Apologies if this has already been answered but in the old P

Question

Apologies if this has already been answered but in the old Prefect we started up agents with the p flag so that we can point them to local modules that our flows needed to import This functionality seems to have gone away with Orion Is there some analogous way to indicate to the agent the location of local files that need to be imported For reference the short stack trace I m getting is `Flow could not be retrieved from deployment ` `Traceback most recent call last ` `File tmp flow script calculate blob size4cemrmn py line 1 in lt module gt ` `ModuleNotFoundError No module named local module `

Kevin Kho · Answer

Can I see what the DeploymentSpec looks like in this case

Mary Clair Thompson · Answer

```DeploymentSpec flow location= path to flow py name= blob size calc test deployment schedule=CronSchedule cron= 30 03 flow name= calculate blob size parameters= st acct oitdatatransferdev container name prefect orion test storage config path path to configs yml tags= test ```

Kevin Kho · Answer

What does this end up running on Subprocess Docker or Kubernetes

Mary Clair Thompson · Answer

subprocess

Kevin Kho · Answer

I would say 1 The agent might be running in a directory without access to those files so you can try running it from the same folder so the subprocess will have access 2 You can add the folder to the PYTHONPATH so it can be imported from other places 3 Similar to two you can make it a module and `pip install` the module

Mary Clair Thompson · Answer

I was hoping to avoid packaging the module for 1 how do i determine where the agent is running

Kevin Kho · Answer

Just the existing directory when you did `prefect agent start work queue id`

Mary Clair Thompson · Answer

ah gotcha thank you Kevin

Kevin Kho · Answer

Maybe it will work Not 100 sure For Prefect 1 that is the case

Mary Clair Thompson · Answer

I ll keep you posted about to test it

Anna Geller · Answer

if you really want to make it easy you can create a conda environment and install your custom module as an editable package inside ```pip install e ```

Mary Clair Thompson · Answer

good call thanks Anna

Mary Clair Thompson · Answer

< Kevin Kho> starting the agent from the directory containing the imported file worked thank you