PROBLEM

As the OCP OKD Engineering group,
we need to be aware of the infrastructure resources required for building and hosting
OKD/SCOS, and fulfill them.

PROPOSAL

Temporarily run SCOS bootimage builds on operate-first, until a long-term solution is found, like possibly the CentOS community infrastructure.

OKD/SCOS-specific container images (installer, ironic) will be built in the internal Prow CI system, and payloads will be distributed via quay.io, so no additional resources are needed there.

DEPENDENCIES

• compute:
  • [ ] run coreos-assembler pipeline for SCOS (short-term: operate-first; long-term: CentOS infra )
  • [x] run OKD/SCOS specific container image builds (OpenShift Prow CI)
• storage:
  • [ ] host SCOS bootimage artifacts, e.g. in an S3 bucket
  • [x] host OKD/SCOS container image payloads (quay.io)