fix: correct manual retry logic. Fixes #14124

[jswxstw]

Fixes #14124

Motivation

In certain scenarios, manual retries do not work properly.

Modifications

• Retry all failed execution nodes
• Reset all group nodes and non-boundary parent nodes if needed
• Do not retry nodes which their descendant nodes are Succeeded

Verification

Case 1:

apiVersion: argoproj.io/v1alpha1
kind: Workflow
metadata:
  generateName: http-template-
spec:
  entrypoint: main
  arguments:
    parameters:
      # good: https://raw.githubusercontent.com/argoproj/argo-workflows/4e450e250168e6b4d51a126b784e90b11a0162bc/pkg/apis/workflow/v1alpha1/generated.swagger.json
      # bad: https://raw.githubusercontent.com/argoproj/argo-workflows/thisisnotahash/pkg/apis/workflow/v1alpha1/generated.swagger.json
      - name: url
        value: "https://raw.githubusercontent.com/argoproj/argo-workflows/thisisnotahash/pkg/apis/workflow/v1alpha1/generated.swagger.json"
  templates:
    - name: main
      steps:
        - - name: fail1
            template: http
          - name: fail2
            template: http
    - name: http
      http:
        url: "{{workflow.parameters.url}}"

Case 2:

apiVersion: argoproj.io/v1alpha1
kind: Workflow
metadata:
  name: workflow-exit-handler-fail
spec:
  entrypoint: echo
  onExit: exit-handler
  templates:
  - name: echo
    http:
      url: "https://raw.githubusercontent.com/argoproj/argo-workflows/4e450e250168e6b4d51a126b784e90b11a0162bc/pkg/apis/workflow/v1alpha1/generated.swagger.json"
  - name: fail
    container:
      image: alpine:3.18
      command: [sh, -c]
      args: ["exit 1"]
  - name: exit-handler
    steps:
      - - name: exit-handler-task
          template: fail

Case 3:

apiVersion: argoproj.io/v1alpha1
kind: Workflow
metadata:
  name: workflow-steps-with-retry-fail
spec:
  entrypoint: main
  templates:
  - name: main
    steps:
    - - name: retry-step-group-case
        template: fail-step-group
  - name: fail-with-rate
    container:
      image: python:alpine3.6
      command: ["python", -c]
      args: ["import random; import sys; exit_code = random.choice([0, 1]); sys.exit(exit_code);"]
  - name: fail-step-group
    steps:
    - - name: step1
        template: fail-with-rate
    - - name: step2
        template: fail-with-rate
    - - name: step3
        template: fail-with-rate
    retryStrategy:
      limit: "1"

Case 4

apiVersion: argoproj.io/v1alpha1
kind: Workflow
metadata:
  name: dag-contiue-on-fail
spec:
  retryStrategy:
    limit: 1
  entrypoint: workflow
  templates:
  - name: workflow
    dag:
      tasks:
      - name: A
        template: hello-world
      - name: B
        depends: "A"
        template: intentional-fail
      - name: C
        depends: "A"
        template: hello-world
      - name: D
        depends: "B.Failed && C"
        template: hello-world
      - name: E
        depends: "A"
        template: intentional-fail
      - name: F
        depends: "A"
        template: hello-world
      - name: G
        depends: "E && F"
        template: hello-world

  - name: hello-world
    container:
      image: busybox
      command: [echo]
      args: ["hello world"]

  - name: intentional-fail
    container:
      image: alpine:latest
      command: [sh, -c]
      args: ["echo intentional failure; exit 1"]

Documentation

[jswxstw]

/retest

[isubasinghe]

Nice, thanks for simplifying the logic and fixing the bugs in the handling of http and plugin nodes.

[jswxstw]

/retest