Hey there, I'm running into a perms issue that I can't seem to figure out. Anyone know what is needed to resolve the following?

Insufficient permissions to enable logging (Service: AmazonApiGatewayV2; Status Code: 400; Error Code: BadRequestException

ooh man.. my API stack is getting rather full 😅 I think CF has a 500 resource limit, right? anyone else run into this limit yet? I wonder if it might make sense to just centralize the permissions so they all share the same policies (like Serverless Framework does). it’s not ideal, but with every single API endpoint having its own policies, it definitely eats up that resource quota

I’m using API GW v2 with an HTTP Mapping, but the

path params

apparently are not working? in the Console is not showing anything for those mappings? I’m mapping from a subdomain to another domain. This is my code

routes: {
        'GET /projects/{projectId}/filters': {
          url: `${BASE_URL}/hey-xxxx/projects/{projectId}/filters`,
          method: HttpMethod.GET,
          timeout: 30,
        },
      },

this is always a strange error (when running `sst start`:

Rebuilding infrastructure...
Rebuilding infrastructure failed

no details, no error msg. I just kill the process & rerun it, and it’s usually fine

I’ve created an API mapping for a custom domain, and this works fine. But I want to define

GET /careers

, and doesn’t seams to be possible? the URL will be

/careers/

.. notice the

/

at the end. Is there a way to tell the API to map

/careers

and

/careers/

? Also, is there any convention, or best practice on this topic? I have been defining APIs without the ending

/

all my life, but maybe is not correct.

Is this new?

❌ mycomp-vpc-stack failed: The mycomp-vpc-stack stack contains no resources.

I’m using an stack to lookup for the existing VPC… never failed until I did the update?

I can't seem to be able to set permissions to

permissions: [sst.PermissionType.ALL]

in either defaultFunctionProps at the app level or at an individual function level. I get the error:

Error: Action '*:*' is invalid. An action string consists of a service namespace, a colon, and the name of an action. Action names can include wildcards.

Can I see

sst.Cron

instances in AWS? I've got some basic code and it doesnt seem to be working and I'm not sure how to debug it 🤔

Hola! I created my sst using

--language typescript

but I failed to read through docs to add

--use-yarn

and my

sst add-cdk

is complaining (and using npm). Where is this set? I’d like to change this after the boilerplating my project 🙏

Is there a way to define dependencies between stacks? For example I have 1 stack that has some main foundation components.. if this fails to create, I dont want other stacks to be executed or created.

Does sst do anything to help seed dynamodb tables, or do we need to script that?

Is there a way to avoid type-checking over dependencies? Middy apparently imports x-ray, but I’m not using that.

Not sure if anybody experienced this… but sometimes there are resources that takes like 20 minutes to be deleted. In this case it’s an authorizer and it’s associated resources. Anybody experienced delays when deleting roles and things like that? This happened to me on SLS too.

Is there a way to configure the

sst test

command? In my case Jest is finding some folders outside of my SST project that I don’t want to test.

Running

sst test --testPathPattern=test/

for my SST stacks tests… any ideas what could be this issue?

Hey, should I be using

cdk.Fn.import_value

for sharing values across sst projects? for context, I have a SNS Topic setup in project “EventsCore”, and I want “ReaderApi” to subscribe.

this.addOutputs(..)

from EventsCore and import value in ReaderAPI? Or any cleverness/best-practices I could be considering?

^ btw, I’m trying to avoid the restrictions on the Export option of outputs in https://docs.aws.amazon.com/AWSCloudFormation/latest/UserGuide/outputs-section-structure.html

I have two stacks that point to the same folder for lambda functions/api routes. The

Auth

stack uses a function in the singers module, because it creates a singer from the user record in the Cognito UserPool in the

postConfirmation

Cognito trigger. The

Singers

stack points to the singers module, too, because it handles the rest of the singers functionality. This wasn't an issue back in my Node days, but now seems to be a hassle with Python. Since Python uses

srcPath

to point to the folder, and then the route just points to the lambda file itself, it seems that

Auth

stack is claiming all the lambda files in the singers folder, and then when the

Singers

stack tries to claim the remainder of the lambda files, it can't have any because:

Building Lambda function src/services/singers/KAR_SNG_get_singer_by_id.handler
Error: Asset is already associated with another stack 'test-karaoke-auth'. Create a new Code instance for every stack

I could create a separate folder for that one function that Auth needs, and I will if I have to, but it feels like a bit of a hack. Is there a configuration refinement I can use to not let

Auth

be greedy and steal all the lambdas in

Singers

?

File

cdk.context.json

should be added to source control?

I’m getting this warning, but what is supposed to be the type on those

catch

? Looks like has to be

any

or

unknown

but the linter doesn’t allow it.

71:19  warning  Unexpected any. Specify a different type  @typescript-eslint/no-explicit-any

Is there a way to specify input path on an EventBridge Rule? Porting some stuff over from serverless, where it looks like:

IndexLedgerLinesRule:
      Type: AWS::Events::Rule
      Properties:
        EventBusName: some=eb
        EventPattern:
          source:
            - some-source
          detail-type:
            - some-detail
        Targets:
          - Id: IndexLedgerLinesQ
            Arn: !GetAtt IndexLedgerLinesQ.Arn
            InputPath: '$.detail'

Does anybody know how to do Throttling/rate-limiting on API GW v2 APIs? maybe with the

Api

construct? Found that documentation shows this example, but HOW we do it with CDK?

aws apigatewayv2 update-stage \
    --api-id a1b2c3d4 \
    --stage-name dev \
    --route-settings '{"GET /pets":{"ThrottlingBurstLimit":100,"ThrottlingRateLimit":2000}}'

Found this on CDK, but this is for GW v1 😩 . https://docs.aws.amazon.com/cdk/api/latest/docs/@aws-cdk_aws-apigateway.ThrottleSettings.html

I'm currently on version

0.35.1

for

@serverless-stack/cli

and

@servlerless-stack/resources

. I'd like to upgrade to

v0.40.4

. What is the best yarn command I can run to upgrade, given that my dependencies look like:

"dependencies": {
    "@aws-cdk/core": "1.111.0",
    "@serverless-stack/cli": "^0.35.1",
    "@serverless-stack/resources": "^0.35.1",
    "uuid": "^8.3.2"
  }

(My confusion seems to be around there being two dependencies that make up serverless-stack)

I may have found a little buggy issue with sst and Python lambdas: • I have a Lambda, called

KAR_ROT_complete_performance.py

, which puts a

PerformanceCompleted

event on my event bus. Works great, works great, returns a result with

"FailedEntryCount": 0

- all good. • Then, I have a Lambda, called

KAR_HST_add_performance.py

, that consumes said event and places it in a performance history dynamodb table. • If I do a

npx sst deploy --stage test

and then POST to the API to complete a performance, everything works great. No errors in the CloudWatch logs, and I can see my performance in the performance table. So, it works end-to-end with a

deploy

. • If I do a

npx sst start --stage test

, my

KAR_ROT_complete_performance.py

still works well, but my

KAR_HST_add_performance.py

throws the following exception in the terminal:

2d5b30dd-1daa-4bb4-af6b-da4ea4e975cf REQUEST test-karaoke-performance-his-historylogger7C190503-GGOSGWNuWwtH [src/services/performance-history/KAR_HST_add_performance.handler] invoked
SyntaxError: Unexpected token n in JSON at position 0
    at JSON.parse (<anonymous>)
    at createStrictSyntaxError (/home/lukewyman/Documents/code/karaoke-backend-sst/node_modules/body-parser/lib/types/json.js:158:10)

• I've also simplified the subscribed lambda so that I'm not trying to do anything at all with the event:

def handler(event, context):
    print('Inside the add performance handler!')

and I still get the exception shown above, while also printing

Inside the add performance handler!

to to the terminal. So, somehow, the subscribed lambda is falling apart with Live Lambda, but working as expected when deployed.

I need to reference a function from one stack into other stack, but I can’t share the

Function

?

I think removing “empty stacks” doesn’t work very well. I have 2 “empty stacks” which actually are doing a “VPN lookup”, mostly because I don’t want to create the VPC at the moment. Those are deployed fine now, but I tried removing them and it failed.

Stack amouly-xxx-vpc-stack
  Status: failed
  Error: The amouly-xxx-vpc-stack stack failed to destroy.

But removing from CF console works fine.

Does SST or CDK-pros have any guidance on potentials around “post deploy” hooks for scripting additional

aws-sdk

work against resources? One limitation with CDK/Cloudform in my experience is the

sync

(vs.

async

) nature of it all - it’s the key reason I do love terraform, ability to interrogate deployed resources and use APIs to modify them (outside of the creation). Specifically, I’m looking at some additional SNS Topic properties, which led me to https://stackoverflow.com/a/65703536 (and I’ve verified cloudform still does not support modifying these properties).

Hi all. I'm not sure what's changed but I can't upgrade my stacks past 0.40.1. If I switch to anything higher (I've tried 0.40.2 -0.40.5) I get the following errors:

RangeError: Maximum call stack size exceeded
    at Object.normalize (path.js:265:12)
    at Object.join (path.js:429:18)
    at getManager (C:\Users\pa-de\Work\dentr\sst\node_modules\@serverless-stack\core\dist\packager\packager.js:43:33)
    at getManager (C:\Users\pa-de\Work\dentr\sst\node_modules\@serverless-stack\core\dist\packager\packager.js:48:12)
    at getManager (C:\Users\pa-de\Work\dentr\sst\node_modules\@serverless-stack\core\dist\packager\packager.js:48:12)
    at getManager (C:\Users\pa-de\Work\dentr\sst\node_modules\@serverless-stack\core\dist\packager\packager.js:48:12)
    at getManager (C:\Users\pa-de\Work\dentr\sst\node_modules\@serverless-stack\core\dist\packager\packager.js:48:12)
    at getManager (C:\Users\pa-de\Work\dentr\sst\node_modules\@serverless-stack\core\dist\packager\packager.js:48:12)
    at getManager (C:\Users\pa-de\Work\dentr\sst\node_modules\@serverless-stack\core\dist\packager\packager.js:48:12)
    at getManager (C:\Users\pa-de\Work\dentr\sst\node_modules\@serverless-stack\core\dist\packager\packager.js:48:12)

npm ERR! code ELIFECYCLE
npm ERR! errno 1
npm ERR! dentr-sst@0.1.0 deploy: `sst deploy`
npm ERR! Exit status 1
npm ERR!
npm ERR! Failed at the dentr-sst@0.1.0 deploy script.
npm ERR! This is probably not a problem with npm. There is likely additional logging output above.

npm ERR! A complete log of this run can be found in:
npm ERR!     C:\Users\pa-de\AppData\Roaming\npm-cache\_logs\2021-08-31T10_40_48_799Z-debug.log

I'm running on Windows 10, if that makes a difference? I also have a stack that has an environment property check around if to determine whether to deploy it or not, like this:

if (!(skipStatic || isLocal)) {
  new AdminSite(app, `${stackIdPrefix}${AdminSite.STACK_SUFFIX}`);
}

Based on one of the bug fixes that went into 0.40.2, about handling empty stacks, I wonder if this could be the issue?

Hello! I'm trying to recreate a serverless.yaml setup in SST for my WebSockets. In my Yaml file i have the following:

functions:
  authorizer:
    handler: fn_authorizer.lambda_handler
    runtime: nodejs14.x
  connection-manager:
    handler: fn_connection_manager.lambda_handler
    events:
      - websocket:
          route: $connect
          authorizer:
            name: authorizer
            identitySource:
              - "route.request.querystring.Auth"
      - websocket:
          route: $disconnect

This applies the custom authorizer for the $connect route. It's an IAM response. I've noticed that SST does handle IAM authorizer for WebSockets but it doesn't look to handle a custom lambda function to handle the authentication. Is there maybe a CDK solution? Or an SST solution that I'm just missing? My current solution was to deploy the Websockets via SST and then use the above yaml to attach the authorizer and routes. But this solution relies on cloud watch and cloud watch is lacking proper logging for websockets in my experience. note: connection-manager has a python runtime (not that this should change anything but yeah). Thanks

to confirm, the

@aws-cdk/assert

package, and it’s usage in docs like sst provides - is for interrogating the stack template, NOT deployed resources right? My team is writing some integration/e2e tests, and I think it would be nice to have some basic tests that poll/wait for some resources to be available - MAYBE it’s not required, but for completedness I want to confirm this is not behaviour of cdk/assert and/or is not best practice 🙏 thanks!