dirkrkotzeml opened a new issue, #45550: URL: https://github.com/apache/airflow/issues/45550
### Apache Airflow Provider(s) amazon ### Versions of Apache Airflow Providers 8.27.0 ### Apache Airflow version 2.7.2 ### Operating System Amazon Linux ### Deployment Amazon (AWS) MWAA ### Deployment details Standard MWAA deployment - V2.7.2 ### What happened SageMakerProcessingJobs have a hard limit of 64 characters for the ProcessingJobName. In the SageMakerBaseOperator there is a check for uniqueness for the name. In the case that a name is not unique it adds a timestamp to prevent a potential collision, however there is no check to prevent the updated <jobname>-<timestamp> from exceeding 64 characters. This causes the creation of the SageMakerProcessingJob to fail. ### What you think should happen instead In the SageMaker Pipelines SDK they truncate the base name before adding the timestamp, therefor we recommend taking a similar approach for consistency purposes. ### How to reproduce Create a SageMaker ProcessingJob using the SageMakerProcessingOperator with a name of longer than 50 characters, and trigger it more than once. On the second time it is triggered the time stamp will be added and in the airflow logs it will show the error stating the sagemaker processing job failed to create due to the name exceeding the character limit ### Anything else Every time that a scheduled run occurs with the same name (after the first run) ### Are you willing to submit PR? - [X] Yes I am willing to submit a PR! ### Code of Conduct - [X] I agree to follow this project's [Code of Conduct](https://github.com/apache/airflow/blob/main/CODE_OF_CONDUCT.md) -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
