Db2 12.1.2 random test-data inserter

这个工具读取一个表名和数据量列表，向指定 schema 下的表插入随机数据。

直接执行

列表文件格式：

CUSTOMER,100
ORDERS,200
PRODUCT,50

执行前准备好 Db2 CLP 和数据库连接环境，然后传入 schema 和列表文件：

chmod +x random_insert.sh
./random_insert.sh APP list.txt

通过 npm 全局安装后也可以执行：

db2-random-data APP list.txt

脚本安装过程一次、设置 schema 一次、逐行插入，最后提交。

不需要在 Docker 容器内执行。只要运行脚本的机器具有 Db2 CLP，它可以连接本机 或远程 DB2。Docker 只是开发阶段可选的测试数据库。

Implemented mapping

| Db2 type | Generated value | | --- | --- | | CHAR, CHARACTER, VARCHAR | random ASCII digit string, bounded by SYSCAT.COLUMNS.LENGTH | | DECIMAL, NUMERIC | random value derived from precision (LENGTH) and SCALE, with an overflow safety margin | | integer and floating numeric types | random value within a conservative type-safe range | | DATE | CURRENT DATE | | TIMESTAMP | CURRENT TIMESTAMP | | BLOB | fixed X'ABCD1234' bytes | | any other type | NULL |

Identity and generated columns are omitted from the insert list. Inserts use a single recursive-CTE INSERT ... SELECT except for the rare table containing only generated columns.

Expected limitations

This is type-aware rather than business-aware. Foreign keys, unique keys, check constraints, triggers, row/column access control, and unsupported NOT NULL columns can still reject the insert. Very large DATACOUNT values also create a large transaction; commit batching is intentionally not hidden inside the procedure.

SQL PL 已依据 IBM Db2 12.1 文档检查；当前版本没有在真实 Db2 12.1.2 实例中 执行过，首次在目标环境运行前应先对测试表进行一次 smoke test。