@Marvin I'm using Prefect 3 and trying to deploy my first flow on an ECS fargate push work pool created using the prefect cli. I'm baking the flow and task code directly into a docker image stored on a private ECR registry. Below is my prefect.yaml file where the image name and tag are properly populated (it is stored in ecr).

name: test
prefect-version: 3.1.6

deployments:
- name: test_flow_x_hourly_deployment
  version: "1"
  tags: null
  description: This deployment runs flow x hourly
  schedules:
    - cron: "0 0 * * *"
      timezone: "Europe/Paris"
      active: "true"
  flow_name: flow_1
  entrypoint: src/projects/test/prefect/flows/flow_1.py:sample_flow
  parameters:
    input_data: "Hello, World!"
  work_pool:
    name: ecs-fargate-workpool
    work_queue_name: default
    job_variables:
      image: '{{ $IMAGE_NAME }}:{{ $IMAGE_TAG }}'

Below is my dockerfile from which the image is built

FROM prefecthq/prefect:3.1.6-python3.11

ARG PROJECT_PATH

WORKDIR /app

RUN pip install uv

COPY uv.lock .
COPY pyproject.toml .

RUN uv sync

COPY src/common src/common
COPY ${PROJECT_PATH} ${PROJECT_PATH}

ENV PATH=".venv/bin:$PATH"

ENTRYPOINT []

My flow crashes with the error

Flow run infrastructure exited with non-zero status code:
 Exited with non 0 code. (Error Code: 1)
This may be caused by attempting to run an image with a misspecified platform or architecture.

What could I be missing?

I appear to be up and running, but noticed the flow run does not appear on the Prefect web console. I am authenticated and was under the impression that even if I execute a flow locally, the logs/results of the flow would go to prefect. Is that not the case with Controlflow? I apologize if this is a basic question, but still trying to wrap my head around Prefect + Controlflow.

@Marvin I have a docker compose setup with a prefect server and a prefect worker, all in a remote machine. I've setup PREFECT_API_URL on the worker to point to http://server:4200/api, and set the PREFECT_SERVER_API_HOST to 0.0.0.0. I'm able to connect to the Prefect UI using docker port-forwarding, but not without it. I get "Can't connect to PREFECT_API_URL http://localhost:4200/api"

Does prefect have a good way to scale horizontally in Prefect Server?

@Marvin I would like to install dependencies for my flows in the pull step of my

prefect.yaml

. I have tried to do the following but I am still getting an error saying that modules are missing. How can I install the dependencies of the repo that I just cloned?

- prefect.deployments.steps.git_clone:
          <<: *clone_repo
      - prefect.deployments.steps.run_shell_script:
          id: install-dependencies
          script: pip install poetry && poetry config virtualenvs.create false && poetry install --no-interaction --no-ansi
          stream_output: false

Hello Prefect, I'm getting an error

TypeError: cannot pickle '_thread.lock' object

(I'll paste full error code in the comments) and I'm out of ideas how to properly solve it. It is related with

DaskExecutor

and it appears when trying to proceed this task(I'll upload code sample in the comments): 1. Read file content from a yml file which is located in the S3 storage. For file reading I'm using raw

boto3

implementation. 2. Read bytes from the downloded yml file 3. Load yml file and convert it into list Does anyone knows a solution to this problem?

I’d like to trigger some flows whenever a new file is added to a Google Drive folder. My current plan is to run a small python server that receives webhook calls from the Drive API, then uses the Prefect API to trigger the flows. Questions: 1. Is the right Prefect API to use the GraphQL API mutations API

createFlowRun

? 2. Is there anything like this around already? Would love to not reinvent the wheel and deploy a new service.

We are seeing increased latency when starting multiple prefect jobs when submitting on the sub minute level (or running every minute). In Prefect 1 we never experienced this issue, but now are seeing more instances of conflicting runs starting anywhere from 1 to 3 minutes late. Has anyone had success in improving this? We have unlimited concurrency on our work pools

Hey I was using a Prefect2's sdk func

build_from_flow

to deploy a flow by passing

entrypoint

and `path`:

build_deployment = await Deployment.build_from_flow(
        flow=hello_world_flow,
        name=name, 
        ...
        entrypoint=entrypoint,
        path=path)

Im starting now on 3.0 but now build_from_flow is missing:

prefect.exceptions.PrefectImportError: `prefect.deployments:Deployment` has been removed. Use `flow.serve()`, `flow.deploy()`, or `prefect deploy` instead.

Docs mention this args but it is unclear with no examples. Does someone know how to pass entrypoint & path when creating deployment from a flow?

Does anybody have flows that act as infinite loops running the same tasks again and again instead of scheduling a flow to run every few seconds? Just wondering if there’s any downside to doing that vs. scheduling it.