[bug] <Bug Name>

[HumairAK]

Environment

• How do you deploy Kubeflow Pipelines (KFP)? standalone and multi-user

• KFP version: 2.4

To find the version number, See version number shows on bottom of KFP UI left sidenav. -->

• KFP SDK version: 2.11

Steps to reproduce

• Use a simple dag pipeline like the following:

@dsl.component()
def fail():
    import sys
    sys.exit(1)

@dsl.component()
def hello_world():
    print("hellow_world")

@dsl.component()
def post_msg():
    print(f"this is a message")

@dsl.pipeline
def pipeline():
    post_msg_task = post_msg().set_caching_options(enable_caching=False)
    with dsl.ExitHandler(exit_task=post_msg_task):
        hello_task = hello_world().set_caching_options(enable_caching=False)
        fail_task = fail().set_caching_options(enable_caching=False)
        fail_task.after(hello_task)


compiler.Compiler().compile(
    pipeline_func=pipeline,
    package_path=__file__.replace('.py', '-v2.yaml'))

Or any other control flow that utilizes sub dags, i.e. loops, nested pipelines, exit handlers, etc.

Expected result

The subdag status should stay as loading until all tasks are completed.

Materials and reference

dag-status-behavior.webm

Labels

/area backend

---

Impacted by this bug? Give it a 👍.