[SUPPORT] Spark3 无法读取hudi表

[shenbinglife]

sparksql 3.2
hudi0.9.0

有对hudi从spark2 升级到 spark3了解的吗？ 使用sparksql语法， 升级后，org.apache.spark.sql.hive.HiveExternalCatalog#restoreDataSourceTable 强制从 spark.sql.sources.schema 属性中读取schema ，但是hudi表中没有这个属性

[xushiyan]

@shenbinglife hudi 0.9.0 does not support spark 3.2; 0.9.0 is supposed to work with spark 3.0.x
hudi 0.10.0 supports spark 3.1.x
hudi 0.11.0 supports spark 3.2 and the PR is close to land #4270
Please use the right spark / hudi combinations.

cc @YannByron

[shenbinglife]

如果我想使用hudi0.10 和spark3.1 ，我是否应该自己编译hudi 并启用 -Pscala-2.12 -Pspark3， 而不是直接用maven仓库的hudi-spark-bundle-0.10 ？@xushiyan

[YannByron]

@shenbinglife
you can use the jar in maven repo directly.

[xushiyan]

@shenbinglife I updated the website and hope it clarifies. Closing this now.
https://hudi.apache.org/docs/next/quick-start-guide#setup

hudi 0.10.0 is built against spark 3.1 by default. Only when users are on spark 3.0.x, they build hudi jars for themselves