[archived] "Resource Template" docs discussion

[giscus[bot]]

argo-workflows/walk-through/kubernetes-resources/

https://argoproj.github.io/argo-workflows/walk-through/kubernetes-resources/

[b-luu]

Is it possible to pass the manifest via an input artifact?

[terrytangyuan]

Yes, use manifestFrom instead of manifest.

[ONGmartino]

@terrytangyuan could you please provide an example?

[ciiiii]

@ONGmartino FYI

- name: apply
  inputs:
   artifacts:
   - name: manifest
     path: /tmp/manifest    
  resource:
    action: apply
    manifestFrom:
      artifact:
        name: manifest

[mikekantzer-drizly]

Is it possible to get the generated name of a resource?

[mikekantzer-drizly]

Never mind: just saw the link to

argo-workflows/examples/k8s-jobs.yaml

Lines 44 to 50 in 83501a4

	outputs:
	parameters:
	# job-name is extracted using a jsonPath expression and is equivalent to:
	# `kubectl get job <jobname> -o jsonpath='{.metadata.name}'`
	- name: job-name
	valueFrom:
	jsonPath: '{.metadata.name}'

[robinsmedberg]

Is it possible to access input artifacts from created resources? I'm creating a spark job and would like to access the input artifact instead writing code to download it from s3

[agilgur5]

It works like any other template, so you do have access to Workflow Variables.

[ebuildy]

What about the crucial information => Which service-account is used here?

[agilgur5]

Same as other template types, whichever SA you've assigned to the template or Workflow. So make sure your SA has proper RBAC for your resource's kind and action

[KingsonKai]

how to use complex condition in the successCondition, like below, it can't work

      resource:
        action: create
        failureCondition: status.status == failed
        successCondition: status.phase == recover && status.status == success

[KingsonKai]

I have get my answer:

      resource:
        action: create
        failureCondition: status.status == failed
        successCondition: status.phase == recover,status.status == success

[yingzhe-qu]

is it possible to get the outputs artifacts , eg manifest contents is tfjob

[yingzhe-qu]

  - name: train
    outputs:
      artifacts:
        - name: step4-code
          path: /usr/local/code
    resource: 
      action: create
      setOwnerReference: true
      failureCondition: status.replicaStatuses.Worker.failed > 0
      successCondition: status.replicaStatuses.Worker.succeeded = 2
      manifest: |
        apiVersion: kubeflow.org/v1
        kind: TFJob
        metadata:
          name: tfjob-enti-test1
          namespace: kubeflow
        spec:
          tfReplicaSpecs:
          # ...

and then ,
i can not get the outputs artifacts step4-code

[agilgur5]

Creating a resource in Kubernetes does not return any files. As in, the output is the same as if you had used a script with kubectl create -f my-tf-job.yaml.

I'm not sure what you're trying to get from /usr/local/code, but that path would be located inside the container that ran kubectl create

[dmarquez-splunk]

Is it possible to use jsonpath as a way of getting a particular field? I am having trouble accessing particular fields within a status condition. Since it is an array I am having trouble using the jsonpath syntax as I receive the following error:

condition:

successCondition: "status.conditions[?(@.type=='Synced')].status=True"

error:

time="2024-05-28T19:12:51.550Z" level=fatal msg="success condition 'status.conditions[?(@.type=='Synced')].status=True' failed to parse: unable to parse requirement: <nil>: Invalid value: \"status.conditions[?\": name part must consist of alphanumeric characters, '-', '_' or '.', and must start and end with an alphanumeric character (e.g. 'MyName',  or 'my.name',  or '123-abc', regex used for validation is '([A-Za-z0-9][-A-Za-z0-9_.]*)?[A-Za-z0-9]')"