Does Seed apply incremental deploys or have some heuristics for doing so with Serverless Stack? Does Seed do this already by default for SST projects?

hi, has anyone tried using cdk pipelines with sst?

also, is it possible to configure the build output directory? i'd like to write up a nx plugin, traditionally the build output goes into a dist folder in the workspace root

@Frank maybe this changed recently? Looks like code in the Doc has a type error. From

HttpMethod

to

CorsHttpMethod

? https://docs.serverless-stack.com/constructs/Api#configuring-cors

If I define vpc property within a Function, is this meant to reference existing VPC/subnets etc or is this going to create them?

What is the point of cdk-context. I just confused myself for 30min because I was wondering why

ssm.valueFromLookup

was giving an old value and it's because it was cached in cdk-context. Doesn't see to be a way to override this

Anybody know if it's possible/how to determine what stage the deploy is running in during runtime? (as in, it could have picked up the stage from the

sst.json

file, or the

--stage

command line arg)

Got a little issue here when I do my lambdas in python with sst code in Typescript. Given a Route like this:

'POST /karaoke/rotations': 'src/services/rotations/KAR_ROT_create_rotation.lambda_handler'

I get the following error on deploy:

error: Could not resolve "/home/lukewyman/Documents/code/karaoke-backend-sst/src/services/rotations/KAR_ROT_create_rotation.js"

So it's assuming it's looking for a file ending in

.js

when the file actually ends in

.py

.

Hey guys, Is there any examples for setting up sst with yarn workspaces in typescript?

Okay. So,

FunctionProps

has an attribute called

bundle

which can take a

FunctionBundlePythonProps

with this interface:

export interface FunctionBundlePythonProps {
    readonly installCommands?: string[];
}

What are those

installCommands

all about? The AWS Lambda documentation for Python discusses using either a zip file or a Docker image to deploy. How might the

installCommands

play ball with either or both of those?

I’m using API GW v2. construct, but after having a 500 error, my service doesn’t work anymore, and start to get 503 “service unavailable”. Looks like something is blocking my API after having a 500, and from so on starting to get 503. Anybody had the same problem?

I think there's a bug with the output of the new

sst update

command. When it's run the package.json is updated to have

"@aws-cdk/core": "^1.114.0",

and

"@aws-cdk/assert": "^1.114.0",

but this causes a warning on build that the package must be the exact version. I.e. it should be (without ^)

"@aws-cdk/core": "1.114.0",

and

"@aws-cdk/assert": "1.114.0",

I crashed SST’s live debug 😳

main ist: [Function (anonymous)]

TypeError: Cannot read property 'length' of undefined
    at Object.truncate (/Volumes/SuperData/Sites/reelcrafter/v2-microservices/node_modules/@serverless-stack/cli/lib/object.js:7:17)
    at printLambdaResponse (/Volumes/SuperData/Sites/reelcrafter/v2-microservices/node_modules/@serverless-stack/cli/scripts/start.js:1540:58)
    at ChildProcess.<anonymous> (/Volumes/SuperData/Sites/reelcrafter/v2-microservices/node_modules/@serverless-stack/cli/scripts/start.js:1455:5)
    at ChildProcess.emit (events.js:376:20)
    at ChildProcess.emit (domain.js:470:12)
    at Process.ChildProcess._handle.onexit (internal/child_process.js:277:12)

I’m driving myself crazy with some CORS issues, hoping someone has had similar experience. My GET requests are resolving fine, but my POSTs are not. Any troubleshooting tips?

feature request/enhancement: it would be nice if the error logger used

util.inspect()

so we can get more depth. the

errors: [Array]

isn’t helpful here

8526a694-c552-4dd1-be61-616982184b83 ERROR Error: Network error: Response not successful: Received status code 401
    at new ApolloError (/Volumes/SuperData/Sites/reelcrafter/v2-microservices/node_modules/src/errors/ApolloError.ts:56:5)
   [snip]
microservices/node_modules/zen-observable/lib/Observable.js:179:3) {
  graphQLErrors: [],
  networkError: {
    response: { size: 0, timeout: 0 },
    statusCode: 401,
    result: { errors: [Array] }
  }
}

Is lambda authorizer caching working for you guys? For me, it’s caching the responses always. I’m sending different bearer JWT tokens, and is caching the response and returning always the same, even for different tokens.

Hey guys, how have you all set your env vars for the

sst test

command? It doesn’t seem that having a

.env.test

file injects the test environment vars into the test environment but I don’t want to overwrite the jest.config under the hood.

Hey @gio, trying to dig into this a bit

SST needs some speed up

By slow, do you mean the handler function is taking long to run locally in

sst start

?

RAM reach 4gb

Can you see which process is this? Is it

esbuild

?

Does somebody get to configure VS Code to lint with the same config as SST? I want to get these kind of warnings on VS Code, not only on SST lint.

34:21 warning Argument ‘value’ should be typed with a non-any type @typescript-eslint/explicit-module-boundary-types

Other question/concern… Is anybody using “Simple” Lambda Authorizer? It works fine for the most part, but this is strange… On my lambda integration, this is what I receive on the `event`:

...
requestContext: {
    accountId: 'xxx',
    apiId: 'xxx',
    authorizer: { lambda: [Object] },
    domainName: 'xxx',
...

As you can see, there is an object

authorizer

which has a child `lambda`… the object inside

lambda

it’s totally fine, it’s exactly what I expect to receive. The problem is, apparently this

lambda

object doesn’t exists on TS interface

APIGatewayProxyEventV2

. As you can see here, this is the interface definition:

...
    requestContext: {
        accountId: string;
        apiId: string;
        authorizer?: {
            jwt: {
                claims: { [name: string]: string | number | boolean | string[] };
                scopes: string[];
            };
        } | undefined;
        domainName: string;
        domainPrefix: string;
...

Apparently the interface

APIGatewayProxyEventV2

only supports

jwt

, but not

lambda

? anybody else had this issue?

Does anyone get this error every now and then while the live lambda env is running?

There was a problem reading the build metafile [Error: ENOENT: no such file or directory, open '/Users/dev/engine/.build/.esbuild.--src-graphql-server-handler.json'] {
   errno: -2,
   code: 'ENOENT',
   syscall: 'open',
   path: '/Users/dev/engine/.build/.esbuild.--src-graphql-server-handler.json'
}

happens somewhat randomly and I have to restart. Might be something with a jest watcher also running

npm test --watch

Are there any plans to support

add-cdk

and

update

in a project inside a workspace?

Seems like using

$default

can screw up cors since it looks like it's routing options requests to it (which in my case fail because it requires auth)

I just experimented with deploying python lambdas. First did a

npx create-serverless-stack@latest [proj nam] --language python

and it worked as expected, creating a JavaScript

MyStack

and a Python lambda (

lambda.py

). All good. Then I did an

npx sst start

and then checked out what goodies I made in the AWS console. I see that my Lambda got created, it's in JavaScript in the console (no Python runtime Lambdas created). Is sst converting from Python to JavaScript?

Anybody else experienced this issue in Lambda? My lambda is timing out, but checking in the logs, there is nothing there… only couple of warnings, but looks like is taking long time to do the cold-start and it doesn’t log anything.

Hey everyone! I was just wondering about the best practice in the

StaticSite

construct for running

npm install

before compiling the SPA? I just literally do

npm install && npm run build

for the

buildCommand

. Is there a better way? What does everyone else do?

I wonder if there’s any way to optimize the switching between live & debug stacks (start vs deploy). I realize it has to update the functions to inject the stub code.. but if you have 150 functions for your HTTP API, that would take a long time. I don’t know how it works under the hood, but I wonder if there’d be some way to just modify the Lambda’s code bundle and not touch CloudFormation at all, which is where the bottleneck is

In these kind of examples, can you guys add the imports on top? Sometimes is hard to find where those belong to, due most of them are from CDK and there so many sub-packages. In this case, was hard to find that

targets

belongs to

@aws-cdk/aws-events-targets

.

We're getting an error when setting accessLogs to true on the Api construct. If we set

accessLog: true

in the snippet below we get the error:

TypeError: Cannot read property 'api' of undefined
    at Object.buildAccessLogData (/home/jamin/Code/OFSC-forms-PDFify/node_modules/@serverless-stack/resources/src/util/apiGatewayV2AccessLog.ts:79:54)
    at new Api (/home/jamin/Code/OFSC-forms-PDFify/node_modules/@serverless-stack/resources/src/Api.ts:217:45)
    at new PdflifierStack (/home/jamin/Code/OFSC-forms-PDFify/lib/PdflifierStack.ts:97:17)
    at Object.main (/home/jamin/Code/OFSC-forms-PDFify/lib/index.ts:13:3)
    at Object.<anonymous> (/home/jamin/Code/OFSC-forms-PDFify/.build/run.js:89:16)
    at Module._compile (internal/modules/cjs/loader.js:1063:30)
    at Object.Module._extensions..js (internal/modules/cjs/loader.js:1092:10)
    at Module.load (internal/modules/cjs/loader.js:928:32)
    at Function.Module._load (internal/modules/cjs/loader.js:769:14)
    at Function.executeUserEntryPoint [as runMain] (internal/modules/run_main.js:72:12)

There was an error synthesizing your app.

It goes away if we set it to false. Any thoughts on what's going on here?

Is it possible to deploy only one function even though if stack has multiple functions?