Is Glue Spark shuffle manager supported with Glue 3.0?

0

Is Glue Spark shuffle manager supported with Glue 3.0? I'm trying to use --write-shuffle-files-to-s3 flag with a Glue 3.0 job and it doesn't seem to work as no files are written.

If it's not supported - will it be supported eventually?

  • Update: after changing the job to Glue 2.0 shuffle seems to work (at least shuffle files are written). So the question is now whether this will be supported in Glue 3.0.

已提問 2 年前檢視次數 1711 次
1 個回答
0
已接受的答案

AWS Glue Spark Shuffle Manager is currently only supported on AWS Glue version 2.0, and Spark 2.4.3. as mentioned in documentation link [1] under "Notes and limitations" Section. Unfortunately this feature will not work if your Glue job is using Glue version 3.0.

I understand that after downgrading the Glue job to Glue version 2.0 you were able to make the shuffle work.

Regarding your query, whether it be supported by Glue 3.0, please note that it is a pending feature request with the Glue team. Unfortunately, there is no ETA or time frame on when this feature or if these features will be rolled out by our service team as each release cycle of features is decided by the demand and urgency of the features. Although, we do appreciate your feedback.

You can keep an eye on our "What's New with AWS" blog to monitor our new releases as they become available.

[+] https://aws.amazon.com/new/

———

Reference:

[1] AWS Glue Spark shuffle manager with Amazon S3 - https://docs.aws.amazon.com/glue/latest/dg/monitor-spark-shuffle-manager.html

AWS
支援工程師
已回答 2 年前
profile picture
專家
已審閱 1 個月前

您尚未登入。 登入 去張貼答案。

一個好的回答可以清楚地回答問題並提供建設性的意見回饋,同時有助於提問者的專業成長。

回答問題指南