API Key length causing deployment to fail

[revgum]

Bug Report

Description

When a specified API Key has fewer than 20 characters, deployment fails without any evidence as to why. I used to AWS Console to try to create the key manually and saw the key length requirement show up there. The error provided on the AWS Console is API Key value should be at least 20 characters.

1. What did you do?
   Set an API Key value having fewer than 20 characters in configuration.

2. What happened?
   CloudFormation repeatedly failed to create ApiGatewayApiKey1 and rolled back deployment with no error details.

3. What should've happened?
   Serverless should have prevented deployment by validating the api key length.

4. What's the content of your serverless.yml file?

provider:
  apiKeys:
    - name: ${self:service.name}-api-key
      value: tooshort 

1. What's the output you get when you use the SLS_DEBUG=* environment variable (e.g. SLS_DEBUG=* serverless deploy)

Template validation passes, typical resource creation happens and eventually:
CloudFormation - CREATE_FAILED - AWS::ApiGateway::ApiKey - ApiGatewayApiKey1 followed by all of the resource deletes.

Similar or dependent issues:

• API gateway generated key name not prefixed with stage #6458 (config example would be affected by this issue as well)

[medikoo]

@revgum thanks for report

CloudFormation repeatedly failed to create ApiGatewayApiKey1 and rolled back deployment with no error details.

I just issued such test deployment and output was:

Serverless Error ---------------------------------------
 
  An error occurred: ApiGatewayApiKey1 - API Key value should be at least 20 characters (Service: AmazonApiGateway; Status Code: 400; Error Code: BadRequestException; Request ID: 3df4aed1-4343-4cca-9e29-f79c90c37a08).

Are you relying on latest version of a Framework?

[revgum]

I was using the latest, however, my actual apiKeys value was coming from an ${env:API_KEY}. That wouldn't make sense why I don't see the same error you're seeing. Any ideas?

[medikoo]

That wouldn't make sense why I don't see the same error you're seeing. Any ideas?

What exactly ends in generated CloudFormation template?

[revgum]

It looks legit;

    "ApiGatewayApiKey1": {
      "Type": "AWS::ApiGateway::ApiKey",
      "Properties": {
        "Enabled": true,
        "Name": "abc-development-api-key",
        "Value": "abc-123-test",
        "StageKeys": [
          {
            "RestApiId": {
              "Ref": "ApiGatewayRestApi"
            },
            "StageName": "development"
          }
        ]
      },
      "DependsOn": "ApiGatewayDeployment123456789012345"
    },

...
The CREATE_FAILED followed by a bunch of DELETE_COMPLETE and finally the only error from Serverless is:

...
CloudFormation - DELETE_COMPLETE - AWS::CloudFormation::Stack - abc-development
Serverless: Stack create finished...
Serverless: Invoke aws:info
Serverless: [AWS cloudformation 400 0.475s 0 retries] describeStacks({ StackName: 'abc-development' })

  Serverless Error ---------------------------------------

  ServerlessError: Stack with id abc-development does not exist
      at /Users/joshgum/Documents/code/abc/node_modules/serverless/lib/plugins/aws/provider/awsProvider.js:331:27
      at processTicksAndRejections (internal/process/task_queues.js:97:5)

  Get Support --------------------------------------------
     Docs:          docs.serverless.com
     Bugs:          github.com/serverless/serverless/issues
     Issues:        forum.serverless.com

  Your Environment Information ---------------------------
     Operating System:          darwin
     Node Version:              12.16.1
     Framework Version:         1.68.0
     Plugin Version:            3.6.8
     SDK Version:               2.3.0
     Components Version:        2.30.2

error Command failed with exit code 1.
info Visit https://yarnpkg.com/en/docs/cli/run for documentation about this command.

[medikoo]

The CREATE_FAILED followed by a bunch of DELETE_COMPLETE and finally the only error from Serverless is:

This error seems related to some other issue with your configuration, and that's why the API key error doesn't surface for you.

It's hard for me to state what can be wrong not know your full configuration, and not having full sls deploy output (with SLS_DEBUG=*)

[revgum]

I think the underlying problem is that there are resources being created that depend on the API key and they fail being created at the same time the API key fails to be created.. the API key length error gets swallowed up by these circumstances and never surfaces in a meaningful way. The stack rollback happens, and you are left not knowing why it really failed.

When I deploy an updated lambda affecting the API Gateway along with the value of the API key to under 20 characters I get the errors I've left in the issue ticket already. The only things you don't see are the lambdas each having CREATE_FAILED, then all of the changes being deleted, the stack rolling back and the bad error indicating the stack doesn't exist even though it does.

I hope this helps!

[revgum]

I wanted to add that I tried a few scenarios;

• Deploy with a custom named API Key and a value < 20 characters

  • The error gets swallowed by Serverless and I only see the aforementioned errors.

• Deploy with an automatically generated name for API Key and a value < 20 characters

  • You will see the error from AWS stating that the key length fails

[medikoo]

The only things you don't see are the lambdas each having CREATE_FAILED, then all of the changes being deleted, the stack rolling back and the bad error indicating the stack doesn't exist even though it does.

@revgum thanks for clarification. This looks very familiar to: #7635 - it appears that it's the source of an issue that hides real error from you

[AhmedFat7y]

@revgum Could you provide list of events for that failed stack? (at least the events statuses without reosuces ids).

Serverless checks for the stack progress/success/failure through the events provided by describe stack-events. So, it might be helpful if you could provide list of events for a failed stack because I get a normal error message when I try to deploy a stack with an error in deployment stage.

[arash-bizcover]

This is 5th time in last year, I'm searching whole Github. how to explicitly set value of API key on this Serverless lambda thing😰
It just doesn't work

[medikoo]

@arash-bizcover try asking at https://forum.serverless.com/ or https://www.serverless.com/slack