@Marvin in prefect.yaml, how can I specify the code location path that is asked for when I run prefect deploy? I mean the answer to this question: ? What is the path to your flow code in your Dockerfile?

To specify the path to your flow code in your

prefect.yaml

, you can add the

build

section with the

context

parameter. Here’s an example configuration:

yaml
# Generic metadata about this project
name: my-prefect-project
prefect-version: 2.x.x

# build section allows you to manage and build docker images
build:
  - prefect_docker.deployments.steps.build_docker_image:
      id: build-image
      requires: prefect-docker
      image_name: my_registry/my_flow_image
      tag: latest
      context: .

# pull section for setting up remote execution
pull:
  - prefect.deployments.steps.git_clone:
      repository: git@github.com:my-org/my-repo.git
      branch: main
      access_token: "{{ prefect.blocks.secret.github-token }}"
  - prefect.deployments.steps.set_working_directory:
      directory: /opt/prefect

# deployments section for deploying flows
deployments:
  - name: my_deployment
    version: latest
    entrypoint: path/to/your/flow.py:flow_function_name
    parameters: {}
    work_pool:
      name: my-work-pool-name
      job_variables: {}

In this configuration,

context: .

specifies the current directory as the context for the Docker build, meaning it includes all necessary files from the current directory into the Docker image. For more detailed guidance, refer to the Prefect Deployment Guide and the Docker Deployment Guide. If you encounter issues with deployments or other errors, you can browse through the relevant GitHub issues. Here are a few relevant issues: 1. ModuleNotFoundError after installing a local package 2. Failed to parse @flow argument 3. Cloning Error during pull step

@Marvin no, I mean the path inside the docker image, by default this is /opt/prefect/[directory-name]

To specify the path to your flow code inside the Docker image in your

prefect.yaml

file, you can use the

working_directory

option in the

pull

step. By default, it is set to

/opt/prefect/your-directory

. Here’s an example configuration:

yaml
# Generic metadata about this project
name: my-prefect-project
prefect-version: 2.x.x

# build section allows you to manage and build Docker images
build:
  - prefect_docker.deployments.steps.build_docker_image:
      id: build-image
      requires: prefect-docker
      image_name: my_registry/my_flow_image
      tag: latest
      context: .

# pull section for setting up remote execution
pull:
  - prefect.deployments.steps.git_clone:
      repository: git@github.com:my-org/my-repo.git
      branch: main
      access_token: "{{ prefect.blocks.secret.github-token }}"
  - prefect.deployments.steps.set_working_directory:
      directory: /opt/prefect/my-directory

# deployments section for deploying flows
deployments:
  - name: my_deployment
    version: latest
    entrypoint: /opt/prefect/my-directory/path/to/your/flow.py:flow_function_name
    parameters: {}
    work_pool:
      name: my-work-pool-name
      job_variables: {}

In this example,

directory: /opt/prefect/my-directory

specifies the path inside the Docker image where your flow code will be placed. For more information, you can refer to the Prefect Deployment Guide.

@Marvin I am also asked "Does your Dockerfile have a line that copies the current working directory into your image?", how to specify that?

To ensure that your Dockerfile copies the current working directory into your image, you can specify this in the

dockerfile

section of your

prefect.yaml

file. Here is an example configuration:

yaml
# Generic metadata about this project
name: my-prefect-project
prefect-version: 2.x.x

# build section allows you to manage and build Docker images
build:
  - prefect_docker.deployments.steps.build_docker_image:
      id: build-image
      requires: prefect-docker
      image_name: my_registry/my_flow_image
      tag: latest
      context: .
      dockerfile: |
        FROM python:3.9-slim
        WORKDIR /opt/prefect
        COPY . /opt/prefect
        RUN pip install -r requirements.txt

# pull section for setting up remote execution
pull:
  - prefect.deployments.steps.git_clone:
      repository: git@github.com:my-org/my-repo.git
      branch: main
      access_token: "{{ prefect.blocks.secret.github-token }}"
  - prefect.deployments.steps.set_working_directory:
      directory: /opt/prefect/my-directory

# deployments section for deploying flows
deployments:
  - name: my_deployment
    version: latest
    entrypoint: /opt/prefect/my-directory/path/to/your/flow.py:flow_function_name
    parameters: {}
    work_pool:
      name: my-work-pool-name
      job_variables: {}

In this configuration, the

dockerfile

section within the

build

step specifies the instructions for building the Docker image. The

COPY . /opt/prefect

line ensures that the current working directory (where the

Dockerfile

is located) is copied into the

/opt/prefect

directory inside the Docker image. For more detailed guidance, refer to the Prefect Docker Guide.

@Marvin I have a build step like this

build:
- prefect_docker.deployments.steps.build_docker_image:
    id: build_image
    requires: prefect-docker>=0.3.0
    image_name: registry/xxx
    tag: latest
    dockerfile: docker/prefect.Dockerfile

the image is built correctly, but when I run the flow it starts the

prefecthq/prefect:2.19.7-python3.12

image, why? how to run it with the image I just built?

The reason your flow is running with the

prefecthq/prefect:2.19.7-python3.12

image instead of your custom image could be due to the worker pool configuration. Make sure that the image specified in the worker pool matches the image you just built. Here’s how you can ensure that your flow uses the custom image you built: 1. Build Step Configuration: Your existing

build

step looks correct, so there’s nothing to modify here.

yaml
build:
  - prefect_docker.deployments.steps.build_docker_image:
      id: build_image
      requires: prefect-docker>=0.3.0
      image_name: registry/xxx
      tag: latest
      dockerfile: docker/prefect.Dockerfile

2. Specify Custom Image in Deployment: Ensure that the deployment configuration specifies the custom image you built.

yaml
# deployments section for deploying flows
deployments:
  - name: my_deployment
    version: latest
    entrypoint: /opt/prefect/my-directory/path/to/your/flow.py:flow_function_name
    parameters: {}
    work_pool:
      name: my-work-pool-name
      job_variables: 
        image: registry/xxx:latest

In this configuration: - Replace

registry/xxx:latest

with the actual name and tag of your built image. - Ensure that the

job_variables

under the

work_pool

section contains the

image

key with the path to your custom Docker image. For more detailed guidance, refer to the Prefect Deployment Guide and the Prefect Docker Guide.