Flyte enables production-grade orchestration for machine learning workflows and data processing created to accelerate local workflows to production.

Flyte

Hi!
I’m trying to add a kubernetes pod label to all pods generated as a result of launching a workflow. Execution code:
```remote.execute(
        workflow,
        inputs=inputs,
        options=Options(labels=common_models.Labels({"flyte.user": os.getenv("USER", "na")}))```
where `remote` is of type `FlyteRemote`.
I can see that the workflow’s pod has the expected label, but none of the subworkflows do. Is there a way to have all pod labels be passed to all child pods of a workflow pod?

Hi <@U03ARAAMQV6>, I'm going to look into this here. Subworkflows should inherit this attribute from their parent workflow during execution. However, it seems that's not happening.

ty! yes, that’s our observation. If you see something different than this, please let me know and we can investigate what might be different

Sorry, did get around to making an issue but I do have this on my list. Hopefully get to it in the next few days.

Would you mind creating an issue? Like I said, I'll pick it up right away.

Is this because we are just using `app` as a label when creating the pods <https://github.com/flyteorg/flytepropeller/blob/master/manager/manager.go#L91-L93> ?

<@USPH1DVD2> so the flytepropeller manager is a way of starting multiple flytepropeller instances. the code you linked is setting the pod labels on those managed instances rather than the pods flytepropeller starts for tasks.

Hey <@U03ARAAMQV6>, I'm not able to reproduce this here. All subworkflows are adopting the parent workflows Pod labels. Using your example `FlyteRemote` commands:
```from flytekit.configuration import Config, PlatformConfig
from flytekit.models.common import Labels
from flytekit.remote import FlyteRemote
from flytekit.remote.remote import Options

import os

config = Config.for_endpoint("localhost:8089", insecure=True)

remote = FlyteRemote(config=config)

workflow = remote.fetch_workflow(project="flytesnacks", domain="development", name="core.control_flow.subworkflows.parent_wf", version="v0.3.100")

remote.execute(workflow, inputs={"a": 1}, options=Options(labels=Labels({"flyte.user": os.getenv("USER", "na")})), project="flytesnacks", domain="development")```
where the workflow is one of our <https://github.com/flyteorg/flytesnacks/blob/6997c290b9795752ae8d3a246c7ba96f69d20324/cookbook/core/control_flow/subworkflows.py#L65|flytesnacks subworkflow tests> I am seeing all the pods (including those launched by the subworkflow) have the correct labels set:
```hamersaw@ragnarok:~$ kubectl -n flytesnacks-development get pod fc09411a31e134117aaa-node-t1-parent-0 -o yaml
apiVersion: v1
kind: Pod
metadata:
  labels:
    flyte.user: hamersaw
// ...

hamersaw@ragnarok:~$ kubectl -n flytesnacks-development get pod fc09411a31e134117aaa-n1-0-n0-0 -o yaml
apiVersion: v1
kind: Pod
metadata:
  labels:
    flyte.user: hamersaw
// ...

hamersaw@ragnarok:~$ kubectl -n flytesnacks-development get pod fc09411a31e134117aaa-n1-0-n1-0 -o yaml
apiVersion: v1
kind: Pod
metadata:
  labels:
    flyte.user: hamersaw
// ... ```

Can you maybe provide a little more information on this? Is the workflow more complex than a single subworkflow?

we just tested this out in flytekit1.1 and it works. It didn’t work in flytekit0.3. Thank you for investigating this and I apologize for the goose-chase

no problem! glad everything is working :smile: