This function demonstrates how to use S3 put event to trigger a Lambda function which picks up the dropped device endpoint list and publishes notification to at a very fast rate.

1. Deploy this function with timeout setting of 5 minutes. Make sure the lambda Role has S3 read/write permissions to the bucket and SNS publish permission. Configure Lambda environment variables e.g REGION=us-east-1, THREADS=1000, LOG_TIME=1

2. Create a S3 bucket and set up events for PUT to trigger this lambda function.

3. In the S3 events configuration, set the event suffix to .gz.

4. Create a payloads file i.e endpoint_list_14032016.json with contents in the below json format.

   {
    "Endpoints": [
      {
        "EndpointArn": "arn:aws:sns:us-west-2:1111122222:endpoint/GCM/MyApp/55a1ffbf-aefc-3e7a-bd84-3af5bca4fc63",
        "Message": "{\\\"data\\\": { \\\"message\\\": \\\"Test from Lambda.\\\", \\\"title\\\": \\\"Test\\\"}}"
      }
    ]
   }
   

5. If sending the same message to all endpoints, use the following json format.

   {
    "SameMessage": true,
    "Message": "{\\\"data\\\": { \\\"message\\\": \\\"Same message to all endpoints.\\\", \\\"title\\\": \\\"Test\\\"}}",
    "Endpoints": [
      {
        "EndpointArn": "arn:aws:sns:us-west-2:1111122222:endpoint/GCM/MyApp/55a1ffbf-aefc-3e7a-bd84-3af5bca4fc63"
      }
    ]
   }
   

6. Compress the file using gzip. e.g gzip -kf endpoint_list_14032016.json' creates 'endpoint_list_14032016.json.gz

7. Upload the file endpoint_list_14032016.json.gz to the S3 bucket which will trigger this lambda function.

8. This function will publish to all endpoints in the json file and log failures in <FILENAME>_error.log.

Tip: Publish even faster by distributing endpoints across multiple smaller json files when the number of endpoints exceed over a few 100,000s or increase max_threads variable value to something higher.