maropu commented on pull request #29368: URL: https://github.com/apache/spark/pull/29368#issuecomment-670255634
> SPARK-27963 Allow dynamic allocation without a shuffle service. Thanks for the info, @dongjoon-hyun . I see and I referred to [the other ticket](https://issues.apache.org/jira/browse/SPARK-24432) and [the document statement](https://github.com/apache/spark/blame/master/docs/running-on-kubernetes.md#L491-L495). But, as you said, the current statement doesn't match it. I will checki it again and update it. > I'd wonder, what actionable difference would this make, to declare it still partly experimental?... Thanks for the comment, @srowen. Yea, this statement update doesn't gurantee anything about behaivour/configuration stability... But, I think it is a bit weird that the docment 2.3.x and 3.0.x have the same description about the development status. So, I thought it is worth updating it according to the current situation (I need to consider more about how to write it though...). > Hi, I am not absolutely sure where and when it was decided, that we need to wait for External shuffle service, for Spark on k8s to be GA, can anyone help me with the discussion link? So far, It has been stated as if already a fact. Any link to the discussion described in the mailing list, @holdenk ? Probably, she refered to the discussion @dongjoon-hyun [described above](https://lists.apache.org/thread.html/r9db328c667a2e4ac6c874dbdc41d3b20f33e0c71cb8992c9bc407789%40%3Cdev.spark.apache.org%3E)? I'm still not sure that we have a general consensus about k8s GA though, do you know something about it, @ScrapCodes ? (I roughly looked around the mailing list logs and jira/PR, but I couldn't find it though...) ---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
