s0nskar commented on PR #3710: URL: https://github.com/apache/celeborn/pull/3710#issuecomment-4660042726
Hi @RexXiong, Thanks for the review. > The use case described (GCP live migration) can already be covered by the existing DecommissionThenIdle + Recommission workflow via the Master's /sendWorkerEvent API. This does not work for us as the notification time we get for such events is very less (Less then 5mins). It is is not possible to use decommission flow in such case. For the other points i definitely agree on your concerns but making it more restrictive might not cover cases like this – https://github.com/apache/celeborn/pull/3696 where we have to force the NONE_PAUSE state. We are using this API for such cases as well. Should i introduce a new flow, just to pause receiving new data on channels and do not have any side effect. I am open to discuss more on this. -- 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]
