How to Create a Task Tree in Snowflake
To create a
task tree in Snowflake, define multiple TASK objects where each task depends on the completion of its parent task using the AFTER clause. This chaining allows tasks to run in sequence automatically, forming a tree of dependent tasks.Syntax
In Snowflake, a task tree is created by defining tasks with dependencies using the AFTER keyword. Each task runs after its parent task completes.
CREATE TASK task_name: Defines a new task.WAREHOUSE = warehouse_name: Specifies the warehouse to run the task.SCHEDULE = 'USING CRON 0 * * * * UTC': Sets a schedule for the root task.AFTER parent_task: Makes this task run after the specified parent task.AS SQL_statement: The SQL command the task executes.
sql
CREATE TASK task1 WAREHOUSE = mywh SCHEDULE = 'USING CRON 0 * * * * UTC' AS -- SQL for root task CALL some_procedure(); CREATE TASK task2 WAREHOUSE = mywh AFTER task1 AS -- SQL for child task INSERT INTO table2 SELECT * FROM table1; CREATE TASK task3 WAREHOUSE = mywh AFTER task1 AS -- SQL for another child task UPDATE table3 SET col = 'value';
Example
This example shows a simple task tree with one root task and two child tasks that run after the root completes. The root task runs every hour, and the child tasks run automatically after it.
sql
CREATE OR REPLACE TASK root_task WAREHOUSE = compute_wh SCHEDULE = 'USING CRON 0 * * * * UTC' AS INSERT INTO logs (event_time, event) VALUES (CURRENT_TIMESTAMP, 'root task ran'); CREATE OR REPLACE TASK child_task_1 WAREHOUSE = compute_wh AFTER root_task AS INSERT INTO logs (event_time, event) VALUES (CURRENT_TIMESTAMP, 'child task 1 ran'); CREATE OR REPLACE TASK child_task_2 WAREHOUSE = compute_wh AFTER root_task AS INSERT INTO logs (event_time, event) VALUES (CURRENT_TIMESTAMP, 'child task 2 ran'); -- Enable tasks to start the chain ALTER TASK root_task RESUME; ALTER TASK child_task_1 RESUME; ALTER TASK child_task_2 RESUME;
Output
Tasks created and enabled. root_task runs hourly; child_task_1 and child_task_2 run after root_task completes.
Common Pitfalls
Common mistakes when creating task trees include:
- Not enabling tasks after creation, so they never run.
- Creating circular dependencies, which cause errors.
- Forgetting to specify
AFTERfor dependent tasks, causing them to run independently. - Using different warehouses inconsistently, which can cause resource issues.
sql
/* Wrong: Circular dependency example */ CREATE TASK task_a WAREHOUSE = wh AFTER task_b AS SELECT 1; CREATE TASK task_b WAREHOUSE = wh AFTER task_a AS SELECT 2; /* Right: Linear dependency */ CREATE TASK task_a WAREHOUSE = wh AS SELECT 1; CREATE TASK task_b WAREHOUSE = wh AFTER task_a AS SELECT 2;
Quick Reference
| Clause | Description |
|---|---|
| CREATE TASK task_name | Defines a new task with the given name |
| WAREHOUSE = warehouse_name | Specifies the warehouse to run the task |
| SCHEDULE = 'USING CRON 0 * * * * UTC' | Sets the schedule for the root task only |
| AFTER parent_task | Runs this task after the specified parent task completes |
| AS SQL_statement | SQL command the task executes |
| ALTER TASK task_name RESUME | Enables the task to run |
Key Takeaways
Create dependent tasks using the AFTER clause to form a task tree.
Always enable tasks with ALTER TASK ... RESUME to start execution.
Avoid circular dependencies to prevent errors.
Schedule only the root task; child tasks run automatically after parents.
Use consistent warehouses for all tasks in the tree.
