add maptask cache clarification (#5340)

* add maptask cache clarification Signed-off-by: Daniel Sola <[email protected]> * grammar Signed-off-by: Daniel Sola <[email protected]> * Update docs/user_guide/advanced_composition/map_tasks.md Co-authored-by: Nikki Everett <[email protected]> Signed-off-by: Daniel Sola <[email protected]> --------- Signed-off-by: Daniel Sola <[email protected]> Signed-off-by: Daniel Sola <[email protected]> Co-authored-by: Nikki Everett <[email protected]>
flyteorg · May 9, 2024 · c0f5b10 · c0f5b10
1 parent aa1f211
commit c0f5b10
Showing 1 changed file with 1 addition and 0 deletions.
diff --git a/docs/user_guide/advanced_composition/map_tasks.md b/docs/user_guide/advanced_composition/map_tasks.md
@@ -58,6 +58,7 @@ def map_workflow_with_metadata(data: list[int] = [10, 12, 11, 10, 13, 12, 100, 1
         data_point=data
     )
 ```
+When `cache` and `cache_version` are used in `TaskMetadata` for a map task, the cache hits occur on individual tasks being mapped over, rather than the parent map task operation. This means that if one input item in a list changes, each previously executed task is read from cache and only the task for the changed item is actually executed, rather than the task being re-executed for every item.  Note that this has the same effect as adding `cache` and `cache_version` in the `@task` decorator for a task being mapped over.
 
 You can also configure `concurrency` and `min_success_ratio` for a map task:
 - `concurrency` limits the number of mapped tasks that can run in parallel to the specified batch size.