[
https://issues.apache.org/jira/browse/FLINK-37941?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17987151#comment-17987151
]
Xuyang Zhong edited comment on FLINK-37941 at 7/3/25 7:14 AM:
--------------------------------------------------------------
I think the effective matrix should be as follows:
||partitions exist on old table||partitions exist on new table||merge
strategy||merge result ||
|yes |yes |INCLUDING |if
partitions on new table are same as those on the old table, use them;
else an error will be thrown. |
|yes |no |INCLUDING
|partitions on old table |
|no |yes |INCLUDING
|partitions on new table |
|no |no |INCLUDING
|none |
|yes |yes |EXCLUDING
|partitions on new table |
|yes |no |EXCLUDING
|none |
|no |yes |EXCLUDING
|partitions on new table |
|no |no |EXCLUDING
|none |
was (Author: xuyangzhong):
I think the effective matrix should be as follows:
|| partitions exist on old table || partitions exist on new table || merge
strategy || merge result ||
| yes | yes | INCLUDING
| ERROR |
| yes | no | INCLUDING
| partitions on old table |
| no | yes | INCLUDING
| partitions on new table |
| no | no | INCLUDING
| none |
| yes | yes | EXCLUDING
| partitions on new table |
| yes | no | EXCLUDING
| none |
| no | yes | EXCLUDING
| partitions on new table |
| no | no | EXCLUDING
| none |
> EXCLUDING PARTITIONS doesn't work when using CREATE TABLE LIKE
> --------------------------------------------------------------
>
> Key: FLINK-37941
> URL: https://issues.apache.org/jira/browse/FLINK-37941
> Project: Flink
> Issue Type: Bug
> Components: Table SQL / Planner
> Affects Versions: 2.0.0, 1.20.1
> Reporter: Xuyang Zhong
> Priority: Major
>
> The following test can reproduce this bug:
>
> {code:java}
> @Test
> def test(): Unit = {
> tEnv
> .executeSql(s"""
> |CREATE TABLE `originalT` (
> | `msisdn` VARCHAR(2147483647),
> | `uuid` VARCHAR(2147483647) NOT NULL,
> | `street_id` VARCHAR(2147483647),
> | `dt` VARCHAR(2147483647) NOT NULL,
> | CONSTRAINT `PK_dt_uuid` PRIMARY KEY (`dt`, `uuid`) NOT
> ENFORCED
> |) COMMENT 'radius日志表，以事件日期为分区'
> |PARTITIONED BY (`dt`)
> |WITH (
> | 'connector' = 'values'
> |)
> |""".stripMargin)
> .await()
> tEnv
> .executeSql(s"""
> |create table `newT` LIKE `originalT`(
> | INCLUDING ALL
> | EXCLUDING PARTITIONS
> | )
> |;
> |""".stripMargin)
> .await()
> val re = tEnv.executeSql("show create table `newT`;")
> println(re.asInstanceOf[TableResultImpl].collect().next())
> } {code}
> The result is:
> {code:java}
> CREATE TABLE `default_catalog`.`default_database`.`newT` (
> `msisdn` VARCHAR(2147483647),
> `uuid` VARCHAR(2147483647) NOT NULL,
> `street_id` VARCHAR(2147483647),
> `dt` VARCHAR(2147483647) NOT NULL,
> CONSTRAINT `PK_dt_uuid` PRIMARY KEY (`dt`, `uuid`) NOT ENFORCED
> ) PARTITIONED BY (`dt`)
> WITH (
> 'connector' = 'values'
> ) {code}
> It seems that `EXCLUDING PARTITIONS` doesn't work actually.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
