This folder contains our infrastructure documentation. These documents may be of interest for the team that maintain the Tekton own CI/CD setup as well as for anyone interested in using Tekton to run (part of) their own CI/CD infrastructure.
The infra system relies on several different kubernetes clusters, three are static and the rest are dynamic (provisioned on demand).
- prow: Prow, Boskos and Tekton run in this cluster.
This cluster runs resources defined in the
prow
folder. CI Jobs that only require a container run in thetest-pods
namespace of this cluster. - dogfooding: Tekton runs in this cluster. This cluster is
setup with resources from the
tekton
folder, plus a few secrets. - robocat: This cluster is our test bed for continuous deployment of services and resources. Everything that runs in this cluster is deployed automatically, which means it must be possible at any time to delete the cluster and recreate it from scratch.
Automation for the tektoncd
org runs in a GKE cluster which
members of the governing board
have access to.
There are several GCP projects used by Tekton:
- The GCP project that is used for GKE, storage, etc. is called
tekton-releases
. It has several GKE clusters:- The GKE cluster that
Prow
,Tekton
, andboskos
run in is calledprow
and is used - The GKE cluster that is used for nightly releases and other dogfooding is called
dogfooding
- The GKE cluster that
- The GCP project
tekton-nightly
is used to hold nightly release artifacts and the robocat cluster
The script addpermissions.py gives users access to these projects.
DNS Names are managed via Netlify. The setup of DNS record, for now, is manual only.