[ https://issues.apache.org/jira/browse/KYLIN-5394?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Jiale He resolved KYLIN-5394. ----------------------------- Resolution: Fixed > Project-level JDBC pushdown queries may not work when multiple projects set > different pushdown configurations > ------------------------------------------------------------------------------------------------------------- > > Key: KYLIN-5394 > URL: https://issues.apache.org/jira/browse/KYLIN-5394 > Project: Kylin > Issue Type: Bug > Components: Query Engine > Affects Versions: 5.0-alpha > Reporter: Jiale He > Assignee: Jiale He > Priority: Major > Fix For: 5.0-alpha > > Attachments: KYLIN-5394 Design & Test.pdf, > image-2023-01-10-11-09-47-237.png, image-2023-01-10-11-12-12-271.png, > image-2023-01-10-11-13-22-052.png > > > h1. 背景 > 按照查询类型分为两个项目, 一个项目配置下压查询外部spark, 另一个配置下压查询ch. > 步骤一:创建两个项目 > project1: 是一套下压引擎 配置 {{PushDownRunnerJdbcImpl}} 下压 > !image-2023-01-10-11-09-47-237.png|width=882,height=193! > project2: 是一套下压引擎 同样配置 {{{}PushDownRunnerJdbcImpl{}}}下压,但是配置不同 > !image-2023-01-10-11-12-12-271.png|width=881,height=174! > 步骤二: > 初次查询 project1 项目下压查询是OK的,然后在查询 project2 > 查询 project2 的下压就会报错 > h1. Root Cause > !image-2023-01-10-11-13-22-052.png|width=877,height=354! > kylin5 在初始化下压引擎的连接管理器时做了非空判断,如果之前已经存在下压引擎,就会复用之前的下压引擎,导致使用之前的下压引擎来查project2报错 > SDK {{{}PushDownRunnerJdbcImpl{}}}下压引擎的连接管理器 manage为单例,缓存了第一个配置, > 导致使用之前的下压引擎来查project2报错 > -- This message was sent by Atlassian Jira (v8.20.10#820010)