Past events are replayed every time the server starts

[jjohnson1994]

When I start serverless offline (serverless offline start), any requests from previous sessions are replayed. I think this then leads Node to run out of memory. Node doesn't run out of memory every time though.

My lambda consuming the DynamoDB stream:

export const handler = async (event, context) => {
  try {
    const promises = event.Records.map(record => {
      const { eventName } = record;
      const { S: model } = { ...record.dynamodb.NewImage, ...record.dynamodb.OldImage }.model;
      const topicArn = generateTopicArn(
        eventName,
        model,
        awsRegion,
        IS_OFFLINE === 'true' ? '123456789012' : awsAccountId
      );

      return SNS.publish({
        Message: JSON.stringify(record),
        TopicArn: topicArn
      })
        .promise()
    });

    await Promise.all(promises)

    return 200;
  } catch (error) {
    console.error("error in stream", error)
    throw error;
  }
}

My serverless config

org: jjohnson1994
app: test_app
service: test_app

plugins:
  - serverless-dotenv-plugin
  - serverless-webpack
  - serverless-dynamodb-local
  - serverless-offline-sns
  - serverless-offline-dynamodb-streams
  - serverless-offline
custom:
  tableName: 'test_app'
  serverless-offline:
    httpPort: 3001
  webpack:
    webpackConfig: 'webpack.config.js'
  dynamodb:
    stages: dev
    start:
      migrate: true
  serverless-offline-dynamodb-streams:
    apiVersion: '2013-12-02'
    endpoint: http://0.0.0.0:8000
    region: ${env:AWS_REGION}
    accessKeyId: root
    secretAccessKey: root
    skipCacheInvalidation: false
    readInterval: 500
  serverless-offline-sns:
    port: 4002
    debug: false

package:
  individually: true

provider:
  name: aws
  runtime: nodejs12.x
  stage: dev
  region: eu-west-1
  iamRoleStatements:
    - Effect: Allow
      Action:
        - dynamodb:Query
        - dynamodb:Scan
        - dynamodb:GetItem
        - dynamodb:PutItem
        - dynamodb:UpdateItem
        - dynamodb:DeleteItem
        - sns:Publish
        - sns:Subscribe
        - s3:PutObject
        - s3:PutObjectAcl
      Resource:
        - "*"
  environment:
    DB: ${self:custom.tableName}
    NODE_ENV: ${env:NODE_ENV}
    AWS_ACCOUNT_ID: ${env:AWS_ACCOUNT_ID}

functions:
  app:
    handler: index.handler
    events:
      - http:
          method: ANY
          path: /
          cors: true
      - http:
          method: ANY
          path: '{proxy+}'
          cors: true
  stream:
    handler: functions/stream/stream.handler
    events:
      - stream:
          enabled: true
          type: dynamodb
          batchSize: 1
          startingPosition: TRIM_HORIZON
          arn:
            Fn::GetAtt:
              - DynamoDBTestApp
              - StreamArn

resources:
  Resources:
    DynamoDBTestApp:
      Type: 'AWS::DynamoDB::Table'
      Properties:
        TableName: ${self:custom.tableName}
        ProvisionedThroughput:
          ReadCapacityUnits: 20
          WriteCapacityUnits: 20
        StreamSpecification:
          StreamViewType: NEW_AND_OLD_IMAGES
        AttributeDefinitions:
          - AttributeName: hk
            AttributeType: S
          - AttributeName: sk
            AttributeType: S
          - AttributeName: model
            AttributeType: S
          - AttributeName: slug
            AttributeType: S
        KeySchema:
          - AttributeName: hk
            KeyType: HASH
          - AttributeName: sk
            KeyType: RANGE
        GlobalSecondaryIndexes:
          - IndexName: gsi1
            KeySchema:
              - AttributeName: model
                KeyType: HASH
              - AttributeName: sk
                KeyType: RANGE
            Projection:
              ProjectionType: ALL
            ProvisionedThroughput:
                ReadCapacityUnits: 5
                WriteCapacityUnits: 5
          - IndexName: gsi2
            KeySchema:
              - AttributeName: model
                KeyType: HASH
              - AttributeName: slug
                KeyType: RANGE
            Projection:
              ProjectionType: ALL
            ProvisionedThroughput:
                ReadCapacityUnits: 5
                WriteCapacityUnits: 5

When starting the offline server, the terminal logs each repeated request:

Then running out of memory:

The expected behavior is that requests are only sent once, (unless the lamba returns an error response, then they can be replayed?)

[thetumper]

Seeing similar with S3. Config in serverless.yml:

  serverless-offline-s3:
    endpoint: http://localstack:4566