This is an automated email from the ASF dual-hosted git repository.
dataroaring pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/doris-website.git
The following commit(s) were added to refs/heads/master by this push:
new 14386d3a691 add doc for group_commit_mode table property (#3465)
14386d3a691 is described below
commit 14386d3a691508b36939af634ebe54097035db81
Author: meiyi <[email protected]>
AuthorDate: Fri Mar 20 17:09:39 2026 +0800
add doc for group_commit_mode table property (#3465)
add doc for pr: https://github.com/apache/doris/pull/61242
## Versions
- [x] dev
- [x] 4.x
- [ ] 3.x
- [ ] 2.1
## Languages
- [x] Chinese
- [x] English
## Docs Checklist
- [ ] Checked by AI
- [ ] Test Cases Built
---
docs/data-operate/import/group-commit-manual.md | 69 ++++++++++++++++++++++
.../data-operate/import/group-commit-manual.md | 68 +++++++++++++++++++++
.../data-operate/import/group-commit-manual.md | 68 +++++++++++++++++++++
.../data-operate/import/group-commit-manual.md | 68 +++++++++++++++++++++
4 files changed, 273 insertions(+)
diff --git a/docs/data-operate/import/group-commit-manual.md
b/docs/data-operate/import/group-commit-manual.md
index 1844c80d51e..692bff42cc6 100644
--- a/docs/data-operate/import/group-commit-manual.md
+++ b/docs/data-operate/import/group-commit-manual.md
@@ -52,6 +52,69 @@ PROPERTIES (
);
```
+### Table Property Configuration
+
+:::info
+`group_commit_mode` table property is supported from **4.1.0**.
+:::
+
+You can set the default Group Commit mode at the table level. When Stream Load
does not set the `group_commit` HTTP Header, the mode from the table property
will be used.
+
+**Configure during table creation:**
+
+```sql
+CREATE TABLE `dt` (
+ `id` int(11) NOT NULL,
+ `name` varchar(50) NULL,
+ `score` int(11) NULL
+) ENGINE=OLAP
+DUPLICATE KEY(`id`)
+DISTRIBUTED BY HASH(`id`) BUCKETS 1
+PROPERTIES (
+ "replication_num" = "1",
+ "group_commit_mode" = "async_mode"
+);
+```
+
+**Modify table property:**
+
+```sql
+# Modify to synchronous mode
+ALTER TABLE dt SET ("group_commit_mode" = "sync_mode");
+
+# Disable Group Commit
+ALTER TABLE dt SET ("group_commit_mode" = "off_mode");
+```
+
+**View table property:**
+
+`SHOW CREATE TABLE` displays the `group_commit_mode` property (unless the
value is `off_mode`):
+
+```sql
+mysql> SHOW CREATE TABLE dt;
++-------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
+| Table | Create Table
|
++-------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
+| dt | CREATE TABLE `dt` (
+ `id` int(11) NOT NULL,
+ `name` varchar(50) NULL,
+ `score` int(11) NULL
+) ENGINE=OLAP
+DUPLICATE KEY(`id`)
+DISTRIBUTED BY HASH(`id`) BUCKETS 1
+PROPERTIES (
+ "group_commit_mode" = "async_mode"
+) |
++-------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
+1 row in set (0.00 sec)
+```
+
+**Priority Description:**
+
+- For Stream Load: If the `group_commit` HTTP Header is set, the Header value
takes priority; otherwise, the table property value is used
+- For INSERT INTO VALUES: Session variable `group_commit` has higher priority
than table property
+
+
### Using JDBC
When users write using JDBC's `insert into values` method, to reduce SQL
parsing and planning overhead, we support MySQL protocol's `PreparedStatement`
feature on the FE side. When using `PreparedStatement`, SQL and its load plan
are cached in session-level memory cache, and subsequent loads directly use the
cached objects, reducing FE CPU pressure. Here's an example of using
`PreparedStatement` in JDBC:
@@ -233,6 +296,8 @@ func logInsertStatistics() {
### INSERT INTO VALUES
+Enable Group Commit by setting the Session variable `group_commit`. Session
variables have higher priority than table properties.
+
* Asynchronous Mode
```sql
@@ -298,6 +363,10 @@ mysql> set group_commit = off_mode;
### Stream Load
+When importing via Stream Load, you can enable Group Commit by setting the
`group_commit` parameter in the HTTP Header.
+
+**Note**: If the `group_commit` Header is not set but `group_commit_mode` is
configured in the table property, the table property mode will be used
automatically.
+
Assuming `data.csv` contains:
```sql
diff --git
a/i18n/zh-CN/docusaurus-plugin-content-docs/current/data-operate/import/group-commit-manual.md
b/i18n/zh-CN/docusaurus-plugin-content-docs/current/data-operate/import/group-commit-manual.md
index f2ce02c9d13..d818c961f66 100644
---
a/i18n/zh-CN/docusaurus-plugin-content-docs/current/data-operate/import/group-commit-manual.md
+++
b/i18n/zh-CN/docusaurus-plugin-content-docs/current/data-operate/import/group-commit-manual.md
@@ -52,6 +52,68 @@ PROPERTIES (
);
```
+### 表属性配置
+
+:::info
+`group_commit_mode` 表属性从 **4.1.0** 版本开始支持。
+:::
+
+可以在表级别设置默认的 Group Commit 模式。当 Stream Load 未设置 `group_commit` HTTP Header
时，将使用表属性中的模式。
+
+**建表时配置：**
+
+```sql
+CREATE TABLE `dt` (
+ `id` int(11) NOT NULL,
+ `name` varchar(50) NULL,
+ `score` int(11) NULL
+) ENGINE=OLAP
+DUPLICATE KEY(`id`)
+DISTRIBUTED BY HASH(`id`) BUCKETS 1
+PROPERTIES (
+ "replication_num" = "1",
+ "group_commit_mode" = "async_mode"
+);
+```
+
+**修改表属性：**
+
+```sql
+# 修改为同步模式
+ALTER TABLE dt SET ("group_commit_mode" = "sync_mode");
+
+# 关闭 Group Commit
+ALTER TABLE dt SET ("group_commit_mode" = "off_mode");
+```
+
+**查看表属性：**
+
+`SHOW CREATE TABLE` 会显示 `group_commit_mode` 属性（除非值为 `off_mode`）：
+
+```sql
+mysql> SHOW CREATE TABLE dt;
++-------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
+| Table | Create Table
|
++-------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
+| dt | CREATE TABLE `dt` (
+ `id` int(11) NOT NULL,
+ `name` varchar(50) NULL,
+ `score` int(11) NULL
+) ENGINE=OLAP
+DUPLICATE KEY(`id`)
+DISTRIBUTED BY HASH(`id`) BUCKETS 1
+PROPERTIES (
+ "group_commit_mode" = "async_mode"
+) |
++-------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
+1 row in set (0.00 sec)
+```
+
+**优先级说明：**
+
+- 对于 Stream Load：如果设置了 `group_commit` HTTP Header，则优先使用 Header 中的值；否则使用表属性中的值
+- 对于 INSERT INTO VALUES：Session 变量 `group_commit` 的优先级高于表属性
+
### 使用 JDBC
当用户使用 JDBC `insert into values`方式写入时，为了减少 SQL 解析和生成规划的开销，我们在 FE 端支持了 MySQL 协议的
`PreparedStatement` 特性。当使用 `PreparedStatement` 时，SQL 和其导入规划将被缓存到 Session
级别的内存缓存中，后续的导入直接使用缓存对象，降低了 FE 的 CPU 压力。下面是在 JDBC 中使用 `PreparedStatement` 的例子：
@@ -233,6 +295,8 @@ func logInsertStatistics() {
### INSERT INTO VALUES
+通过设置 Session 变量 `group_commit` 来启用 Group Commit。Session 变量的优先级高于表属性。
+
* 异步模式
```sql
@@ -298,6 +362,10 @@ mysql> set group_commit = off_mode;
### Stream Load
+通过 Stream Load 进行导入时，可以在 HTTP Header 中设置 `group_commit` 参数来启用 Group Commit。
+
+**注意**：如果未设置 `group_commit` Header 且表属性中配置了 `group_commit_mode`，则自动使用表属性中的模式。
+
假如`data.csv`的内容为：
```sql
diff --git
a/i18n/zh-CN/docusaurus-plugin-content-docs/version-4.x/data-operate/import/group-commit-manual.md
b/i18n/zh-CN/docusaurus-plugin-content-docs/version-4.x/data-operate/import/group-commit-manual.md
index f2ce02c9d13..d818c961f66 100644
---
a/i18n/zh-CN/docusaurus-plugin-content-docs/version-4.x/data-operate/import/group-commit-manual.md
+++
b/i18n/zh-CN/docusaurus-plugin-content-docs/version-4.x/data-operate/import/group-commit-manual.md
@@ -52,6 +52,68 @@ PROPERTIES (
);
```
+### 表属性配置
+
+:::info
+`group_commit_mode` 表属性从 **4.1.0** 版本开始支持。
+:::
+
+可以在表级别设置默认的 Group Commit 模式。当 Stream Load 未设置 `group_commit` HTTP Header
时，将使用表属性中的模式。
+
+**建表时配置：**
+
+```sql
+CREATE TABLE `dt` (
+ `id` int(11) NOT NULL,
+ `name` varchar(50) NULL,
+ `score` int(11) NULL
+) ENGINE=OLAP
+DUPLICATE KEY(`id`)
+DISTRIBUTED BY HASH(`id`) BUCKETS 1
+PROPERTIES (
+ "replication_num" = "1",
+ "group_commit_mode" = "async_mode"
+);
+```
+
+**修改表属性：**
+
+```sql
+# 修改为同步模式
+ALTER TABLE dt SET ("group_commit_mode" = "sync_mode");
+
+# 关闭 Group Commit
+ALTER TABLE dt SET ("group_commit_mode" = "off_mode");
+```
+
+**查看表属性：**
+
+`SHOW CREATE TABLE` 会显示 `group_commit_mode` 属性（除非值为 `off_mode`）：
+
+```sql
+mysql> SHOW CREATE TABLE dt;
++-------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
+| Table | Create Table
|
++-------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
+| dt | CREATE TABLE `dt` (
+ `id` int(11) NOT NULL,
+ `name` varchar(50) NULL,
+ `score` int(11) NULL
+) ENGINE=OLAP
+DUPLICATE KEY(`id`)
+DISTRIBUTED BY HASH(`id`) BUCKETS 1
+PROPERTIES (
+ "group_commit_mode" = "async_mode"
+) |
++-------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
+1 row in set (0.00 sec)
+```
+
+**优先级说明：**
+
+- 对于 Stream Load：如果设置了 `group_commit` HTTP Header，则优先使用 Header 中的值；否则使用表属性中的值
+- 对于 INSERT INTO VALUES：Session 变量 `group_commit` 的优先级高于表属性
+
### 使用 JDBC
当用户使用 JDBC `insert into values`方式写入时，为了减少 SQL 解析和生成规划的开销，我们在 FE 端支持了 MySQL 协议的
`PreparedStatement` 特性。当使用 `PreparedStatement` 时，SQL 和其导入规划将被缓存到 Session
级别的内存缓存中，后续的导入直接使用缓存对象，降低了 FE 的 CPU 压力。下面是在 JDBC 中使用 `PreparedStatement` 的例子：
@@ -233,6 +295,8 @@ func logInsertStatistics() {
### INSERT INTO VALUES
+通过设置 Session 变量 `group_commit` 来启用 Group Commit。Session 变量的优先级高于表属性。
+
* 异步模式
```sql
@@ -298,6 +362,10 @@ mysql> set group_commit = off_mode;
### Stream Load
+通过 Stream Load 进行导入时，可以在 HTTP Header 中设置 `group_commit` 参数来启用 Group Commit。
+
+**注意**：如果未设置 `group_commit` Header 且表属性中配置了 `group_commit_mode`，则自动使用表属性中的模式。
+
假如`data.csv`的内容为：
```sql
diff --git
a/versioned_docs/version-4.x/data-operate/import/group-commit-manual.md
b/versioned_docs/version-4.x/data-operate/import/group-commit-manual.md
index f086e4eb991..1fd293b1bc3 100644
--- a/versioned_docs/version-4.x/data-operate/import/group-commit-manual.md
+++ b/versioned_docs/version-4.x/data-operate/import/group-commit-manual.md
@@ -52,6 +52,68 @@ PROPERTIES (
);
```
+### Table Property Configuration
+
+:::info
+`group_commit_mode` table property is supported from **4.1.0**.
+:::
+
+You can set the default Group Commit mode at the table level. When Stream Load
does not set the `group_commit` HTTP Header, the mode from the table property
will be used.
+
+**Configure during table creation:**
+
+```sql
+CREATE TABLE `dt` (
+ `id` int(11) NOT NULL,
+ `name` varchar(50) NULL,
+ `score` int(11) NULL
+) ENGINE=OLAP
+DUPLICATE KEY(`id`)
+DISTRIBUTED BY HASH(`id`) BUCKETS 1
+PROPERTIES (
+ "replication_num" = "1",
+ "group_commit_mode" = "async_mode"
+);
+```
+
+**Modify table property:**
+
+```sql
+# Modify to synchronous mode
+ALTER TABLE dt SET ("group_commit_mode" = "sync_mode");
+
+# Disable Group Commit
+ALTER TABLE dt SET ("group_commit_mode" = "off_mode");
+```
+
+**View table property:**
+
+`SHOW CREATE TABLE` displays the `group_commit_mode` property (unless the
value is `off_mode`):
+
+```sql
+mysql> SHOW CREATE TABLE dt;
++-------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
+| Table | Create Table
|
++-------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
+| dt | CREATE TABLE `dt` (
+ `id` int(11) NOT NULL,
+ `name` varchar(50) NULL,
+ `score` int(11) NULL
+) ENGINE=OLAP
+DUPLICATE KEY(`id`)
+DISTRIBUTED BY HASH(`id`) BUCKETS 1
+PROPERTIES (
+ "group_commit_mode" = "async_mode"
+) |
++-------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
+1 row in set (0.00 sec)
+```
+
+**Priority Description:**
+
+- For Stream Load: If the `group_commit` HTTP Header is set, the Header value
takes priority; otherwise, the table property value is used
+- For INSERT INTO VALUES: Session variable `group_commit` has higher priority
than table property
+
### Using JDBC
When users write using JDBC's `insert into values` method, to reduce SQL
parsing and planning overhead, we support MySQL protocol's `PreparedStatement`
feature on the FE side. When using `PreparedStatement`, SQL and its load plan
are cached in session-level memory cache, and subsequent loads directly use the
cached objects, reducing FE CPU pressure. Here's an example of using
`PreparedStatement` in JDBC:
@@ -233,6 +295,8 @@ func logInsertStatistics() {
### INSERT INTO VALUES
+Enable Group Commit by setting the Session variable `group_commit`. Session
variables have higher priority than table properties.
+
* Asynchronous Mode
```sql
@@ -298,6 +362,10 @@ mysql> set group_commit = off_mode;
### Stream Load
+When importing via Stream Load, you can enable Group Commit by setting the
`group_commit` parameter in the HTTP Header.
+
+**Note**: If the `group_commit` Header is not set but `group_commit_mode` is
configured in the table property, the table property mode will be used
automatically.
+
Assuming `data.csv` contains:
```sql
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
