I have a dagster job which runs fine and I have wr...
# ask-community
p
I have a dagster job which runs fine and I have wrapped it around a simple scheduler and I hit the error:
Copy code
Attempted to build memoized execution plan without providing a persistent DagsterInstance to create_execution_plan.
I set the tags as
return RunRequest(run_key=None, run_config=base_config, tags={"MEMOIZED_RUN_TAG": "true"})
I could not find much more in terms of docs. Does anyone have any idea how to proceed?
j
hey @Pankaj Daga - for what purpose are you setting the “MEMOIZED_RUN_TAG”?
p
I was getting this error and did a basic google search and this had come up. Sorry the quotes were a mistake but other than that, it still did not work with or without the tag
j
so if you dont have the tag at all it also doesnt work?
return RunRequest(run_key=None, run_config=base_config)
that tag should be entirely unnecessary, unless you’re doing some deprecated stuff with memoized jobs
p
My schedule code is as follows:
Copy code
@schedule(
    name="train_and_predict",
    cron_schedule="10 * * * *",  
    job=train__plus,
)
def train_schedule():
    with open(file_relative_path(__file__, "train_and_predict.yaml")) as f:
        bridge_config = yaml.safe_load(f)

    with open(file_relative_path(__file__, "dev__train.yaml")) as f:
        base_config = yaml.safe_load(f)

    base_config.update(bridge_config)

    return RunRequest(run_key=None, run_config=base_config, tags={MEMOIZED_RUN_TAG: "true"})
yeah, without it, I get the same error. Basically about memoization
j
huh that’s odd. either way, you shouldnt need that tag, so you can remove it. How are you running dagster?
also what version are you on?
p
on the clud. The version is 1.3.13
So turning the scheduler manually from the frontend
let me try again w/o the tag. one sec
j
ok - if that doesn’t work, can you share the code for the
train__plus
job? feel free to dm it if needed
👍 1
p
Just FYI: The job runs fine. I can call the train_and_predict job without issues
j
the train_and_predict job? i see from the code that the schedule is called train_and_predict, but that the job it’s launching is called train__plus. just want to confirm that we’re talking about the same thing