Retry failure on failed workflow.

[XiaoBoya]

I ran a parallel workflow using withParam configuration. When a node fails, the overall status of the workflow is marked as failed. However, the phase field of this node remains 'Running'. It also triggers the hook that gets called when the workflow fails. When I try to retry it from the UI, it tells me that the node's state is 'Running', so it cannot be retry.
Argo Workflow: v3.5.11
Kubernetes: v1.25.16

        - - name: check-node
            templateRef:
              name: drain
              template: check-node
              clusterScope: true
            arguments:
              parameters:
                - name: type
                  value: "{{`{{item}}`}}"
                - name: token
                  value: "{{`{{steps.get-token.outputs.parameters.token}}`}}"             
            withParam: '["NetworkUnavailable","MemoryPressure","DiskPressure","PIDPressure","Ready"]'

[jswxstw]

Sorry, you provided too little information. Have you searched existing issues and could not find a match for this bug?

[XiaoBoya]

Thank you for your support. I have added some information. I'm sorry, but due to issues with my testing environment, I currently cannot reproduce this phenomenon. There doesn't seem to be an exact match in the Issue list. The phenomenon observed is that when a step fails, the corresponding node's phase status remains as 'Running', thereby preventing retry operations. I will keep an eye out for this issue during subsequent usage. If it reoccurs, I will provide additional screenshots along with my comments. Thank you once again for your response.

[jswxstw]

Okay, v3.5.11 has actually fixed many issues related to workflows getting stuck. From the example you provided, there doesn't seem to be any obvious problems.
By the way, may I ask have you enabled Parallelism and FailFast? There is a known issue: #13806 related to this.

[XiaoBoya]

I did indeed configure these two parameters in the template (failFast: true, parallelism: 2), which is very similar to the issue you described. Thank you very much for your response!