Environment Required 1.8 Kyuubi is pre-built with Java 1.8 Spark Distribute SQL Engine Optional 3.0 and above By default Kyuubi is pre- built w/ a Apache Spark release inside at $KYUUBI_HOME/externals metastore client to communicate with the higher version Hive metastore server. For example, Spark 3.0 was released with a builtin Hive client (2.3.7), so, ideally, the version of server should >= 2.3.x component. Nevertheless, if run Kyuubi’s engines on Kubernetes, the ESS is not an option yet. Since Spark 3.0, the DRA can run without ESS. The relative feature called Shuffle Tracking was introduced by SPARK-

Environment Required 1.8 Kyuubi is pre-built with Java 1.8 Spark Distribute SQL Engine Optional 3.0 and above By default Kyuubi is pre- built w/ a Apache Spark release inside at $KYUUBI_HOME/externals metastore client to communicate with the higher version Hive metastore server. For example, Spark 3.0 was released with a builtin Hive client (2.3.7), so, ideally, the version of server should >= 2.3.x component. Nevertheless, if run Kyuubi’s engines on Kubernetes, the ESS is not an option yet. Since Spark 3.0, the DRA can run without ESS. The relative feature called Shuffle Tracking was introduced by SPARK-

metastore client to communicate with the higher version Hive metastore server. For example, Spark 3.0 was released with a built-in Hive client (2.3.7), so, ideally, the version of server should >= 2.3 component. Nevertheless, if run Kyuubi’s engines on Kubernetes, the ESS is not an option yet. Since Spark 3.0, the DRA can run without ESS. The relative feature called Shuffle Tracking was introduced by SPARK- Since v1.1.0, Kyuubi support building with different Spark profiles, Profile Default Since -Pspark-3.0 Yes 1.0.0 -Pspark-3.1 No 1.1.0 1.6. Defining the Apache Mirror for Spark or Flink By default, we

metastore client to communicate with the higher version Hive metastore server. For example, Spark 3.0 was released with a built-in Hive client (2.3.7), so, ideally, the version of server should >= 2.3 component. Nevertheless, if run Kyuubi’s engines on Kubernetes, the ESS is not an option yet. Since Spark 3.0, the DRA can run without ESS. The relative feature called Shuffle Tracking was introduced by SPARK- Since v1.1.0, Kyuubi support building with different Spark profiles, Profile Default Since -Pspark-3.0 Yes 1.0.0 -Pspark-3.1 No 1.1.0 1.6. Defining the Apache Mirror for Spark or Flink By default, we

metastore client to communicate with the higher version Hive metastore server. For example, Spark 3.0 was released with a built-in Hive client (2.3.7), so, ideally, the version of server should >= 2.3 component. Nevertheless, if run Kyuubi’s engines on Kubernetes, the ESS is not an option yet. Since Spark 3.0, the DRA can run without ESS. The relative feature called Shuffle Tracking was introduced by SPARK- the following table. Spark Version Supported Remark master √ - 3.3.x √ - 3.2.x √ - 3.1.x √ - 3.0.x √ - 2.4.x and earlier × PR 2367 is used to track how we work with older releases with scala 2.11

metastore client to communicate with the higher version Hive metastore server. For example, Spark 3.0 was released with a built-in Hive client (2.3.7), so, ideally, the version of server should >= 2.3 component. Nevertheless, if run Kyuubi’s engines on Kubernetes, the ESS is not an option yet. Since Spark 3.0, the DRA can run without ESS. The relative feature called Shuffle Tracking was introduced by SPARK- the following table. Spark Version Supported Remark master √ - 3.3.x √ - 3.2.x √ - 3.1.x √ - 3.0.x √ - 2.4.x and earlier × PR 2367 is used to track how we work with older releases with scala 2.11

metastore client to communicate with the higher version Hive metastore server. For example, Spark 3.0 was released with a built-in Hive client (2.3.7), so, ideally, the version of server should >= 2.3 component. Nevertheless, if run Kyuubi’s engines on Kubernetes, the ESS is not an option yet. Since Spark 3.0, the DRA can run without ESS. The relative feature called Shuffle Tracking was introduced by SPARK- the following table. Spark Version Supported Remark master √ - 3.3.x √ - 3.2.x √ - 3.1.x √ - 3.0.x √ - 2.4.x and earlier × PR 2367 is used to track how we work with older releases with scala 2.11

metastore client to communicate with the higher version Hive metastore server. For example, Spark 3.0 was released with a built-in Hive client (2.3.7), so, ideally, the version of server should >= 2.3 component. Nevertheless, if run Kyuubi’s engines on Kubernetes, the ESS is not an option yet. Since Spark 3.0, the DRA can run without ESS. The relative feature called Shuffle Tracking was introduced by SPARK- the following table. Spark Version Supported Remark master √ - 3.3.x √ - 3.2.x √ - 3.1.x √ - 3.0.x √ - 2.4.x and earlier × PR 2367 is used to track how we work with older releases with scala 2.11

metastore client to communicate with the higher version Hive metastore server. For example, Spark 3.0 was released with a built-in Hive client (2.3.7), so, ideally, the version of server should >= 2.3 component. Nevertheless, if run Kyuubi’s engines on Kubernetes, the ESS is not an option yet. Since Spark 3.0, the DRA can run without ESS. The relative feature called Shuffle Tracking was introduced by SPARK- the following table. Spark Version Supported Remark master √ - 3.3.x √ - 3.2.x √ - 3.1.x √ - 3.0.x √ - 2.4.x and earlier × PR 2367 is used to track how we work with older releases with scala 2.11

metastore client to communicate with the higher version Hive metastore server. For example, Spark 3.0 was released with a built-in Hive client (2.3.7), so, ideally, the version of server should >= 2.3 component. Nevertheless, if run Kyuubi’s engines on Kubernetes, the ESS is not an option yet. Since Spark 3.0, the DRA can run without ESS. The relative feature called Shuffle Tracking was introduced by SPARK- the following table. Spark Version Supported Remark master √ - 3.3.x √ - 3.2.x √ - 3.1.x √ - 3.0.x √ - 2.4.x and earlier × PR 2367 is used to track how we work with older releases with scala 2.11