tvalentyn opened a new issue, #39406: URL: https://github.com/apache/beam/issues/39406
### What happened? ### What happened? We have identified a memory leak that affects Apache Beam Python SDK versions 2.67.0-2.75.0. The leak was caused a code change and will be remediated in Beam 2.76.0. The leak might cause memory buildup for long-running pipelines that have regular processing errors. Eventually SDK process might OOM and restart. It is expected that the Beam runners recover from these OOMs, but it OOMs might introduce processing delay while workers recover. **Mitigation** * Use `apache-beam==2.66.0` or below. * Use `apache-beam==2.76.0` or higher * Reduce processing errors in the pipeline since the presence of errors influence the RAM growth. * If comfortable, consider applying the patch to the Beam SDK you use: https://github.com/apache/beam/pull/39405/changes . **Additional details** The leak was caused by the improvements to error logging added in https://github.com/apache/beam/pull/35403 , which inadvertently stores bundle processors for failed instructions on Python heap. ### Issue Priority Priority: 2 (default / most bugs should be filed as P2) ### Issue Components - [ ] Component: Python SDK - [ ] Component: Java SDK - [ ] Component: Go SDK - [ ] Component: Typescript SDK - [ ] Component: IO connector - [ ] Component: Beam YAML - [ ] Component: Beam examples - [ ] Component: Beam playground - [ ] Component: Beam katas - [ ] Component: Website - [ ] Component: Infrastructure - [ ] Component: Spark Runner - [ ] Component: Flink Runner - [ ] Component: Prism Runner - [ ] Component: Twister2 Runner - [ ] Component: Hazelcast Jet Runner - [ ] Component: Google Cloud Dataflow Runner -- 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]
