amazon-archives / aws-serverless-sqs-event-source Goto Github PK

The "messageAttributes" object is empty on incoming events even though my messages all should contain one attribute.

"messages": [
{
"messageId": "a5df8a12-ae1d-4a1d-b352-c8aafbe4dc4b",
"receiptHandle": "AQEBtYpGgK/MU2rTPCMsV8f+glYkIZ+Jbsptce9qPk4+00y9cOwsP33I9zj8+U0OA0jt4UTkhf/4igPE/mOhz37TFCOEGtkEB12sRNgN+bufASfwFGIUQ5tSYJIZ9QduULuO15IDFH0JAVrOIdq8ZSM3BLcV3EiD7VXha2CIg2mXmqkpcJyfcVl6kVN9WjYCWLBto6RGorejeJ9dAsn7AJcYr546oXk4z5z0q79NRq59F4Q3DESsmNYvGqjOOHOs8aJ5OXzKM4EaH7ZZwhsU+lDocNMqMLOPRPY3J34sbK4A4Te1lsOSkGE9T1CnijwZBnO00mUB/7vpvfPrXs+j/2riz7pqf8mT8Ygwblry+KKz7wTUbLt5fFpgAJqm2QUSvIdu2DI+I4b+4b3ro6OMa/G3Nw==",
"mD5OfBody": "2301f613c4160a6cf522a340da376150",
"body": "{<EXAMPLE_BODY>}",
"messageAttributes": {}
},
{
"messageId": "7c5200d2-22be-4d62-a7b1-9ac6aac2218c",
"receiptHandle": "AQEBo1djdqdLSi5Hwzy8lxcYW9C9H/kxc5pfbbLDaQhOPUFARuOlRMaahEqGU6+GrCUYnzO2CEnoIiBm6DHK3LMN7HO2RfRC/iMLjmkFbLLkWKwL2R9IsH3r6t7kiuYWfbYD0fCh0a/Cl/xkwBrE+WCfJCX3Flkl1zR5zZD/EBmnkx20PKu5u6fEGofJNKm17SAqr8IVh6eDyBb9zHig5sUvCWDrByH9jpW3gaxSKLAHJHAEIj1uyHhxGS39SaqJA6LydxsA+xwZwOjX4PYxuMT7yXgmrVm/vWQ//uL7vy1dX/BydNdSqt+4ZKJ71/phHx5TmrKS1RDDJc/enrKOw6EJTcT9ECtDg2DPjEJyspTp2n2UMTvzLbSf4kysxg1znFffnr2OOpqIRpCglE45cixi2Q==",
"mD5OfBody": "7295178dfd60ce94607294d7460fd4b7",
"body": "{<EXAMPLE_BODY>}",
"messageAttributes": {}
}
]

Currently, the SQSPoller is invoked by the CW Event schedule and invokes the message processor to process batches of messages synchronously. The current algorithm works and is cost-efficient, but cannot support high throughput. An alternative would be if the SQSPoller and MessageDispatcher classes were put into separate lambda functions and the SQSPoller invoked the MessageDispatcher lambda function asynchronously to process a batch of messages, then continued on. The MessageDispatcher lambda function would handle updating the queue with results of the execution of its batch of messages. This would dramatically increase parallelism of message processing. One caveat is that a high volume queue could lead to too many lambdas being invoked, causing the account to hit its concurrency limit. However, setting a concurrency limit on the MessageProcessor function could be used to control how many concurrent invocations would be supported.

Hello,

I've set everything up and the correct lambda function is being successfully executed.
The SQSPoller logs indicate this success is received. INFO MessageDispatcher:62 - Processed 10 messages in 1399ms. perMessageAverage=139ms

Then it crashes before any messages can be removed from the SQS queue. It gives this stack trace

java.lang.NullPointerException: java.lang.NullPointerException
java.lang.NullPointerException
at com.amazonaws.serverless.sqseventsource.MessageDispatcher.dispatch(MessageDispatcher.java:52)
at com.amazonaws.serverless.sqseventsource.SQSPoller.poll(SQSPoller.java:39)
at com.amazonaws.serverless.sqseventsource.lambda.SQSPollerHandler.handleRequest(SQSPollerHandler.java:23)
at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
at java.lang.reflect.Method.invoke(Method.java:498)

Any help would be appreciated!

Thanks,
Andrew