Troubleshooting - Amazon Marketing Cloud Insights on AWS
LoggingCloudTrailAlarmsTest

Troubleshooting

If these instructions don’t address your issue, see the Contact AWS Support section for instructions on opening an AWS Support case for this solution.

Logging

Resources in the stack send logging output to CloudWatch Logs. Entries in these logs can help uncover a misconfiguration or other problem with the stack.

To review log output for different resources:

  1. Navigate to the CloudWatch console.

  2. In the Logs menu, select Log Groups. Log group names for the stack begin with the service prefix, followed by the stack name and resource name.

    /aws/lambda/<StackName>-<LambdaFunctionName>
/aws/vendedlogs/states/<StackName>-<StateMachineExecution>

  3. Use the search filter at the top of the page to find all the log groups for a stack.

Note

All logs deployed by the solution are configured to never expire. These settings can be changed by editing the retention policy of each log group in the CloudWatch console.

CloudTrail

The stack creates a multi-Region CloudTrail trail that stores data to an Amazon S3 bucket.

In the CloudTrail console, choose Event history to find specific API calls relating to metadata including the action, resource name, and type.

Alarms

This stack creates several CloudWatch Alarms to monitor Lambda function invocation and SQS queue processing.

<StackName>-<LambdaFunctionName>-lambda-alarm-throttles

  • The alarm state is ALARM when the named Lambda function encounters a concurrent run limit within a 60-second period.

  • The alarm returns to OK if no throttles are encountered for 60 seconds.

  • You might be encountering the burst run limit for Lambda.

  • You might need to increase the Lambda function concurrent run limit for your account’s Region.

<StackName>-<LambdaFunctionName>-lambda-alarm-errors

  • The alarm state is ALARM when the named Lambda encounters a runtime run error within a 60-second period.

  • The alarm returns to OK if no errors are encountered for 60 seconds.

  • This can be caused by a misconfiguration or failure of resources used by the Lambda Function.

  • This can be caused by a bug in the Lambda function’s code.

<stack-name>-<QueueName>-dlq-a-alarm
<stack-name>-<QueueName>-dlq-b-alarm

  • The alarm state is ALARM when one or more SQS messages visible in the queue.

  • The alarm returns to OK if no messages are visible in the queue.

  • Message arriving in this queue can be caused by a misconfiguration or failure of resources used by a related Lambda function. This may be caused by a bug in a Lambda Function’s code.

Test

This stack includes a set of functional tests that can be used to help verify successful deployment and troubleshoot issues. You can run these tests by using the following steps.

  1. Download the Amazon Marketing Cloud Insights user scripts.

  2. Open a new terminal session and navigate to the test_scripts directory.

  3. Run the following command. Replace the <stack-name> , <profile-name> , and <region-name> variables.

    $ sh run-test.sh --<stack-name> STACK_NAME --<profile-name> PROFILE --<region-name> REGION