4

I've created a new SageMaker pipeline using AWS Python SDK, and everything is working fine, I can trigger my pipeline and it works perfectly using the SDK with these simples commands:

pipeline.upsert(role_arn=get_execution_role())
execution = pipeline.start()

enter image description here

Now, I would like to schedule the pipeline execution to run every day during the morning (let's say 8 a.m for example). And here's my problem. I configured the EventBridge as shown in this tutorial: https://github.com/aws-samples/scheduling-sagemaker-processing-with-sagemaker-pipelines, but instead of creating a new role, I used an existing one (the same returned from the command get_execution_role() above). My event is triggered in the correct hour (every day at 8 am), but the pipeline doesn't execute. When checking the logs on Cloud Watch, It shows that I got a FailedInvocations for the event, but I don't know how to get the logs from this failed execution. I tried to search on cloud trail but don't found nothing.

Anyone could help me?

  • Can you share the logs you see in CloudWatch? Have you confirmed your IAM role has the permission to start a pipeline execution? Can you share the event rule you configured? – Kirit Thadaka Apr 11 '22 at 17:25

1 Answers1

0

You need to give permissions to the pipeline in the EventBridge Role created.If you check it must only have the pipeline name for which you have created the role. For me it worked on adding my pipeline to the Role.

  • 1
    This is helpful but is there any way you could be more explicitly? What does the permission statement look like for the pipeline and/or event bridge role? – Sledge Aug 05 '22 at 20:55