Search code examples
pythongoogle-cloud-platformgoogle-cloud-functionsgcloud

GCP cloud functions with "gcloud functions deploy" fails with --source as file rather than directory

I'm trying to structure my GCP gcloud functions (in python) as a separate file per function. It sounds like from the documentation it should be supported but when I try to use --source as a file name instead of a directory it fails. Is there something I'm doing wrong?

Here is the command I use:

gcloud functions deploy createAccount --runtime python38 --trigger-http --source=postgres/createAccount.py --region us-central1

and the error I get back is:

ERROR: (gcloud.functions.deploy) argument --source: Provided path does not point to a directory

But if I put my "createAccount" python function in main.py inside the postgres directory and use this command the function deploys perfectly:

gcloud functions deploy createAccount --runtime python38 --trigger-http --source=postgres --region us-central1

Here it loojs as though it should accept file names in the --source option:

https://cloud.google.com/functions/docs/first-python

See this section:

enter image description here

Any ideas if there is a way to not make main.py one big monolith of all my cloud functions?

Solution

  • If we look at the documentation of the gcloud command for deploying functions and the --source flag within:

    https://cloud.google.com/sdk/gcloud/reference/functions/deploy#--source

    We find that it unambiguously says that it wants a directory as a parameter and not a source file. The link you gave which seems to say that we can specify a file looks to be in error. I think that is a mistake and it can only be a directory that is supplied with --source.

    This would seem to imply that you can create multiple directories ... where each directory contains just the function you wish to deploy.