mootoday / sapper-on-cloud-run Goto Github PK

Mike, your blog article was really helpful! My sapper app connects to a database and uses some other secrets. I understand that I should use Google Cloud Secrets Manager for most of the secrets, however, for some things I still want environment variables. (I figure, for now using envs is simpler since I'd need to set up IAM and all that for my app to use secrets manager).

My rollup.config.js does something like this:

const processEnv = {
  "process.browser": true,
  "process.env.NODE_ENV": JSON.stringify(mode),
  "process.env.APP_SECRET": JSON.stringify(process.env.APP_SECRET),
  // ... more
};

export default {
  client: {
    input: config.client.input().replace(/\.js$/, ".ts"),
    output: { ...config.client.output(), sourcemap },
    plugins: [
      replace(processEnv), // <---- this

So, these variables would need to exist during the Cloud Build phase, then they get hardcoded into the code which is sent over to Cloud Run. So, using the environment vars in Cloud Run wouldn't work, unless I didn't use the replace() plugin for production builds.. so I suppose that's an option.

Either way, I'm curious how you handle these cases or if you already have some code you could add into this boilerplate? Just like in your guide, my project simply uses a Dockerfile, not a cloudbuild yaml.

Another question too, do you normally run DB migrations from your Dockerfile during Cloud Build as well?