发布记录 - 1.7.0

重大更新

• 基于 Spark 3.1, 3.2 和 3.3 做了充分验证, 初步支持即将发布的 3.4
• 基于 Flink 1.14, 1.15 和 1.16 做了充分验证
• 新增 Trino 接入协议 (实验特性)
• 新增 Spark 血缘插件 (实验特性)
• 新增 Spark Hive 连接器 (实验特性)
• Spark 引擎支持基于 Arrow 序列化的高性能结果集传输
• Spark 引擎支持执行 Python/PySpark 脚本 (实验特性)
• 诸多 Kyuubi Helm Chart 和 Spark on K8s 改进
• REST 和 Batch 接口在功能、客户端和稳定性上进一步增强
• Spark 权限插件重大改进，并包含了对 Apache Iceberg 的支持
• 包括LDAP认证、内部通信加密在内的诸多安全改进

Server

• [KYUUBI #2887] Add a POLLING balance policy for engine pool
• [KYUUBI #3089] Support custom event handler
• [KYUUBI #3449] Change default server info provider to ENGINE
• [KYUUBI #3545] [KYUUBI #3563] Support restrict spark configurations
• [KYUUBI #3554] REST API functions/command-line tool enhancements
• [KYUUBI #3577] Transfer connection url when opening session
• [KYUUBI #3615] Retry opening the engine when encountering a special error
• [KYUUBI #3658] Support SSL for Kyuubi thrift binary connection
• [KYUUBI #3659] Support alternative keys in ConfigBuilder
• [KYUUBI #3663] Support auto build Kubernetes client from env when Kyuubi running in Pod
• [KYUUBI #3742] Add FileSessionConfAdvisor to manage session level configuration
• [KYUUBI #3766] Support real user for KyuubiSession
• [KYUUBI #3835] Allow to use spark-internal as resource for batch job
• [KYUUBI #3839] Introduce signature mechanism to protect session variable on engine side
• [KYUUBI #3844] Forward the server ip in openSession
• [KYUUBI #3847] Add jdbc-shaded profile to support IDE debug
• [KYUUBI #3863] Arrow-based results serialization
• [KYUUBI #3867] Init SQL scripts should create table if table does not exist
• [KYUUBI #3887] Provide kyuubiServerPrincipal as alias for principal in JDBC parameter
• [KYUUBI #3897] Supplying pluggable GroupProvider
• [KYUUBI #3901] Introduce Trino frontend (experimental)
• [KYUUBI #3922] Only the ApplicationInfo with non-empty id is valid for BatchJobSubmission
• [KYUUBI #3923] Add dedicated batch session idle timeout
• [KYUUBI #3926] Introduce antlr4 to parse query statement
• [KYUUBI #3928] Application state mapping considers FinalApplicationStatus
• [KYUUBI #3950] Fix the batch metadata in-consistent issue on open batch session failure
• [KYUUBI #3975] Support to post batch session/operation event
• [KYUUBI #3983] [KYUUBI #3982] Introduce refreshing user defaults configs
• [KYUUBI #4021] Fix async start engine failure resulting in inaccurate operation metrics
• [KYUUBI #4106] Introduce resource file uploading in batch creation via REST API
• [KYUUBI #4119] Return app submission time for batch
• [KYUUBI #4145] Change lock and polling seq_num path on service discovery
• [KYUUBI #4151] Support to ignore subdomain when engine pool conditions are met
• [KYUUBI #4152] Enhance LDAP authentication
• [KYUUBI #4241] Only force close engine ref when open session failed
• [KYUUBI #4322] MySQL URL configuration joiner should use & instead of ;
• [KYUUBI #4352] Support System.gc() with periodic GC interval
• [KYUUBI #4360] Support to refresh the unlimited users for session limiter
• [KYUUBI #4372] Support to return null value for OperationsResource RowSet
• [KYUUBI #4390] Allow user to provide batch id on submitting batch job
• [KYUUBI #4418] Allow disable async retry and fail fast on unrecoverable DB error
• [KYUUBI #4419] Implement simple EngineSecuritySecretProvider

Client

• [KYUUBI #3973] Fix conf fallback for kyuubi-ctl list server
• [KYUUBI #4006] Fix the incorrect execution of the source command
• [KYUUBI #4067] Reset the operation log before fetching new one
• [KYUUBI #4216] Support to transfer client version for JDBC and REST client
• [KYUUBI #4221] REST client for creating batch with uploading resource file
• [KYUUBI #4311] Fix the wrong parsing of JVM parameters in JDBC url
• [KYUUBI #4334] REST client should catch NoHttpResponse Exception and retry

Spark Engine

• [KYUUBI #3128] Support CostMode for PlanOnlyStatement
• [KYUUBI #3385] Set executor Pod name prefix if missing in Spark on K8s case
• [KYUUBI #3441] Change default Spark version to 3.3.1
• [KYUUBI #3487] Support Kyuubi as Spark dialect
• [KYUUBI #3776] Revise Kyuubi Spark engine shaded
• [KYUUBI #3780] Add support for executing Python/PySpark scripts (experimental)
• [KYUUBI #3792] Engine UI support grace stop
• [KYUUBI #3801] Correctly calculate active stages in SQLOperationListener
• [KYUUBI #3851] Support auto set up spark.master when Kyuubi running inside K8s Pod
• [KYUUBI #3885] Fix memory leak when using incremental collect mode
• [KYUUBI #3892] Remove sensitive kyuubi.engine.credentials from spark conf
• [KYUUBI #4035] Post SparkOperationEvent and show sessionId for statements
• [KYUUBI #4150] Support to execute Scala statement synchronized to prevent conflicts
• [KYUUBI #4316] Fix returned Timestamp values may lose precision
• [KYUUBI #4336] Avoid listing all schemas for Spark session catalog on schema pruning
• [KYUUBI #4407] Adapt SLF4J 2.x
• [KYUUBI #4412] Align the session handle between server and engine for Spark engine

Spark Authz Plugin

• [KYUUBI #3424] Access privilege checks for namespaces and tables of DatasourceV2
• [KYUUBI #3515] Support Iceberg commands and skip apply row-filter to output tables
• [KYUUBI #3325] Privilege checks for permanent views and skipping shadowed tables
• [KYUUBI #3300] Support user group based policies from Ranger
• [KYUUBI #3371] Ranger client support sending requests in batch
• [KYUUBI #3581] Support row filter and data masking on permanent views
• [KYUUBI #3607] Support {OWNER} variable defined in Ranger policy
• [KYUUBI #3904] New Authz Plan Serde Layer
• [KYUUBI #4076] Modified query plan should correctly report stats
• [KYUUBI #4270] Register shutdown hook for plugin cleanup
• [KYUUBI #4255] Handle describe relation for V2 relations
• [KYUUBI #4262] Rename table requires ALTER privilege of the source table
• [KYUUBI #4202] Fix reference resolution when data masking enabled for V2 relations
• [KYUUBI #4437] Fix dependencies conflict by replacing jersey-bundle with jersey-client

Other Spark Extensions

• [KYUUBI #2282] Introduce Spark Lineage plugin (experimental)
• [KYUUBI #3365] Introduce Spark Hive connector (experimental)
• [KYUUBI #3601] Support infer columns for rebalance and sort
• [KYUUBI #3962] Add two conditions to decide if add shuffle before writing
• [KYUUBI #3988] Final stage config isolation support write only
• [KYUUBI #3893] UnspecifiedDistribution does not have default partitioning

Flink Engine

• [KYUUBI #3514] Support Flink 1.16
• [KYUUBI #3604] Propagate HADOOP_PROXY_USER on launching Flink engine
• [KYUUBI #3717] Support Flink engine see primary keys

Other Notable Changes

• [KYUUBI #3473] Add Docker Compose based Kyuubi playground
• [KYUUBI #3903] Support generate kyuubi-version-info.properties on Windows
• [KYUUBI #4000] JDBC engine supports Apache Phoenix
• [KYUUBI #4288] Use eclipse-temurin:8-jdk-focal as default base image
• [KYUUBI #4320] Trino engine supports GetPrimaryKeys
• [KYUUBI #4473] Helm Chart improvements

Credits

Last but not least, this release would not have been possible without the following contributors:

• AdithyaPadmashali
• Alex Wiss-Wolferding
• Bowen Liang
• Brandon Grams
• CHzxp
• Cheng Pan
• Deng An
• Fei Wang
• Fu Chen
• HYBG
• Hanna Liashchuk
• Hyukjin Kwon
• John Caveman
• Kent Yao
• Luning Wang
• Min Zhao
• Nietzsche_w
• Paul Lin
• Ruguo Yu
• SteNicholas
• Thomas Prelle
• Tianlin Liao
• Wang Zhen
• William Tong
• Xuedong Luan
• Yikf
• asagjj
• df_liu
• dnskr
• dongkelun
• edddddy
• firefox
• hongdongdong
• hongkunyoo
• jiaoqingbo
• jinsilei
• lcy999
• liangbowen
• liaopeng
• maming
• odone
• pengqli
• runzhliu
• senmiaoliu
• sundy xiong
• sychen
• ulysses-you
• wForget
• wangjunbo
• waywtdcc
• wxmimperio
• xuefeimiaoao
• yehere
• yexianxun
• yikf
• yongqian
• yuqi
• zhouyifan279
• zwangsheng
• 毕博
• 佘志铭
• 胡安忠
• 袁福元