E
Easy Caterpillar
Current Behavior:
API /postExecuteStages:
Can execute a specific stage.
Requires sending the entire runtimeInputYaml payload.
This defeats the purpose of InputSets, since all inputs must still be manually provided.
API /postPipelineExecuteWithInputSetList:
Accepts only inputSetReferences and stageIdentifiers.
Always executes the entire pipeline, even if a stageIdentifier is specified.
Not designed for stage-level execution.
The Ask:
Trigger only a specific stage in a pipeline.
Reference an existing InputSet for runtime inputs.
Avoid rebuilding and passing the full runtimeInputYaml.
Example Goal:
Instead of sending:
runtimeInputYaml: |
pipeline:
identifier: my_pipeline
stages:
- stage:
identifier: my_stage
spec:
service:
identifier: my_service
The customer wants to run:
{
"stageIdentifiers": ["my_stage"],
"inputSetReferences": ["my_input_set"]
}
… and have the execution resolve inputs automatically from the InputSet.
Impact:
Current behavior creates overhead for customers with complex pipelines and existing InputSets.
Inconsistent with the intended value of InputSets (avoiding re-supplying inputs).
Forces additional scripting/automation outside of Harness to generate runtimeInputYaml.
Proposed Enhancement:
Extend /postExecuteStages to support inputSetReferences directly (along with stageIdentifiers).
Allow execution of a single stage using only the InputSet, without requiring the full runtime YAML.
Created by Pedro Mastelaro
·