, - Dataphin - Alibaba Cloud Documentation Center

A periodic instance is an instance generated when a periodic task runs based on its scheduling configuration. Each scheduled run generates one instance. You can perform O&M on generated instances, such as viewing their run status, rerunning them, and viewing node code. This topic describes the general features of periodic instances.

Instance generation and run rules

Every day at 23:00, Dataphin generates the periodic instances for the following day. By default, these instances are in the not started state.

After all of an instance's upstream dependencies succeed, its status changes from not started to waiting for scheduling time.
The status of the periodic instance changes from waiting for scheduling time to running when its scheduled time is reached, sufficient resources are available, and it passes all throttling rule checks.
Note
Pausing an instance affects only the current run and does not affect other instances of the same task. To suspend multiple periodic instances, you can change the task's scheduling property to pause scheduling. For more information, see Manage integration and compute tasks.

Access periodic instances

In the top menu bar of the Dataphin homepage, choose Develop > O&M.
In the top menu bar, select the production or development environment.
In the left-side navigation pane, choose Instance O&M > Periodic instances.

Periodic instances page

The Periodic instances page displays lists of Integration and Compute Tasks and Modeling Tasks. You can perform O&M operations on the listed instances from this page.

Area	Description
① Search and filter area	Filter: You can quickly filter by selecting my instances, failed instances, or unfinished instances, or by selecting a business date and project. The following filter options are also supported: My Instances: All instances owned by the currently logged-in account. Failed Instances: All failed instances in the projects where the currently logged-in user has the Operations - Access Directory permission. Unfinished Instances: All unfinished instances in projects where the currently logged-in user has O&M - Access Directory permissions. Project: Filter by the project where the task is located. You can select multiple projects for which the current user has O&M-Access Directory permissions. Business date: You can select Today, Yesterday, All, or a specific day or a business date range. Instance type: You can select Integration and compute tasks or Modeling tasks. Sub-type: Depending on the selected business type, the supported sub-types are as follows: Integration and Compute Tasks: The available subtypes vary depending on the bound compute source type. For specific options, refer to the UI. Modeling tasks: The available subtypes are dimension logical table, fact logical table, and aggregate logical table. Owner: You can filter by the instance owner. Run status: You can filter by seven run statuses: Succeeded, Failed, Running, Waiting for Scheduling Time, Throttling, Waiting for Scheduling Resources, and Not Started. For more information about each status, see Periodic Instance Run Status Description. Priority: Supports filtering by five instance running priorities: Highest, High, Medium, Low, and Lowest. Note If the baseline feature is enabled, only Medium, Low, and Lowest task priorities are supported. Scheduling Cycle: You can select multiple scheduling cycles for filtering. Six scheduling cycles are supported: Minute, Hour, Day, Week, Month, and Year. Run Mode: Filters instances by their run mode. The options are Normal run, dry run, and Paused run. Start Time: Filters periodic instances based on their start time. Options include Last hour, Today, Yesterday, or a specific day or a time range (accurate to the second). Instances on non-scheduling days: These are no-op instances that are automatically generated by the system on non-scheduling days. This does not include manually configured no-op instances, such as those created by conditional scheduling configurations. The available actions are Show or Hide. Note For logical table instances, an instance is hidden only when all its fields are non-scheduling-day dry runs. End Time: Filters instances by their end time. You can select Last hour, Today, Yesterday, or a specific date or business time range (accurate to the second). Scheduled Time: You can filter periodic instances by their scheduled running time. You can select 00:00-06:00, 06:00-12:00, 12:00-18:00, or a specific time range (accurate to the second). Runtime Duration: You can filter instances by their total runtime or a runtime range. Scheduling Resource Group: You can filter by the scheduling resource group of the instance's specified runtime. Retry Count: The range for the number of retries in the format `a~b`. You can enter 0 or a positive integer for both a and b. The default value for a is 0. By default, b is empty, which indicates all values greater than a. Note This filter option is not supported for modeling tasks. Auto-retry Attempts: The range for the number of auto-retries, in the format `a~b`. The default value for a is 0. If b is empty, the range includes all values greater than a. You can enter 0 or a positive integer for both a and b. Note This filter option is not supported for modeling tasks. Tags: Select one or more task tag values to filter. Directory: Filters the directories in the selected project that contain integration tasks, compute tasks, data collection, and workflow instances. Note Only integration and compute tasks support filtering by a specific directory. Click Reset to clear all filters and restore the system defaults. Search: You can search for periodic instances by entering a node ID or node name. You can also click the icon in the search box to open the Batch Search dialog box. In the Batch Search dialog box, you can search by manually entering multiple task IDs or uploading a document. You can combine a batch search with other filter conditions. Batch Search - Manual Input: In the Batch Search dialog box, select Manual Input, enter up to 5000 task IDs separated by line breaks, and click OK to search. Batch search by uploading a document: In the Batch Search dialog box, select Upload Document and click Upload to upload an Excel file. The system automatically reads data from rows 2 to 5001 of the first column (the task ID column) from the first sheet. You can click Download Task ID List Template to download and use the template.
② List operations	Customize list columns: Click the icon to select the columns to display and adjust their order. Note The first column and the Actions column cannot be hidden or reordered. If you clear your browser cache or switch browsers, the column settings revert to their defaults. Refresh list: Click the icon to refresh the current list.
③ Periodic instance list	The Periodic Instance page lists all periodic tasks from projects for which you have O&M - Access Directory permissions. These tasks are divided into Integration and Calculation Tasks and Modeling Tasks. The Actions column displays the supported O&M operations for each periodic instance. You can adjust the width of each column. Your settings are saved in your local browser and are cleared if you clear your browser history and cookies. The O&M operations supported for each periodic instance type are as follows: For O&M operations supported by integration and compute task instances, see Supported operations for integration and compute task instances. For O&M operations supported by modeling task instances, see Supported operations for periodic modeling task instances. Note The periodic instance list displays only instances from the current tenant.
④ Batch Actions	You can use the batch actions area to perform operations such as rerun, terminate, mark as succeeded and continue, pause, resume, modify scheduling resource group, and modify priority on multiple periodic instances. For batch O&M operations supported by integration and compute task instances, see Batch operations for integration and compute task instances. For batch O&M operations supported by modeling task instances, see Logical table fields and batch operations.
⑤ Time zone conversion	When the scheduling time zone of a tenant and the system time zone (which is the time zone displayed in the User Center) do not match, you can use time zone conversion to view the Scheduled Run Time, Start Run Time, and End Run Time under either the scheduling time zone or the system time zone. Note The Start Time, End Time, and Scheduled Time filters only support filtering based on the System Time Zone.

Periodic instance indicators

Icon	Label	Description
	Delayed	For an event fact logical table with data latency detection enabled, this indicator appears on a successful periodic instance when data latency is detected.
	Dry run	The scheduling property of this periodic instance task is set to dry run.
	Paused	The scheduling property of this periodic instance task is set to pause scheduling.

Periodic instance run status

Status icon	Run status	Description
	Not started	An instance that has not started running.
	Waiting for scheduling resources	All upstream node dependencies have succeeded, and the instance is waiting for scheduling resources.
	Throttling	The instance is being throttled.
	Waiting for scheduling time	All upstream node dependencies have succeeded, and the instance is waiting for its scheduled time.
	Running	The instance is currently running.
	Failed	The instance failed to run. If you terminate an instance that is waiting or running, its status changes to failed.
	Succeeded	The instance ran successfully. An instance from a dry run is automatically marked as succeeded by the system.

Periodic instance DAG

Click an instance object name in the periodic instance list to display a DAG on the right, centered on the corresponding node. The DAG provides a dynamic, visual representation of the node's upstream and downstream dependencies. By default, it shows the Main node (the selected node) and its immediate upstream and downstream nodes. If an instance has no upstream or downstream nodes, the DAG shows only the current Main node. You can also perform O&M on upstream and downstream nodes.

You can adjust the DAG view:

Click the icon to hide the instance list and enlarge the DAG.
Click the icon to hide the DAG.
Hover over the icon and drag to expand or shrink the DAG view.

The DAG interface for modeling task instances is the same as for integration and compute task instances. The following figure uses an integration and compute task instance as an example:

Area	Description
① Node information	Displays summary information for the currently selected node. Click View Node Details to view more node information, including instance probability, run diagnostics, run logs, operation logs, node code, and task parameters. Instance Overview: Displays basic information (such as the node name, ID, instance ID, tenant, and project), run information (such as run status and run time), and resource consumption (memory/CPU allocation and actual memory/CPU usage). Note Except for SQL tasks (compute engine SQL tasks and database SQL tasks), resource consumption information is displayed in the node details of periodic instances and backfill instances only if the Scheduling Resource Dashboard is enabled under System Settings > Resource Settings for the lakehouse tenant. Gantt charts and the run diagnosis feature are not available for external tenant nodes (nodes that are not in the current tenant). Run Diagnosis: View the current instance's upstream dependencies, scheduled time, throttling rules, scheduling resources, and instance execution. For more information, see Instance run diagnosis. Run Log: View the run log for a succeeded or failed instance. Operation Log: View the operation log for the current instance, including the operation time, operator, and specific action. Node Code/Materialized Code: View the code for the current instance. Integration and compute task instances: Includes Instance Overview, Run Diagnosis, Run Log, Operation Log, and Node Code. Modeling task instances: Includes Instance Overview, Run Diagnosis, Run Log, Operation Log, and Materialized Code. External tenant node (a node not in the current tenant): Displays only the Instance Overview.
② Node filter area	If the view becomes crowded, you can search by node name to locate a specific node.
③ Scheduling dependency graph	Displays the scheduling dependency graph for the instance. You can expand more upstream and downstream nodes and perform O&M operations on them. Hover over a DAG node to view its name, type, scheduling cycle, owner, and description. Nodes from other tenants are marked with a cross-tenant label and only support expanding their parent and child nodes. For O&M operations supported by DAG nodes of periodic integration and compute instances, see Supported operations for DAG nodes of integration and compute task instances. For O&M operations supported by DAG nodes of periodic modeling tasks, see Supported operations for DAG nodes of modeling task instances.
④ Canvas controls	You can quickly adjust the display scale of the DAG by setting the scale (100% by default), zooming in (up to 200%), zooming out (down to 20%), fitting the DAG to the canvas, or enabling a full-screen display. Additionally, the node ID and name of the Main node in the current DAG are displayed.
⑤ Instance object list	When you view the DAG, the instance object list displays the instance object name and business date. Hover over an instance object to view its node name, node ID, scheduling cycle, business date, and scheduled run time. See the following figure: