This message was deleted.

I'm not aware of a best practice. If I was doing this, I'd probably consider wrapping the build and pulumi app in a sinlgle script, rather than putting the build into the deploy. I'd be worried about sharing node_modules accidentally, using wrong versions of dependencies, etc.

I am using pulumi only to deploy infrastructure (s3 in this case) and I use circleci for building the client and pushing it to the s3 bucket

Not sure if running it as part of the pulumi deploy would be a good idea, but you should be able to do it using Dynamic Resource Providers. There is an example that may be useful to adapt - https://github.com/pulumi/examples/tree/master/aws-ts-ec2-provisioners. Likely to much more work that just doing it as a build step or wrapper script though...

@wide-jackal-86020 the default way, when using Pulumi, is that Pulumi invokes

node

to create the needed resources. If you want to drive Pulumi from within your NodeJS code, have a look a the Automation API. In that setup, the Pulumi engine will be started by your NodeJS code and you can programmatically set the environment, stack config, etc. https://www.pulumi.com/docs/guides/automation-api/

Thank you all!

wrapping the build and pulumi app in a sinlgle script,

This is what we are doing now. The problem is that the API endpoints are generated in other pulumi projects, they are not accessible when building the REACT application. @little-cartoon-10569 @miniature-king-36473

use circleci for building the client and pushing it to the s3 bucket

CI/CD works, But we don't have CI/CD setup for every stack. @sparse-spring-91820

Pulumi invokes 

node

 to create the needed resources.

This is what I am thinking to give it a try. Thank you for sharing the other way of driving Pulumi from within NodeJS code using the Automation API. 👍 @limited-rainbow-51650