Dynamic Child Pipelines
You'll learn: How to programmatically generate child pipeline configurations at runtime, enabling conditional multi-pipeline execution based on dynamic conditions.
What Are Dynamic Child Pipelines?
Building on the parent-child pipeline relationship you've learned, dynamic child pipelines take it further: instead of triggering a static .gitlab-ci.yml file, the parent pipeline generates the child's configuration on-the-fly based on runtime conditions, repository state, or external data.
Think of it like a restaurant manager who doesn't just delegate tasks to chefs—they write custom recipes for each chef based on today's available ingredients and customer orders.
How It Works
A parent job executes a script (bash, Python, etc.) that:
- Analyzes conditions (changed files, environment variables, API responses)
- Generates a new YAML configuration file as an artifact
- Triggers a child pipeline using that generated configuration
generate-child:
stage: prepare
script:
- ./scripts/generate-pipeline.sh > child-pipeline.yml
artifacts:
paths:
- child-pipeline.yml
trigger-dynamic-child:
stage: deploy
trigger:
include:
- artifact: child-pipeline.yml
job: generate-child
strategy: depend
Your generate-pipeline.sh might detect changed microservices and only create jobs for those services, or adjust deployment targets based on branch names.
Why Use This Pattern?
- Efficiency: Only run jobs relevant to current changes
- Scalability: Handle hundreds of microservices without bloated config files
- Flexibility: Adjust pipeline structure based on external data or complex logic
- Modularity: Keep the parent pipeline clean while generating complex child workflows
Key Takeaway: Dynamic child pipelines let you programmatically create pipeline configurations at runtime, enabling intelligent, condition-based workflow generation that adapts to your repository's state.