Goal

As a developer, I want to control the concurrency model for the consecutive pipeline executions when using pipelines-as-code so that I they execute serially when my use-case requires or limit the number of parallel pipelinesruns.

The following are some of the use-case for serial execution:

• Interacting with legacy systems for example for system testing purposes and these legacy systems do not necessarily support parallelism. Therefore one pipelinerun at a time can execute in order to avoid unknown outcomes when interacting with these legacy systems.
• Concurrency for a single pipeline, so that git push triggers a build+deploy with other pushes queued up
• Concurrency across a chain of pipelines (less common?)
• Configure concurrency - e.g. specify the number of concurrent executions allowed

Acceptance Criteria

• Developer can specify the pipelineruns to execute serially or concurrently in pipeline as code
• Developers can specify in case of serial execution, if last pipelinerun should start executing or the pipelineruns should queue up and execute in order

Related Work

• https://github.com/tektoncd/pipeline/issues/2828
• https://github.com/tektoncd/experimental/issues/699

Questions

• Should the # of concurrent pipelines be added back to the scope?