Currently, the following are the four modes that Data IDE offers that can enable the codes in a task to take effect for the data: page direct run, test run, system automatic and periodic run, and data completing run. For the differences among these modes and their use cases, see Data development overview.
Page direct run is suitable for code debugging and modification, scenarios without the need of configuring scheduling properties, or objects such as Script development that are run directly without the need of being submitted. This article describes how to run a task directly on the code editing page by taking ODPS_SQL node task as an example.
Page direct run is only supported for ODPS_SQL, OPEN_MR, ODPS_MR, and SHELL node types.
The task is running on the default resource group if you select to run directly on the page. To run a task on a custom resource group, see Test run a task.
Double-click an ODPS_SQL task to enter the editing area. Select the part of statements you want to run and click Run in the operation area to trigger the running process of the selected codes. If you do not make your selection and click Run all the codes of the current task will run by default.
Running ODPS_SQL node task consumes a certain amount of computing and storage resources. Therefore, costs are generated. For this reason, before Pay-As-You-Go users run an ODPS_SQL node task, a Consumer Reminder dialog box appears. It displays the estimated fees you may be billed for. The task does not run until you confirm the fees.
The estimated fees displayed in the Consumer Reminder dialog box is only a rough estimate of the fees involved in the current run. The actual fee may differ and is subject to the provided bill.
Currently, among the task and node types supported by Data IDE, billing only applies to MaxCompute. Therefore, the Consumer Reminder feature is only supported for ODPS_SQL tasks.
When the task is triggered, the log page appears at the end of the editing area. If the statement running result returns to the data set, the result page appears next to the log page. It supports features such as copying by rows or columns. In the project configuration, once you enable the download feature, the results support being downloaded on the result page.
Only one log page is provided, regardless of how many times the task runs. The page displays only the logs of the last run, with all the previous logs overwritten. A maximum of 20 result pages are displayed in the statement running order to make it easier for you to perform operations such as data comparison.
When multiple statements are triggered, they are run in a serial order, with logs displayed on the log page in sequence and results displayed on different result pages based on the running order of statements.
If the code selected to be run contains SET statements, when it runs on the page, these SET statements run in a sequence before each non-SET statement runs. This also applies when all the codes in the task run. In other running modes, this logic does not exist and statements run in a serial order.
Any data returned after a run is displayed on the result page at the end of the editing area. You can view the returned data sets by page, select and, copy some of the data, or download all the data in the result page.
The download feature is available only when it is enabled by the administrator in project configuration. If this feature is disabled in the current project, the Download button is invisible.
When a Select statement is run, only the first 10,000 records in the data set are fetched by default. Therefore, limit the number of records generated for each query. To fetch more than 10,000 records at a time, see Export SQL operation result.