Pipeline architecture

Diagram of the dataflow throughout the pipeline

You have a decent (albeit somewhat obsolete) high-level introduction in under the hood part 1 and under the hood part 2. See also the diagrams and notes below.

Triggers dataflow

The triggers are the way pipelines are started. There are several different actors that can start them automatically, depending on where the code change that needs to be tested is made: RHEL kernel merge requests, new commits in upstream kernel trees, GitLab bots to test merge requests, etc.

The trigger code is defined in the com/pipeline-trigger and com/cki-tools repositories, whereas the tree configurations and trigger variables used in the triggers are defined in com/pipeline-data and cee/pipeline-data.

Triggers dataflow
Triggers dataflow

Results dataflow

Results dataflow
Results dataflow

Last modified September 9, 2021: Update pipeline architecture figure and add some info from internal docs (574b395)