Hi folks, does

@task

support both

name

and

task_run_name

at the same time? It seems that

name

takes precedence over

task_run_name

, am I right?

The

name

param is a prefect core feature, and this name is displayed on things like the prefect generated logs for your flow run, and

flow.visualize()

. It should help you understand a flows structure and what each task does in a high level visualization. The

task_run_name

is I believe a prefect cloud or prefect server feature, and is used to tag a specific task run with a human readable name. If you do not set a

task_run_name

for a task, you should see that in different runs of the same flow, the same task gets a random task run name each time. Two word stuff like,

lunatic-landlord

or

emotional-alligator

. I haven’t ever set a

task_run_name

explicitly, but I believe setting it would use the same value for that tasks task run name between each flow run.

Thanks @emre

Yeah I tried that to. This will actually rename the task_run, I verified that when looking into graphql. However somehow the UI doesn't update or pull the real task_run_name

Maybe someone from the Prefect Team can help us sort it out 🙂

Hi all, hopefully I can help clear this up a bit. As @emre mentioned,

task_run_name

is a concept that only exists in Cloud/Server since it requires some sort of persistence. There are two objects being discussed here: task and task run; the first is the blueprint you give to prefect in the form of an extended or decorated task class, and the second is the actual instantiation (the run) of that task. When you provide a task with a

name

, you're naming its blueprint, but when you give it a

task_run_name

, you're templating its instantiation. The UI handles this in a few ways, both for backwards compatibility (task run names were removed as a column in the db early in 2020 but were reintroduced ~last month) and for readability/context: anything describing a

task

or

flow

will show the blueprint task name. Anything describing a run (

task_run

or

flow_run

) will show the templated

task_run_name

if it exists or will show a combination of the

task

(blueprint) name and its parent

flow_run

name. Since older versions of Cloud/Server didn't have task run names in the db, older versions of the UI won't show task run names.

Hi @nicholas, thanks for the clarification. However, I don’t get it, sorry 😅 With the piece of code I shared above I was expecting to see something like

Create table <table_name> in DWH

. But what I actually see is

create_table

. The flow runs in Prefect Server. What am I missing?

Which version of Prefect Server are you running?

0.13.13

Hm, I'm not sure why that wouldn't work then, I just tried it with this code:

import prefect
from prefect import Flow, task, Parameter


@task(task_run_name=lambda **kwargs: f"Create table {kwargs['table_name']} in DWH")
def create_table(table_name):
    # create some table
    return


with Flow("Task Run Names") as flow:
    table_name = Parameter("table_name", default="user")

    create_table(table_name=table_name)

I’ll double check Prefect Server and UI version

Same, your code works fine for me

Thanks again @nicholas

Did you get it sorted @ale?

Waiting for my SRE team to do the check

Gotcha, keep me posted!

👍

Fantastic 🙂