TheodoreLx commented on PR #3099: URL: https://github.com/apache/celeborn/pull/3099#issuecomment-2668447570
Here are some test results about #3018 I want to share： After triggering resumeByPinnedMemory, it will enter pause immediately. The duration of resumeByPinnedMemory is less than 10ms, and the worker will soon fall into the pause deadlock state again until resumeByPinnedMemory is triggered again in 10 seconds. In other words, worker wastes about 10 seconds staying in the pause state. <img width="2440" alt="image" src="https://github.com/user-attachments/assets/afc078d1-2eba-4c90-823c-b29b104542f5"; /> <img width="1285" alt="image" src="https://github.com/user-attachments/assets/9834a2a7-01e2-4725-865c-a23c4030930c"; /> 2025-02-19 16:53:11,053 resumeByPinnedMemory 2025-02-19 16:53:11,353 pause immediately 2025-02-19 16:53:15,749 stuck in pause deadlock again, diskBuffer=0 and all channels pause 2025-02-19 16:53:21,067 resumeByPinnedMemory again after 10s -- 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]
