en/stable/examples/workflows/upstream/volumes_emptydir/

[giscus[bot]]

en/stable/examples/workflows/upstream/volumes_emptydir/

https://hera.readthedocs.io/en/stable/examples/workflows/upstream/volumes_emptydir/

[sourcx]

I'm trying to understand the Hera syntax but where is the @script here and the w.create()?
Also, is it me or does this example seem very verbose and not very idiomatic?

[flaviuvadan]

Hey @sourcx! There's no script here because the example uses a container template, not a script template. Also, w.create() is not added in examples because the workflows are not submitted to any Argo server in CI (yet).

For more context, the w variable is tested by Hera hera, and the tests compare examples with raw YAML examples from the upstream Argo Workflows repo

[flaviuvadan]

If you have questions about script usage I am happy to help!

[sourcx]

Thanks for the reply! I like the concept of Hera a lot but I'm not familiar enought with Argo yet to understand everything.I appreciate your offer to help, so here goes:

I'm in the process of converting an Argo workflow to Hera to see how that would work, it basically boils down to this:

kind: WorkflowTemplate
spec:
  - name: step1
    volumes:
    - name: workdir
      emptyDir: { }
    script:
      volumeMounts:
      - name: workdir
        mountPath: /workdir

So I was trying to use the @script, with Workflow, with Steps-syntax but could not figure out where to add the emptyDir.

[flaviuvadan]

In the YAML example above the workdir is set on the spec of the workflow template! In Hera something like

from hera.workflows import WorkflowTemplate, EmptyDirVolume, models as m, script

# the `volume_mounts` of the `script` informs Argo that this specific script template
# should mount the volume called `workdir` at path `/workdir` inside the script environment
@script(volume_mounts=[m.VolumeMount(name="workdir", mount_path="/workdir")])
def foo():
    print(42)


# here, the workflow template sets an empty dir "globally", at the workflow level
with WorkflowTemplate(name="my-wf-template", entrypoint="foo", volumes=EmptyDirVolume(name="workdir")) as w:
    foo()

print(w.to_yaml())

will print out

apiVersion: argoproj.io/v1alpha1
kind: WorkflowTemplate
metadata:
  name: my-wf-template
spec:
  entrypoint: foo
  templates:
  - name: foo
    script:
      command:
      - python
      image: python:3.8
      source: 'import os

        import sys
        # this gets added by Hera, can disable via `@script(..., add_cwd_to_sys_path=False)`
        sys.path.append(os.getcwd())

        print(42)'
      volumeMounts:
      - mountPath: /workdir
        name: workdir
  volumes:
  - emptyDir: {}
    name: workdir

[sourcx]

Thanks for taking the time to help me out here, much appreciated!

[codex3094]

I've put in place a workflow in which we perform a task for every date within the given date range- Each task is completed one after the other. Currently, If a job fails no failure message is sent and the task is skipped on all the subsequent dates. Would you kindly assist me how to send the failure message when the task fails.

[flaviuvadan]

how to send the failure message when the task fails.

Maybe something like this might help! Essentially you can set up an exit procedure that runs irrespective of the workflow result. Alternatively, you can set the on_failure field of tasks, and you can make those other tasks that run scripts that print out error messages to anywhere