Circuit Breakers for CronWorkflows

[simster7]

Summary

What change needs making?

Implement a circuit breaker for cron workflows. For example, if a workflow has had multiple failures in a row, pause the workflow automatically. Or pause the workflow after a certain number of executions (https://github.com/argoproj/argo-workflows/issues/7201).

Either an expression:

apiVersion: argoproj.io/v1alpha1
kind: CronWorkflow
metadata:
  name: hello-world
spec:
  schedule: "* * * * *"
  backOff: "{{cronworkflow.failures}} > 3"

Or a field:

apiVersion: argoproj.io/v1alpha1
kind: CronWorkflow
metadata:
  name: hello-world
spec:
  schedule: "* * * * *"
  backOff:
    failures: 2

Use Cases

When would you use this?

Cron workflows are the only CRD that act on their own (i.e. no explicit action is needed to spawn workflows from cron workflows), so a circuit breaker could be useful.

---

Message from the maintainers:

Love this enhancement proposal? Give it a 👍. We prioritise the proposals with the most 👍.

[terrytangyuan]

Duplicate of https://github.com/argoproj/argo-workflows/issues/5659?

[simster7]

More like an extension

[simster7]

@alexec Any thoughts on the spec proposal or this feature overall? Want to make sure we are all aligned before implementation begins.

[alexec]

I think the concept is fine. I'm not sure many people are asking for it, so I don't see the core team working on it soon.

[tooptoop4]

@eduardodbr since https://github.com/argoproj/argo-workflows/pull/12305 was merged do u think this issue can be closed?

[eduardodbr]

I believe it can be closed. @agilgur5 do you agree?

[agilgur5]

Per https://github.com/argoproj/argo-workflows/issues/5659#issuecomment-2381628641, I'm not sure either of these two were resolved, since they both specifically say "pause" and "suspend", which is intentionally different from stopStrategy per https://github.com/argoproj/argo-workflows/pull/12696#discussion_r1516670248