose has an Open Source support pledge where qualifying Open Source projects and nonprofits may receive a needs matched paid plan for free. I would be happy to arrange that for DBackup own scheduled infrastructure if you'd find that useful.

Kind regards,
Bartek Rutkowski