@Marvin how the option

prefect.worker.daemon.reuse_virtual_environments

works ?

✅ thought for 30.7 seconds

@Marvin what is the

reuse_virtual_environments

job variables in a

deploy.yml

file ?

✅ thought for 69.4 seconds

i dont think that's a real thing? unless you wrote your own worker where thats on the job config

Based on the research, I need to be honest that I cannot find detailed official documentation specifically about the

reuse_virtual_environments

job variable. However, I can provide some context about job variables in deployment configurations: 1. Job variables are configuration options that can be specified in your

deploy.yml

file to customize how your flow runs are executed. 2. The general structure in a

deploy.yml

file would look something like this:

yaml
deployments:
  - name: my-deployment
    jobs:
      - variables:
          reuse_virtual_environments: true  # or false

Given the name and context, this setting likely controls whether the worker should reuse existing virtual environments across multiple flow runs or create fresh environments for each run. However, since I cannot find authoritative documentation about this specific setting, I recommend: 1. Testing the behavior with and without this setting to verify its effects 2. Checking with your specific work pool type's documentation, as this might be infrastructure-specific 3. Consider reaching out to the Prefect community on Discord or GitHub discussions for clarification Would you mind sharing: 1. Which work pool type you're using? 2. Where you encountered this setting? 3. What specific behavior you're trying to achieve? This would help me provide more targeted guidance for your use case.

@Nate I don't remember where I found it, but in fact, if I comment this in my prefect.yml, my deployment doesn't work anymore... and I'd like to understand why :

job_variables:
  prefect.worker.daemon.reuse_virtual_environments: true

sorry, are you saying if you comment out that job variable then

prefect deploy

no longer works? can you elaborate on what you mean by not working? ie do you get an error?

prefect deploy

works, but when I execute my deployment, I have import errors like

ModuleNotFoundError("No module named 'pandas'")

do you have your python dependencies installed in the runtime where your flows run? that's a very common failure mode, ie forgetting to install deps at runtime or not including them in your image etc

I think this variables enable to reuse venv created by a previous flow

what type of worker are you using? im asking because the variable you listed does not exist anywhere in prefect https://github.com/search?q=repo%3APrefectHQ%2Fprefect%20reuse_virtual_environments&type=code

I didn't customize my image, but made a lot of tests on how to install dependencies in the pull section

so the only way this is a valid job variable is if you created your own worker type

Hmmm... I'm using the official docker image as a process worker

prefecthq/prefect:3-python3.12

yea if you go into the Work Pools section in the UI and edit it you can see what valid job variables are for a given work pool type, here are the valid ones for process pools