Hi, I’m using ApolloApi with nexus and prisma and next.js and some recent additions are causing some issues when deploying via seed. I have 2 stacks - one for the next.js client and one for my server components (ApolloApi and a queue and a consumer lambda) I am getting this error in CloudWatch which is manifesting in the client as a 503 error:

undefined ERROR Error: Root typing path "/var/task/src/context.ts" for the type "context" does not exist

So any attempts to invoke the API routes from my Next.js client results in the above being written into Cloudwatch. Has anyone seen this error before, and would you be able to point me in a direction of solving it ? For reference the schema for Apollo is as follows:

export const schema = makeSchema({
  types,
  outputs: {
    schema: path.join(process.cwd(), 'src', 'generated', 'schema.gen.graphql'),
    typegen: path.join(process.cwd(), 'src', 'generated', 'nexus-types.gen.ts'),
  },
  plugins: [fieldAuthorizePlugin()],
  contextType: {
    module: path.join(process.cwd(), 'src', 'context.ts'),
    export: 'ContextType',
    alias: 'ctx',
  },
});

Hey Mark - will need Frank to chime in as he's more familiar with the nextjs construct.

Hey, its in all modes from local debugging through to seed deployment. In all cases the API is accessible and functional, its just the client api routes domain.com/api/xxx are failing with the above error

Ah gotcha, this sounds similar to what @Yousef was asking about so will let Frank chime in when he has a sec

Thanks @thdxr. Hey @Mark Faulkner, are the Next.js api routes failing only inside

sst start

? It you stopped

sst start

and then do a

sst deploy

, the api routes works right?

Should they work with

sst start

i figured that workflow was unsupported

The entire site shows a placeholder page when using

sst start

and asks for an

sst deploy

to be run

hmm… there hasn’t be any significant changes to the

NextjsSite

since 0.45.2.

So this seems to be all solved now running on the latest 0.49.2 build. Everything was fine with earlier versions when running via localhost but it was with a domain name (and deployed via Seed) that the issues started.

Oh I see. Glad it’s working!