I'm converting a legacy Serverless Framework app written in Python to SST. Things are going fairly smoothly, but my deployed lambdas are large (30+MB) due to bundled dependencies. I've successfully created lambda layers for some of the larger dependencies (psycopg2), but don't understand how to prevent them from being bundled with the deployed lambda. Any tips for how this can be achieved?

Hey @Seth Geoghegan, this is the docker image that bundles the lambda functions. And you can override the install commands by configuring the

bundle.installCommands

flag here.

Oh, interesting. I hadn't seen this option

Yeah, lemme know if you have better ways to do this.

I'll see how far I get with this. I'm pretty new to the python ecosystem, but I have teammates that may have some ideas

yup, how SST bundles Python is similar to CDK’s aws-lambda-python https://docs.aws.amazon.com/cdk/api/v1/docs/aws-lambda-python-readme.html

huh, I didn't know there was such a construct!