magland / dendro

4 stars 2 forks source link

Failed jobs disappear from UI #6

Closed luiztauffer closed 1 week ago

luiztauffer commented 1 week ago

Failed jobs disappear from the UI after new jobs are submitted

magland commented 1 week ago

Here's the signature for submit_job, and I have just now added the docstring

def submit_job(
    *,
    service_name: str,
    job_definition: DendroJobDefinition,
    required_resources: DendroJobRequiredResources,
    target_compute_client_ids: Union[List[str], None] = None,
    tags: List[str] = [],
    skip_cache: bool = False,
    rerun_failing: bool = False,
    delete_failing: bool = False
):
    """
    Submit a job to the Dendro service.

    Parameters
    ----------
    service_name : str
        The name of the service to submit the job to.
    job_definition : DendroJobDefinition
        The job definition.
    required_resources : DendroJobRequiredResources
        The required resources for the job.
    target_compute_client_ids : List[str], optional
        The compute client IDs to target for the job. If None, all compute
        clients are targeted.
    tags : List[str], optional
        The tags to associate with the job.
    skip_cache : bool, optional
        Whether to skip the cache for the job. By default, if another job with
        the same job definition hash exists on this service, then that job will
        be returned instead of creating a new job. See also rerun_failing and
        delete_failing.
    rerun_failing : bool, optional
        Whether to resubmit jobs that have failed. This only applies when
        skip_cache is False.
    delete_failing : bool, optional
        Whether to delete jobs that have failed when resubmitting. This only
        applies when skip_cache is False and rerun_failing is True.

    Returns
    -------
    DendroJob
        The job that was submitted, or the existing job if found in the cache.
    """

What do you have set for skip_cache, rerun_failing, and delete_failing?

luiztauffer commented 1 week ago

ah alright, that was the problem, delete_failing=True, thanks!