将SQL查询结果存储在pentaho变量中

This can indeed be confusing for SSIS users. In PDI, you don't create a recordset variable as you do in SSIS. Simply creating a job creates one for you. Each job has two different types of "Results". One for recordset rows and one for filenames.

These variables are not directly accessible; they are just part of the job. There are steps that interact with them directly. For example under the "Job" branch when you're creating a transform, there is a Get rows from results step and a Copy rows to results step. They work directly with the job's row results.

Be aware that you must manually manage the metadata for the results. This is a pain, but over-all I find PDI's method of doing this more intuitive and easier than SSIS. I find SSIS more flexible in this regard.

There are also Get files from result and Set files in result. These interact with the job's built in file results. This is simply a list of every file touched by any step configured in the job. On the job tab there are tasks that deal with it directly such as Process result filenames, Add filenames to result and Delete filenames from results. These tasks operate on the built in file results list for the job and provide an easy way to, say, archive all the files loaded by the transform you just ran.

请注意，在使用这些步骤时，它们会记录作业中每个步骤所涉及的每个文件。如果您浏览处理文件的转换（数据流）中的大多数步骤，通常会选中“添加文件到结果”复选框。如果取消选中此选项，则不会将文件名添加到作业文件结果中。您也可以通过以下Delete filenames from result步骤从文件结果中删除特定文件。