Bring your towel and join one of the fastest growing data communities. Welcome to our second-generation open source orchestration platform, a completely rethought approach to dataflow automation.

Prefect Community

Hello! I am trying to test a few flows in our prod environment by running a local agent in a docker container that simulates the environment. For some reason the agent is not picking up the registered flows even though they share labels. Does anyone know why this would be happening? For reference, it runs successfully if I run a local agent in my dev environment

Hi <@U01GEBKFTLP> does your flow use the default <https://docs.prefect.io/orchestration/flow_config/storage.html|local storage>? If so then it won’t be getting picked up because flows that are registered with Local storage are given an extra label of the hostname of the machine they were registered on so they don’t accidentally get picked up by agents running elsewhere

I am definitely using local storage so this sounds like the issue

would adding the host-name as a label to my "docker agent" solve the problem?

It would solve the problem in the agent being able to pick up the flow run however since your flow is stored on your local machine the process that the agent creates in your container won’t be able to fine the flow file because it doesn’t have access to it.

This is where the other storage options come into play. They allow you to store your flows in central locations (e.g. S3, docker registry, etc.) that you can give other platforms access to in order to deploy your flow runs

when I try using Docker storage im getting this error
`docker.errors.DockerException: Install pypiwin32 package to enable npipe:// support`

have you seen people having this error on windows when using Docker storage?

I have installed the package it does not solve the issue

Hmm the only time it looks like this has come up is in this PR where we discussed adding Python 3.8 support a while ago (<https://github.com/PrefectHQ/prefect/pull/2080#issuecomment-591017041>).

This other repo someone said that conda installing the package fixes it <https://github.com/twosixlabs/armory/issues/156>