Yikun commented on PR #37802: URL: https://github.com/apache/spark/pull/37802#issuecomment-1239590984
@zheniantoushipashi Much thanks for you feedback. Most of spark + volcano users who I have come into contact with before, they also configure each job through configuration items. There is no doubt that `PodGroupTemplate`'s mechanism will be greatly improved for generality. Another previous consideration is to **reduce the frequent addition of configuration items**. But it is worth noting that `PodGroupTemplate` is different from `PodTemplate`: the CRD of `Pod` has a lot of options, but the CRD of `PodGroup`, especially the commonly used configurations, is relatively fixed (Therefore, configuration items will not be added frequently). Commonly used are `queue`, `minResource`, `priority`. BTW, please note that `minMember` is a bit special, because `minMember` is set to `1` to monitor the driver/job life cycle (see more in https://github.com/volcano-sh/volcano/pull/2038), then to ensure the accuracy of scheduling. Therefore, this parameter should not be set arbitrarily in spark + volcano case. Therefore, I personally think it is reasonable to use `PodGroupTemplate` to cover the whole scene, and then increase flexibility through configuration items. But we still need more voice from spark community before 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: reviews-unsubscr...@spark.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org
