I am building a serverless function using the serverless framework. However im having an issue with running it locally prisma/schema.prisma

generator client {
    provider      = "prisma-client-js"
    binaryTargets = ["native", "rhel-openssl-1.0.x"]
}

serverless.ts

package: {
        individually: true,
        patterns: [
            "!node_modules/.prisma/client/libquery_engine-*",
            "node_modules/.prisma/client/libquery_engine-rhel-*",
            "!node_modules/prisma/libquery_engine-*",
            "!node_modules/@prisma/engines/**",
        ],
    },

npx prisma generate && npm install

sls invoke local -f main

✖️ Error: ENOENT: no such file or directory, open ''/.esbuild/.build/node_modules/.prisma/client/schema.prisma'

What am i doing wrong here? attempting to follow this example: https://github.com/prisma/prisma-examples/tree/latest/deployment-platforms/aws-lambda • note: i am using the serverless

aws-nodejs-typescirpt

template which uses serverless-esbuild and not serverless-webpack

Hi @Kay Khan I was facing the same situation, avoided it with something like including the files needed for local, and excludend them for deployment, also created a lambda layer for prod.

inject: ['../../prisma/schema.prisma',
'../../node_modules/.prisma/client/libquery_engine-rhel-openssl-1.0.x.so.node',
'../../node_modules/.prisma/client/libquery_engine-debian-openssl-3.0.x.so.node',
 ],
  assetNames: './[name]',
  loader: { '.prisma': 'file', '.node': 'file' },

this is what I added in my custom esbuild config for local, and, of course when deploying, not considering it.

@Marcos Iglesias hmm i dont understand, can i see your entire serverless.ts?

nop, just trial after trial, searched info in esbuild page and in prisma docs also

custom esbuild object?

sorry, like this one

I dont touch that

custom: {
        esbuild: {
            bundle: true,
            minify: false,
            sourcemap: true,
            exclude: ["aws-sdk"],
            target: "node16",
            define: { "require.resolve": undefined },
            platform: "node",
            concurrency: 10,
        },
    },

Try to add my lines, something like this:

custom: {
  esbuild: {
      bundle: true,
      minify: false,
      sourcemap: true,
      exclude: ["aws-sdk"],
      target: "node16",
      define: { "require.resolve": undefined },
      platform: "node",
      concurrency: 10,
      inject: [
        '../../prisma/schema.prisma',
        '../../node_modules/.prisma/client/libquery_engine-rhel-openssl-1.0.x.so.node',
        '../../node_modules/.prisma/client/libquery_engine-debian-openssl-3.0.x.so.node',
      ],
      assetNames: './[name]',
      loader: { '.prisma': 'file', '.node': 'file' },
  },
},

you will need to resolve the correct paths and where the "assets" are going to be injected accoding to your project structure of course

I need it myu error message tells me i need that

Query engine library for current platform "debian-openssl-1.1.x" could not be found.
You incorrectly pinned it to debian-openssl-1.1.x

in the "inject" you have to point the path to the file wich is going to be "copied", in the "assetsName" put it like that, but preceding the folder you need it. like "./node_modules/[name]" and the loader is for handling the files as what they are, we need them to be copied

i guess your solution is doing something similar except injecting it after?

yes, correct

really?

if you include them in the deployment the lamba size will be huge

right they are 80mb total

huge

So are you telling me if i inject in the custom.esbuild they dont show up in deployment?

no i didn't told you that, since we are in a .ts file I added a conditional with 2 esbuild custom configs, one with the inject for local, and the other one "clean", for aws deploy

If i need to include these "engines" in the output build if i want to run

sls invoke local -f main

when this is deployed why does it not require thsese engines?

so many lambdas are using that layer

oh~ thats just sounds like a pain 😕 still not quite understanding why it would not need them when its deployed

sorry I don't know the answer to your last question. If sls invoke uses the aws lambda then you are including them in the final package, right?

right

so that's why they are not need while they are deployed

?

they are included, and that's why your lambda is 80mb, right?

No~ i have not deployed this yet

ahh

You said you have 2 builds, one for local and 1 for production. You are saying your production build does not include these "engines"

oh, yes, you/we need them. I included them in a lambda layer!

right

you can still have them in the deployment build, but the lambda size is huge! like you point

Then im fine with just this, hopefully

package: {
        individually: true,
        patterns: [
            "!node_modules/.prisma/client/libquery_engine-*",
            "node_modules/.prisma/client/schema.prisma",
            "node_modules/.prisma/client/libquery_engine-debian-*",
            "!node_modules/prisma/libquery_engine-*",
            "!node_modules/@prisma/engines/**",
        ],
    },

yeah, maybe

You can't get around the size of it

I already avoided it. And my lambdas weight less than 2MB

really?

But I did the trick after a lot of researching, fail & trial, and finally with lambda layers SHARING these binaries

can you point me to some documentation you researched

thanks i will read these

sure no worries

@Marcos Iglesias your final deployment on aws for this lambda function is how many mb?

less than 2MB

are you sure your other layer is not keeping the size of the binary?

yes, one layer is bigger than the other

how big is that layer?

can't tell rn, apologies, but I think is something like 20MB

okay

most of that 16mb should be from the lib engine file...

yep

and if you include another one... i dont want to imagine lol

i think those layers was wayu to much of a complicated solution, compared to what i needed

https://dev.to/eddeee888/how-to-deploy-prisma-in-aws-lambda-with-serverless-1m76 https://dev.to/sabinthedev/managing-lambda-size-with-multiple-prisma-clients-4m8l?hmsr=joyk.com&utm_source=joyk.com&utm_medium=referral

if you use layers , each layer still need to conform to the size limits of a lambda function

for how many lambdas?

I have a single lambda function, no layers

in that case... because that guys was separatin every model with 1 client...

I ony ever have 1 schema file and 1 prisma client

but the other article took a different approach