Hi all, I am wondering if this is a NextJS issue or a SST issue. Somehow I am not able to pass my env variables to Amplify configure. but it works if I set the configure directly. The

process.env

is printed alright in my terminal. Do my miss something? Do I have to prefix

NEXT_PUBLIC_

to these variables?

Is there a way to declare the local version of node SST should use when running live dev? After updating node to v17 I now see the error below in the console. I’ve installed node v16 as well and tried adding a

.nvmrc

file at the project root (e.g. with ‘16’) but it doesn’t seem to affect things.

I'm working with the guys at Thundra to evaluate their APM product as a replacement for Epsagon. When wrapping a lambda using their library SST deploy errors out with the following:

> node_modules/jest-pnp-resolver/index.js:46:4: error: Cannot assign to "defaultResolver" because it is a constant
    46 │     defaultResolver = getDefaultResolver();
       ╵     ~~~~~~~~~~~~~~~
   node_modules/jest-pnp-resolver/index.js:13:18: note: "defaultResolver" was declared a constant here
    13 │   const {basedir, defaultResolver, extensions} = options;
       ╵                   ~~~~~~~~~~~~~~~

 > node_modules/fsevents/fsevents.js:13:23: error: No loader is configured for ".node" files: node_modules/fsevents/fsevents.node
    13 │ const Native = require("./fsevents.node");
       ╵                        ~~~~~~~~~~~~~~~~~

 > node_modules/jest-pnp-resolver/getDefaultResolver.js:8:32: error: Could not resolve "jest-resolve/build/default_resolver" (mark it as external to exclude it from the bundle, or surround it with try/catch to handle the failure at run-time)
    8 │       defaultResolver = require(`jest-resolve/build/default_resolver`...
      ╵                                 ^

 > node_modules/jest-config/build/vendor/jsonlint.js:909:22: error: Could not resolve "file" (mark it as external to exclude it from the bundle, or surround it with try/catch to handle the failure at run-time)
    909 │     var cwd = require('file').path(require('file').cwd());
        ╵                       ~~~~~~

Is there anything that I can go back to the team with in terms of things they could change to make it work?

Hey guys… I want to re-use a function, to be called by a Cron and also by API GW… Do I need to define the

permissions

on each integration? or I can jut define a Function and assign the permissions to it, and later attach that function to both Cron and GW?

Question about shared-libraries for micro-services: If you create a shared-library that is used by many different functions…. and in that logic, for example, you need to trigger an Event on EBridge… That means every Lambda consumer of that library, will need to have access to the EBridge, with permissions. And possibly also an environment variable with the EBridge location, due the library doesn’t know which is the AWS Resource. And other stuff could be needed too. And that’s not clear for the consumer… you are “consuming” a library which might be a “black box” ideally. For example… “Users Library” which contains “Update User” which triggers “User Updated” event. Also this creates coupling between those “consumers” of the library and every resource that is needed… Which could be OK, but that can be messy if you have many things shared. This is why I prefer to encapsulate that logic in a different Lambda instead of Library… doing something like “Lambda calling Lambda” (or maybe a REST API) that other functions can consume, that way we encapsulate the logic and permissions of the “shared code”. Thoughts? @thdxr @Frank

In the Notes example, the whole kitchen sink is rolled into a single project. Is there any wisdom to breaking it into 2 separate projects. One for the API and Data and another for the front end React app? I can see where I might ultimately have multiple front end apps all talking to the same API. Also, I have development teams that are very focused on backend and database and others that are all about the front end. Separation of concerns seems wise to me. Thoughts?

Question for the core team: Have you guys considered / Do you have a living + public roadmap for SST?

Not sure if I’m doing something wrong but suddenly our

.env.local

env vars are not being loaded into the process.env in the

sst.Stack

we just moved from

0.48.0

to

0.50.2

Hey guys, curious to hear what people's thoughts were around running an express server connected to the api-gw such that each route would have only one lambda function which handles routing etc? Heard the idea thrown around a bit recently, the main pro I can see for this is that it would reduce cold start times since the function would be getting used more, but that may be negligible if you had enough throughput on your lambdas? I imagine though there are other reasons like being able to use a bunch of express middleware and have access to that ecosystem?

I didn't realized until recently after reading some of the comments here, that you can actually call localhost from within your SST Lambdas on development! I think that's real game changer for dev exp. We have a python api server running on fargate in prod and it's easier on dev to run it locally with docker together with local postgres, but with SST I can call and connect to them from lambdas. Another usecase that I just build was having to test incoming webhooks, but with api not being serverless, instead of using something like ngrok to tunnel remote calls or deploying to check. I used a SST lambda that basically proxy remote call to my local api server and it just works. Just wanted to pointed that out here, incase someone find it useful. Love the community here and amazing work from the SST team.

It appears that using Datadog’s CDK construct to wrap lambdas seems to break SST’s live lambda dev, swallowing the output… so I’m not getting nothing in the console. Perhaps adding monitoring like this should only happen when not

IS_LOCAL

?

I’m getting an error with sst start (but not sst deploy):

invokeError [1636085771837] started /Users/omi/workspaces/libs/lambda-utils/index.ts:1
import LogUtils, { CoveLogger, ChildLogger } from '@fragment/log-utils';
^^^^^^

SyntaxError: Cannot use import statement outside a module

I’ve tried to dig into this a bit, this is the generated import in `.build`:

var import_lambda_utils = __toModule(require("@fragment/lambda-utils"));

So for some reason esbuild is thinking that module doesn’t need to be bundled and transpiled

I'm working with the guys at Thundra to get the travel time debugging (TTD) option working (tracing now works with just the inclusion of a layer! @Adrián Mouly) The way TTD works is you give it an env var that represents the path of the handler so that it can slurp up the code. We're struggling, we think, because of the output structure from esbuild. They are considering an esbuild plugin to set this up (they already have one for webpack) Therefore, for debugging purposes I'm trying to build unbundled but I'm getting the error

Error: Bundle cannot be disabled for the "AuthorizerFunction" function since the "srcPath" is set to the project root. Read more here — <https://github.com/serverless-stack/serverless-stack/issues/78>

Looking at the github issue I'm confused as to how to workaround this. Shifting my

src

down a level doesn't make any difference. Does the code need to live in an entirely different folder structure for this to work?

This is likely a dumb question but here’s to becoming less dumb - I’ve noticed I can import various

aws-cdk

packages in a stack (e.g.

@aws-cdk/aws-events

) and have been able to successfully run and build the stack without having to install these packages first in the project. Is this ok, or should they be listed in the project’s package.json? And if so, is it ok for them to be devDependencies only?

I was setting up a new stack that references an SSM variable that I created manually in the AWS console. Therefore, the stack creation failed when executing

sst start

.

sgeoghegan-kas-api-stack | CREATE_FAILED | AWS::CloudFormation::Stack | sgeoghegan-kas-api-stack 

 ❌  sgeoghegan-kas-api-stack failed: ERROR Parameter Name /sgeoghegan/kas/api-url with a different configuration already exists.


Stack sgeoghegan-kas-db-stack
  Status: deployed

Stack sgeoghegan-kas-api-stack
  Status: failed
  Error: ERROR Parameter Name /sgeoghegan/kas/api-url with a different configuration already exists.

Failed to deploy the app
error Command failed with exit code 1.
info Visit <https://yarnpkg.com/en/docs/cli/run> for documentation about this command.

So I’m playing with

NextjsSite

construct, with an oauth flow which requires an https redirect - so rather than struggle to setup a local cert, I’ve been deploying to test my code - which doesn’t provide the sexiest workflow. Normally, with sst,

sst start

would get me what I want - but with nextjs a boilerplate temp site is deployed, and nextjs is run locally at

localhost:3000

which doesn’t satisfy my oauth partner requirements. I’m considering ngrok to my localhost, so I can edit code locally while I test this oauth flow, but anyone have any other suggestions? Workarounds for the lack of local tunneling for

next dev

sites?

Hi, what’s the SST approach to Provisioned Concurrency?

It looks like SST’s

.addOutputs

method publishes outputs without an

export name

value set – making the value inaccessible to

cdk.Fn.importValue

. Is there are way to add an export name to a SST created output or should you duplicate these via

cdk.CfnOutput

? I’m finding myself creating outputs via

.addOuputs

for consumption by other SST stacks within the same service and by

cdk.CfnOutput

for consumption by other Services’ stacks in the same account – which leads to duplicate outputs. Not a huge issue, but would be nice remove the duplication in the code and console if possible.

I've got an odd problem that the debugger is no longer hitting my breakpoints. I think I remember a Visual Studio Code update yesterday (1.62.0) but other than that, I can't think of what's changed. Anyone know of a way to debug the debugger?

I have a single function definition that can be included in a stack multiple times with different names.

datasetIds.forEach((datasetId) => {
      
      const populatorFunction = new sst.Function(this, `populatorFunction-${datasetId}`, {
        handler: 'src/main/handler/populator.startLookingForMissingUrls',
        environment: {
          OFSC_INSTANCE: datasetId,
        },
      });

      new sst.Cron(this, `OmwPopulatorCron=${datasetId}`, {
        schedule: process.env.POPULATOR_CRON,
        job: {
          function: populatorFunction,
        },
      });
})

So you can see that the OFSC_INSTANCE env var is set according to the value from a string array defined earlier. When running this using

sst start

I'm seeing some odd behaviour regarding the env var OFSC_INSTANCE. On the first kick off of each lambda via the cron, sometimes the value of the env var is correct. However, after that, both lambdas have the same value for that env var. Sometimes they're the same on first start too. Is there something about the local execution environment that would env vars end up getting shared between invocations of shared code?

Another dumb question from me 😄 What implementation do i need to allow a user to signup using cognito. Like for now, I am using shell to sign up a user i.e.,

aws cognito-idp sign-up

... but i have to provide a rest endpoint for user so that they can do signup themselves (and do challenge to confirm signup). I mean there must be a library from

aws-cdk

which provides user signup from backend code?

Is there a way to break up stacks so that changes to one stack do not force a redeploy of the other? For instance DatabaseStack and ApiStack. Then I could redeploy only changes to ApiStack, hopefully speeding development. This seems to be related, but I haven’t sorted it out yet: https://docs.serverless-stack.com/constructs/Api#sharing-an-api-across-stacks

Currently getting a 503 on all of my NextJsSite API endpoints. Has anyone else had this issue?

Is it possible for an sst stack to connect to a local database instead of one running and created in aws ? Our concern is that if we have multiple engineers working locally, sst will create multiple databases which doesn’t seem right. https://serverless-stack.com/examples/how-to-use-postgresql-in-your-serverless-app.html

I keep getting these logs on the new sst version (v0.50.2) in live lambda development. Along with general slowness after not hitting the lambda endpoint for a few minutes. Any ideas on what this is about?

Warning: Ignoring extra certs from `/etc/pki/tls/certs/ca-bundle.crt`, load failed: error:02001002:system library:fopen:No such file or directory
and
Unexpected Top Level Error: Error: socket hang up

Question regarding using relational database. I see in this example, It is said that I have to go to aws ui and create tables manually. Is there a way to add db migrations as part of sst?

Question: Does the Api construct allow wildcarding for the routes? i.e.

/foo/*

would match

/foo/bar

and

/foo/bar/baz

Hi. We're trying to automate a docker compose down that stops local configs and the current running lambda at the same time. Right now we are basically running a "kill -9 <sst start process id>", but we are wondering if there is a much safer way to do that

Anyone uses Webstorm debug with SST (Node and Typescript) successfully? The best I have is to create npm debug configurations, which do stop on breakpoints but only on transpiled files, not on the original source.

Question: Is it possible to rename/configure the directory name for the lambdas? I'd like to change it from

src

to

lambdas