https://outerbounds.com/ logo
Join Slack
Powered by
  • w
    Hi team, I am trying to use devstack to run one of my flow using kubernetes but getting the following error, wondering if I can get some help with this
    Copy code
    poetry run python metaflow_jobs/flows/train.py --with kubernetes --environment=pypi run
Metaflow 2.19.7 executing CovtypeTrainingPipelineFlow for user:chen.liang
Project: covtype, Branch: user.chen.liang
Validating your flow...
    The graph looks good!
Running pylint...
    Pylint not found, so extra checks are disabled.
2025-11-14 08:51:19.058 Bootstrapping virtual environment(s) ...
2025-11-14 08:51:19.213 Virtual environment(s) bootstrapped!
2025-11-14 08:51:19.577 Workflow starting (run-id 2), see it in the UI at <http://localhost:3000/CovtypeTrainingPipelineFlow/2>
2025-11-14 08:51:20.588 [2/start/4 (pid 81134)] Task is starting.
2025-11-14 08:51:21.352 [2/start/4 (pid 81134)] [pod t-07ea3803-xkdgx-jth29] Task is starting (Pod is pending, Container is waiting - ContainerCreating)...
2025-11-14 08:51:21.689 [2/start/4 (pid 81134)] [pod t-07ea3803-xkdgx-jth29] Setting up task environment.
2025-11-14 08:51:26.033 [2/start/4 (pid 81134)] [pod t-07ea3803-xkdgx-jth29] Downloading code package...
2025-11-14 08:51:26.498 [2/start/4 (pid 81134)] [pod t-07ea3803-xkdgx-jth29] Code package downloaded.
2025-11-14 08:51:26.526 [2/start/4 (pid 81134)] [pod t-07ea3803-xkdgx-jth29] Task is starting.
2025-11-14 08:51:26.913 [2/start/4 (pid 81134)] [pod t-07ea3803-xkdgx-jth29] Bootstrapping virtual environment...
2025-11-14 08:51:33.352 [2/start/4 (pid 81134)] [pod t-07ea3803-xkdgx-jth29] Bootstrap failed while executing: set -e;
2025-11-14 08:51:34.444 [2/start/4 (pid 81134)] Kubernetes error:
2025-11-14 08:51:34.444 [2/start/4 (pid 81134)] Error: Setting up task environment.
2025-11-14 08:51:34.541 [2/start/4 (pid 81134)] Downloading code package...
2025-11-14 08:51:34.542 [2/start/4 (pid 81134)] Code package downloaded.
2025-11-14 08:51:34.542 [2/start/4 (pid 81134)] Task is starting.
2025-11-14 08:51:34.542 [2/start/4 (pid 81134)] Bootstrapping virtual environment...
2025-11-14 08:51:34.542 [2/start/4 (pid 81134)] Bootstrap failed while executing: set -e;
2025-11-14 08:51:34.542 [2/start/4 (pid 81134)] tmpfile=$(mktemp);
2025-11-14 08:51:34.542 [2/start/4 (pid 81134)] echo "@EXPLICIT" > "$tmpfile";
2025-11-14 08:51:34.542 [2/start/4 (pid 81134)] ls -d /metaflow/.pkgs/conda// >> "$tmpfile";
2025-11-14 08:51:34.542 [2/start/4 (pid 81134)] export PATH=$PATH:$(pwd)/micromamba;
2025-11-14 08:51:34.542 [2/start/4 (pid 81134)] export CONDA_PKGS_DIRS=$(pwd)/micromamba/pkgs;
2025-11-14 08:51:34.542 [2/start/4 (pid 81134)] export MAMBA_NO_LOW_SPEED_LIMIT=1;
2025-11-14 08:51:34.542 [2/start/4 (pid 81134)] export MAMBA_USE_INDEX_CACHE=1;
2025-11-14 08:51:34.542 [2/start/4 (pid 81134)] export MAMBA_NO_PROGRESS_BARS=1;
2025-11-14 08:51:34.542 [2/start/4 (pid 81134)] export CONDA_FETCH_THREADS=1;
2025-11-14 08:51:34.542 [2/start/4 (pid 81134)] micromamba create --yes --offline --no-deps                 --safety-checks=disabled --no-extra-safety-checks                 --prefix /metaflow/linux-aarch64/538301e1d2a9acc --file "$tmpfile"                 --no-pyc --no-rc --always-copy;
2025-11-14 08:51:34.542 [2/start/4 (pid 81134)] rm "$tmpfile"
2025-11-14 08:51:34.542 [2/start/4 (pid 81134)] Stdout:
2025-11-14 08:51:34.542 [2/start/4 (pid 81134)] Stderr: ls: cannot access '/metaflow/.pkgs/conda//': No such file or directory
2025-11-14 08:51:34.542 [2/start/4 (pid 81134)] 
2025-11-14 08:51:34.542 [2/start/4 (pid 81134)] (exit code 1). This could be a transient error. Use @retry to retry.
2025-11-14 08:51:34.542 [2/start/4 (pid 81134)] 
2025-11-14 08:51:33.352 [2/start/4 (pid 81134)] [pod t-07ea3803-xkdgx-jth29]             tmpfile=$(mktemp);
2025-11-14 08:51:33.352 [2/start/4 (pid 81134)] [pod t-07ea3803-xkdgx-jth29]             echo "@EXPLICIT" > "$tmpfile";
2025-11-14 08:51:33.352 [2/start/4 (pid 81134)] [pod t-07ea3803-xkdgx-jth29]             ls -d /metaflow/.pkgs/conda/*/* >> "$tmpfile";
2025-11-14 08:51:33.352 [2/start/4 (pid 81134)] [pod t-07ea3803-xkdgx-jth29]             export PATH=$PATH:$(pwd)/micromamba;
2025-11-14 08:51:33.352 [2/start/4 (pid 81134)] [pod t-07ea3803-xkdgx-jth29]             export CONDA_PKGS_DIRS=$(pwd)/micromamba/pkgs;
2025-11-14 08:51:33.352 [2/start/4 (pid 81134)] [pod t-07ea3803-xkdgx-jth29]             export MAMBA_NO_LOW_SPEED_LIMIT=1;
2025-11-14 08:51:33.352 [2/start/4 (pid 81134)] [pod t-07ea3803-xkdgx-jth29]             export MAMBA_USE_INDEX_CACHE=1;
2025-11-14 08:51:33.352 [2/start/4 (pid 81134)] [pod t-07ea3803-xkdgx-jth29]             export MAMBA_NO_PROGRESS_BARS=1;
2025-11-14 08:51:33.352 [2/start/4 (pid 81134)] [pod t-07ea3803-xkdgx-jth29]             export CONDA_FETCH_THREADS=1;
2025-11-14 08:51:33.352 [2/start/4 (pid 81134)] [pod t-07ea3803-xkdgx-jth29]             micromamba create --yes --offline --no-deps                 --safety-checks=disabled --no-extra-safety-checks                 --prefix /metaflow/linux-aarch64/538301e1d2a9acc --file "$tmpfile"                 --no-pyc --no-rc --always-copy;
2025-11-14 08:51:33.352 [2/start/4 (pid 81134)] [pod t-07ea3803-xkdgx-jth29]             rm "$tmpfile"
2025-11-14 08:51:33.352 [2/start/4 (pid 81134)] [pod t-07ea3803-xkdgx-jth29] Stdout:
2025-11-14 08:51:33.352 [2/start/4 (pid 81134)] [pod t-07ea3803-xkdgx-jth29] Stderr: ls: cannot access '/metaflow/.pkgs/conda/*/*': No such file or directory
2025-11-14 08:51:33.352 [2/start/4 (pid 81134)] [pod t-07ea3803-xkdgx-jth29]
2025-11-14 08:51:34.553 [2/start/4 (pid 81134)] Task failed.
2025-11-14 08:51:34.562 Workflow failed.
2025-11-14 08:51:34.562 Terminating 0 active tasks...
2025-11-14 08:51:34.562 Flushing logs...
    Step failure:
    Step start (task-id 4) failed.
    0
    • 1
    • 1
  • q
    Hey team! blob excited The official Metaflow-on-GCP setup references this metadata service image:
    Copy code
    jackieob/metadata_service:gcp.rc1
    link to code Is this image still maintained? I was looking for the source repository of this image and it appears to be this one, but most of the commits are from years ago. We are trying to take advantage of the newly released 
    spin
    feature, but it looks like that requires a newer version of the metadata sevrice.
    0
    t
    • 2
    • 4
  • s
    Hello, we are running into a problem with flows with conditional steps in argo. Details with reproducing flowspec in 🧵.
    0
    t
    • 2
    • 6
  • a
    Hey team, We're using Metaflow and Argo Workflows v3.6.7 We noticed that Argo Retry command doesn't work properly for some workflows, either using Argo Workflows UI (retry button) and CLI (
    argo retry
    ). When I retry the workflow, Argo changes workflow status to Running, delete all failed pods, but it gets stuck without rerunning any step. This is happening specially with workflows that use foreach instruction. Is this the expected current behavior? Should I instead rely on Metaflow retry config (
    --with retry
    ) only?
    0
    a
    • 2
    • 6
  • h
    Hello everyone, We are running metaflow on AWS and have been doing good. Recently have been experimenting with PyG (Graph Based model) and when deploying on AWS it give me this warning. 
    NOTE: The SHMEM allocation limit is set to the default of 64MB. This may be
    insufficient for PyG. NVIDIA recommends the use of the following flags:
    docker run --gpus all --ipc=host --ulimit memlock=-1 --ulimit stack=67108864 ...
    and without this flag it is throwing bus error basically unable to allocate a large graph memory. How do i set 
    --ipc=host --ulimit memlock=-1 --ulimit stack=67108864
    in our batch jobs?
    ➕ 1
    0
    a
    f
    • 3
    • 9
  • m
    Hi, question regarding 
    @retry
    on the foreach step and with 
    --max-workers 1
    , metaflow will process 1 child step at a time. if one of the child steps fails, if I run the flow locally, it would retry on the failed child step with the interval, but if I run the flow with argo workflow, it would skip the retry and run other child steps, and then come back to retry on the failed step. Is this expected behavior with argo workflow?
    0
    a
    • 2
    • 3
  • c
    Hi I got a question regarding the 
    @timeout
    decorator combining with 
    @catch
    . According to the documentation:
    Copy code
    It will cause the step to be retried if needed and the exception will be caught by the @catch decorator, if present.
    But that is not behaviour I'm observing. When the timeout is reached the whole process is aborted and the pipeline fails.
    👀 1
    0
    a
    t
    • 3
    • 6
  • s
    Hello, we are running metaflows in argo and have been doing so successfully for quite a while. We upgraded to metaflow 
    2.19.5
    , mainly for the conditional step transitions, but now run into an issue that makes absolutely no sense to us. Details in 🧵, thanks in advance.
    ✅ 1
    0
    a
    t
    • 3
    • 9
  • f
    Hi, while using baseflow pattern with flow mutators, I get some error. Details in 🧵
    ✅ 1
    0
    a
    d
    • 3
    • 5
  • a
    Hi everyone, I have a couple questions regarding dependency management in the case of a monorepo. What we have: • A monorepo with different projects that rely on a common set of dependencies defined in a pyproject.toml at the root • A CI/CD pipeline that builds a custom Docker image, installing all common dependencies using poetry, everytime a PR is approved • When a flow is deployed, using AWS Step function, it will uses the custom Docker image where the (global) virtual environment has already been created From my understanding (please correct me if I'm wrong), Metaflow will install custom sets of dependencies at run time, for a given flow, only when using the 
    @conda
    and 
    @pypi
    decorators. In our case, every flow that is run on AWS Batch, whether it's deployed or not, will use the custom Docker image, which creates two main pain points: • When adding new dependencies, the inability to run a flow with batch without first having updated the pyproject.toml on the main branch (so that the CI/CD triggers itself and the Docker image is re built, including freshly added packages) • The obvious bottlenecks it will create as more projects are created, depending on larger and probably most likely conflicting dependencies What I would like is/my questions are: • The ability to specify dependencies at the project level ◦ From what I've seen, I feel like the recommended way is essentially to use the 
    @conda
    or 
    @pypi
    decorators, is that correct? ◦ Is there an official alternative if for some reason someone would like to stick with poetry or uv? • If it's not yet possible, I was wondering what is everyone's opinion on eventually having an official decorator, let's say 
    @uv(pyproject="pyproject.toml")
    or 
    @uv(lock="uv.lock")
    , that would allow to specify the dependencies at the flow level while allowing the creation of the venv at runtime. ◦ Similar to, from the doc about netflix's metaflow extensions, 
    A more full-fledged environment command allowing you to resolve environments using external requirements.txt or environment.yml files as well as inspect and rehydrate environments used in any previously run step.
    ? In any case, happy to hear your thoughts and recommendations on that topic, thanks a lot in advance! PS: the spin feature looks amazing 🤩
    👀 1
    ✅ 1
    0
    d
    • 2
    • 8
  • w
    Has anyone setup a metaflow pipeline to read off an SQS queue and aws batch? The benefit I would see to doing this is that we can keep a pipeline alive if there are still things to process on the queue instead of having to spin up an entire another aws batch job.
    ✅ 1
    0
    f
    v
    • 3
    • 12
  • b
    Hi We're facing a persistent scheduling issue for Metaflow flows deployed with 
    @schedule
    to our modern Argo Workflows cluster. The core problem is a schema incompatibility in the generated Argo CronWorkflow manifest. Key Issue: Singular vs. Plural Fields 1. Metaflow Output: Metaflow is generating the CronWorkflow using the deprecated singular field: 
    spec.schedule: '*/15 * * * *'
    (a string). 2. Argo Controller Requirement: Our Argo Controller (v3.6+) *r*equires the current plural field: 
    spec.schedule*s*: ['*/15 * * * *']
    (a list). 3. The Failure: As a result, the Argo Controller sees an empty list (
    schedules: []
    ) and throws the error: 
    "cron workflow must have at least one schedule"
    .
    ✅ 1
    0
    a
    t
    • 3
    • 16
  • b
    Are their recommendations for setting up a high availability deployment of metaflow (metadata service + ui backend) on k8s? Are they both horizontally scalable? Any recommended resource requests/limits? Apologies if this is already documented somewhere. I was having a tough time finding it
    ✅ 2
    0
    s
    • 2
    • 5
  • e
    Hi, I am currently trying to debug, why our Metaflow steps sometimes are slow during "bootstrap of virtual environment". Is there any debugging variable I can set to get more logging out of the bootstrapping process? I am trying to locate any potential bottleneck in network, or k8s cluster resources. In the attached screenshot the bootstrapping takes minutes, but other times it is closer to 20 seconds We are using the --environment=pypi flag and the pypi decorator in the example
    0
    f
    a
    d
    • 4
    • 12
  • f
    I'm running into a strange issue and am a bit stumped. doesn't seem like 
    host_volumes
    is working properly. i've got an AWS setup, using fargate. the recent change is that i'm using a custom container. the batch job is able to launch the ec2 instance, and the task starts, but for some reason the mount isn't passed through to the container properly. more details and things in thread.
    ✅ 1
    0
    f
    • 2
    • 19
  • c
    When running Metaflow on ECS Fargate, is there a client hook to add AWS Tags to the Metaflow-generated AWS Batch JobDefinition? Goal is to propogate tags to the generated Fargate workers (I see propogateTags is set to true for the JobDefinition already)
    ✅ 1
    0
    f
    • 2
    • 3
  • q
    Hey team! is the 
    metaflow_metadata_service
    Docker image the same in ECR (under the Outerbounds org) and in DockerHub (under the netflix-oss org) or are there differences? https://gallery.ecr.aws/outerbounds/metaflow_metadata_service https://hub.docker.com/r/netflixoss/metaflow_metadata_service/tags
    ✅ 1
    0
    s
    t
    • 3
    • 3
  • h
    Hi, is there a way to resume/reun a flow after a fix with argo-workflows?
    ✅ 1
    0
    s
    • 2
    • 2
  • g
    Hey Folks! Nice to meet you all. I am relatively new to working with Metaflow and was hoping to pick the collective brain. Currently running Metaflow with Argo Workflows on GCP GKE. We are seeing increasingly more occurrences of pods getting preempted as part of Autoscaler pod consolidation events. More often than not these are getting preempted before populating into Metaflow and so I thought to implement Argo Workflow retries to assist with but I am being told by the team, this is possibly not a great solution. I supposed closest issue I found was here. Have any of you ran into this on GKE and any suggestions for a more resilient deployment?
    ✅ 1
    0
    a
    b
    f
    • 4
    • 6
  • a
    Hello! I'm facing a problem installing dalex in Metaflow 2.12.5 using @pypi_base. I wrote:
    Copy code
    @pypi_base(
packages={
"dalex": "1.7.2",
...
},
python="3.12.6",
)
    But when Metaflow bootstraps the virtual environment, I get: 
    ERROR: Could not find a version that satisfies the requirement dalex==1.7.2
    (from versions: 0.1.0, 0.1.2, ..., 0.2.0)
    It seems that Micromamba / Metaflow is only seeing the old dalex versions (0.1.x → 0.2.0), even though dalex 1.7.2 exists on PyPI and installs fine outside Metaflow. For example, if I run locally: 
    pip install dalex==1.7.2
    It works perfectly, but inside Metaflow’s Micromamba environment it fails. It looks like Metaflow 2.12.5 with Python 3.12 + Micromamba cannot find the dalex 1.7.x wheel!!
    ✅ 1
    0
    d
    f
    • 3
    • 6
  • p
    Hello ! I'm facing a problem with AWS Batch running GPU. When I use 
    @batch(memory=512, cpu=1, gpu=1)
    the job associated remains in 
    RUNNABLE
    state indefinitely : • I have double-checked that compute ressources are sufficient • I have an ECS Cluster with a g4dn.xlarge instance attached to it and launched by the Auto Scalling Group that Metaflow or Batch launched. • I use the AMI 
    ami-02124cf261ef1e336
    so I have cuda installed My CloudTrails shows a RunTask event that fails with response element :
    Copy code
    "responseElements": {
        "tasks": [],
        "failures": [
            {
                "arn": "arn:aws:ecs:eu-west-3:*******:container-instance/62da090445a44320811473cd2c0e4055",
                "reason": "RESOURCE:GPU"
            }
    It can't understand why, it's like my Batch couldn't see the GPU ressource attached to my EC2 and there is no log to help... When I launch the same step without 
    gpu=1
    it works perfectly well.
    ✅ 1
    0
    c
    f
    • 3
    • 5
  • a
    Hi team, i am facing a memory issue with metaflow-argoWF integration. When users submit jobs directly to K8s, there is no explicit ephemeral-storage request, so k8s uses the full allocatble storage. With argoWF, pods explicitly request ephemeral-storage, limited to 10GB and k8s enforces that. So even if the job gets scheduled to a good-enough instance, there is this ephemeral storage limitation...Anyone faced this before? any insight on best way to solve this would be appreciated! My plan is to change workflow defaults in argoWF or inejct in workflow template but not sure it will work
    ✅ 1
    0
    a
    • 2
    • 1
  • q
    Hello! It' appears that something in the 
    google-api-core
    release yesterday is not compatible with Metaflow since we have started seeing massive failure of all our pipelines this morning with this message:
    Copy code
    [2025-10-28, 07:01:50 UTC] {pod_manager.py:471} INFO - [base]   File "/usr/local/bin/download-gcp-object", line 5, in <module>
[2025-10-28, 07:01:50 UTC] {pod_manager.py:471} INFO - [base]     from simple_gcp_object_downloader.download_gcp_object import main
[2025-10-28, 07:01:50 UTC] {pod_manager.py:471} INFO - [base]   File "/usr/local/lib/python3.12/site-packages/simple_gcp_object_downloader/download_gcp_object.py", line 2, in <module>
[2025-10-28, 07:01:50 UTC] {pod_manager.py:471} INFO - [base]     from google.cloud import storage
[2025-10-28, 07:01:50 UTC] {pod_manager.py:471} INFO - [base]   File "/usr/local/lib/python3.12/site-packages/google/cloud/storage/__init__.py", line 35, in <module>
[2025-10-28, 07:01:50 UTC] {pod_manager.py:471} INFO - [base]     from google.cloud.storage.batch import Batch
[2025-10-28, 07:01:50 UTC] {pod_manager.py:471} INFO - [base]   File "/usr/local/lib/python3.12/site-packages/google/cloud/storage/batch.py", line 43, in <module>
[2025-10-28, 07:01:50 UTC] {pod_manager.py:471} INFO - [base]     from google.cloud import exceptions
[2025-10-28, 07:01:50 UTC] {pod_manager.py:471} INFO - [base]   File "/usr/local/lib/python3.12/site-packages/google/cloud/exceptions/__init__.py", line 24, in <module>
[2025-10-28, 07:01:50 UTC] {pod_manager.py:471} INFO - [base]     from google.api_core import exceptions
[2025-10-28, 07:01:50 UTC] {pod_manager.py:471} INFO - [base]   File "/usr/local/lib/python3.12/site-packages/google/api_core/__init__.py", line 20, in <module>
[2025-10-28, 07:01:50 UTC] {pod_manager.py:471} INFO - [base]     from google.api_core import _python_package_support
[2025-10-28, 07:01:50 UTC] {pod_manager.py:471} INFO - [base]   File "/usr/local/lib/python3.12/site-packages/google/api_core/_python_package_support.py", line 28, in <module>
[2025-10-28, 07:01:50 UTC] {pod_manager.py:471} INFO - [base]     from packaging.version import parse as parse_version
[2025-10-28, 07:02:00 UTC] {pod_manager.py:471} INFO - [base] ModuleNotFoundError: No module named 'packaging'
    The failure is coming from this line. I've looked through the metaflow codebase and it looks like none of the GCP dependencies here are pinned to specific versions so they are probably pulling in the latest 
    google-api-core
    version
    ✅ 2
    this 1
    0
    b
    d
    +2
    • 5
    • 9
  • f
    Hello. I have a quick question about the @trigger_on_finish decorator. Can I use this to essentially re-trigger the same flow and basically keep it running continuously instead of using a scheduled approach? I was going to attempt it but just thought I'd check with the group if anyone knows if this would work.
    ✅ 1
    0
    a
    • 2
    • 6
  • c
    Hello! Brand new Metaflow user here with potentially boneheaded question. I'm hoping to use Metaflow on a long-running petabyte-scale data transformation job with potentially up to a couple million steps/tasks depending on how I chunk my manifest. (I've used similar approaches with Step Functions DistributedMap & ECS Fargate before, so feel like Metaflow will be able to handle this, but please correct me). The DAG is a simple fanout to workers, with no additional joins / reductions necessary. My question is, since I don't have any need to interrogate step artifacts after foreach/map branches complete, how should I handle my 
    join
    ? Is it absolutely required, or is there some type of no-op I can specify to avoid loading a million artifacts from the datastore if I don't need to? Am I missing something / thinking about this incorrectly?
    ✅ 1
    0
    a
    d
    • 3
    • 8
  • c
    Hi, Did anyone else have issues with cards timing out in the metaflow UI? 
    Timeout: loading cards
    Most of the times it works after refreshing the page a few times but not always. Running on AWS ECS
    0
    a
    r
    • 3
    • 4
  • l
    Anyone want to move to Utah and use metaflow with me? Job 😆
    ⛷️ 2
    🏔️ 2
    excited 1
    ⛪ 1
    🧂 1
    0
  • a
    Is it possible to add typing support, or use pydantic models for parameter definition? For example, the fact that the parameter is an int and not a string as desired does not get picked up by mypy.
    Copy code
    def building_block_echo(input: str) -> None:
    print(f"echo... {(input)}.")

class MyEchoPipeline(FlowSpec):
    test = Parameter(
        "test",
        help="A test parameter to show mypy typing works",
        default=1,
        type=int)

    @step
    def start(self):
        print(type(self.test))
        building_block_echo(self.test)

        self.next(self.end)

    @step
    def end(self):
        pass

if __name__ == "__main__":
    MyEchoPipeline()
    0
    h
    • 2
    • 2
  • h
    announcement speaker Metaflow Office Hours 📆 When: Tuesday Oct 28th, 9am PST 📍 RSVP: https://luma.com/6q3m64uf Title: Metaflow for Platform Engineers and Agent Builders: Mutators, Custom Decorators, and Beyond! Speaker: Romain Cledat (Netflix) When Metaflow was first released, it was built for data scientists and ML researchers. As adoption grew, it evolved into a framework that can be extended and customized to meet each company’s needs. In this session, Romain will show how platform teams can use mutators, user-defined decorators, and other advanced features to make Metaflow work seamlessly for their own users—unlocking new power for platform engineers and MLEs.
    myaowlcat flexq 3
    netflix 5
    thankyou 2
    0
    g
    • 2
    • 1
  • f
    Hello. I have a question about running flows. My flows have need to include some configuration files and other important artifacts so I use 
    --package-suffixes
    and I'm also using a uv environment via 
    --environment=uv
    . Since these arguments come before the metaflow commands (run, show, etc.), it doesn't appear that I can set these in a configuration some way. Is there some way to set these via a configuration or even hard code them into the flow script? I tried argument injection in 
    __main__
    but it didn't work.
    0
    a
    • 2
    • 7