is there a way to get additional insight to what api calls are being made under the hood of the FivetranSyncTask? i’m running it like so and for some reason it’s running all of my syncs as manual syncs even though i have manual set to

False

. Code in thread

What version of Prefect are you on? Could you move the code part to the thread to make the main channel cleaner when you get the chance?

@task
def fivetran_sync_map_fn(connector_info: tuple[str, str, bool]) -> dict[str]:
    """
    This represents a task in a flow that would trigger a Fivetran sync.
    To use this task, you first need to set up your connector in the Fivetran dashboard:
        <https://fivetran.com/dashboard>
    You can retrieve the `connector_id` parameter by clicking into the connector under the "Setup" tab - it's titled "Fivetran Connector ID".
    Setting this task will override the sync schedule set in the UI by default. If you don't want this, then set the param `manual=True`.

    :param connector_info (tuple[str, str, bool]): tuple of (fivetran connector id, task name, manual)

    :return (dict): connector_id (str) and succeeded_at (timestamp str)
    """
    connector_id, _, manual = connector_info
    fivetran = FivetranSyncTask(connector_id=connector_id)
    status = fivetran.run(
        api_key=API_KEY,
        api_secret=API_SECRET,
        poll_status_every_n_seconds=20,
        manual=manual,
    )
    return status

even when i hard code it and just run

status = fivetran.run(
        api_key=API_KEY,
        api_secret=API_SECRET,
        poll_status_every_n_seconds=20,
        manual=False,
    )

it’s kicking off as a manual run. but i’m not sure what tools i have to dig into this deeper

I think the manual stuff was added in 0.15.7

i’m on prefect 1.0

I don’t have a better suggestion for the calls under the hood other than reading the task code 😅

what debug logs should i add though? i basically want to see if that

if manual

block is being run but i don’t think i can add a debugger there right?

Yes you can locally by editing the prefect package and just see which

if

path it took

okay, after debugging i actually think that maybe fivetran changed something but now your documentation may be incorrect:

Args:
            - api_key (str): `API key` per <https://fivetran.com/account/settings>; should be secret!
            - api_secret (str): `API secret` per <https://fivetran.com/account/settings>; should be secret!
            - connector_id (str, optional): if provided, will overwrite value provided at init.
            - poll_status_every_n_seconds (int, optional): this task polls the Fivetran API for status,
                if provided this value will override the default polling time of 15 seconds.
            - manual (bool, optional): if provided, will overwrite Prefect's changes
                to the Fivetran connector's schedule, keeping it on Fivetran auto scheduling

^ i interpret that as if

manual=True

then prefect won’t override fivetran’s scheduling. from my tests it seems that setting

manual=True

does override fivetran’s scheduling, and setting

manual=False

won’t override their scheduling https://github.com/PrefectHQ/prefect/blob/master/src/prefect/tasks/fivetran/fivetran.py

Gotcha ok will take a look at it. Thanks for raising! Does that mean manual=True is what you intend?

yes, but i was stuck doing the opposite of what i wanted to do but at least we have a fix that doesn’t require a patch on your side. i think you just have to change the documentation. for what it’s worth i tested this earlier and it worked as you documented it. but i’m not sure if something changed on their side because i don’t see it in the release notes

Thanks for the info and looking into it