Files
mesh-lab/scenarios/openai-bed.yml
T
jschoubben 5c91c0ecd2 Scenarios: egress where images are needed, and images: cut to what is ours
Every scenario that places a container runtime gives each of its machines
`egress: true` — a node that runs modules pulls images from the internet, which is what
a node does. The underlay-only scenarios (bootstrap-single, behind-nat,
segmented-and-unforwardable, the-ordinary-shape, two-on-a-segment) stay sealed on
purpose: an extra NIC would change the very reachability they are asserting about.

`images:` keeps only the mesh's own — 55 third-party entries leave whole-mesh-full
alone, and the machine fetches them itself by the digest its module.json already pins.
The whole-mesh beds also say per machine which of ours they get: novox the substrate
control plane and its own fifteen runtimes, ace its twenty-four, the two workstations
one each. That is not a lab economy. An operator's workstation holds the images its own
modules need, and giving these two the union would put some thirty gigabytes onto a
thirty-gigabyte disk.

bootstrap-with-registry.yml is deleted. It existed only to demonstrate the lab's
registry, nothing referenced it, and there is nothing left for it to demonstrate.

Claude-Session: https://claude.ai/code/session_01LrgweAeERJYBg88c5cKDzF
2026-09-10 23:16:23 +02:00

40 lines
1.7 KiB
YAML

# One machine that becomes a mesh, then plays out the STATIC-KEY half of model-access (novox/hq
# ADR 0050) for OpenAI. Where the Anthropic bed proves the refreshable-grant shape (a manager holds a
# refresh token, the mesh delivers only the access token), this proves the OTHER shape ADR 0050
# defines and no vendor had exercised: a single operator-supplied key, sealed to the consumer holder,
# unsealed by the host, and written where an OpenAI/Codex client reads it. There is no manager, no
# refresh, and no usage — a static-key vendor exposes none, which ADR 0054 says is fine.
#
# The flow the test drives:
# the operator sets the API key on the licence -> the mesh seals it to the consumer holder ->
# the host unseals it and mounts it at the consumer's secret path -> the consumer writes it as
# OPENAI_API_KEY (env file + the Codex auth.json). The test asserts the written key equals the one
# the operator set — through the sealed delivery path, unchanged.
#
# MESH_LAB_HOST_BINARY=.../mesh-host MESH_LAB_BUNDLE=.../examples/substrate-first-node.lock
# Build the consumer runtime image into the local daemon first (raise() stocks it from there):
# scripts/build-module-runtime.sh openai-consumer /tmp/openai-consumer.tar
scenario: openai-bed
segments:
hosting:
kind: public
cidr: [192.0.2.0/24]
machines:
anchor:
at: { segment: hosting, address: [192.0.2.10] }
egress: true
inbound: allow
memory: 3GiB
cpus: 2
images:
- mesh-control:development
# The static-key consumer runtime, built by scripts/build-module-runtime.sh into the local daemon and
# loaded onto the machine, which holds it by its own image ID.
- mesh-runtime-openai-consumer:development
place:
all: [host, runtime]