jschoubben
5c91c0ecd2
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
59 lines
2.6 KiB
YAML
59 lines
2.6 KiB
YAML
# Two machines: one is the substrate, the other carries a lavinmq PROVIDER and a consumer of it.
|
|
#
|
|
# lavinmq is the mesh's own control-plane broker (mesh-broker), and it is ALSO offered as a
|
|
# user-facing capability: a module that needs a message queue gets its OWN broker — a scoped vhost and
|
|
# user on a lavinmq PROVIDER — not an account on the control broker. That makes lavinmq the sharp
|
|
# two-node case, for the same reason two-node-db is: the substrate's broker publishes 5672 on the node
|
|
# it runs on, and a lavinmq provider must publish 5672 too for its consumers to reach it — so the two
|
|
# cannot share a machine. The moment a real amqp provider must own a node's 5672, it collides with the
|
|
# control broker already there, and the chain is blocked single-node.
|
|
#
|
|
# This is the split that unblocks it. `anchor` runs the substrate (store, broker, control) and NOTHING
|
|
# else. `laptop` runs the whole chain: the lavinmq PROVIDER and the amqp-ping CONSUMER that requires
|
|
# it. Provider and consumer are co-located on laptop, so the grant never crosses a node boundary; only
|
|
# enrolment crosses to anchor, over the underlay both machines share. And because the substrate broker
|
|
# is on the OTHER node, the provider owns laptop's 5672 uncontested.
|
|
#
|
|
# It needs a host binary and the substrate bundle:
|
|
#
|
|
# MESH_LAB_HOST_BINARY=.../mesh-host MESH_LAB_BUNDLE=.../examples/substrate-first-node.lock
|
|
#
|
|
# HELPER — stock the two runtimes into the local daemon before the run (some may already be there):
|
|
# scripts/build-module-runtime.sh lavinmq /tmp/lavinmq.tar
|
|
# scripts/build-module-runtime.sh amqp-ping /tmp/amqp-ping.tar
|
|
# The service image cloudamqp/lavinmq:latest must be in the local daemon too — it is already stocked as
|
|
# the substrate's own broker image; each node pulls what it runs from the internet, over its own
|
|
# uplink.
|
|
scenario: lavinmq-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
|
|
laptop:
|
|
at: { segment: hosting, address: [192.0.2.11] }
|
|
egress: true
|
|
inbound: allow
|
|
memory: 3GiB
|
|
cpus: 2
|
|
|
|
images:
|
|
- mesh-control:development
|
|
# The lavinmq provider's runtime (reused for its run-once bootstrap and its provisioner) and the
|
|
# amqp-ping consumer's runtime, both built by scripts/build-module-runtime.sh into the local daemon
|
|
# and loaded onto the machines, which hold them by their own image IDs.
|
|
# The lavinmq SERVICE image is cloudamqp/lavinmq:latest, already stocked above.
|
|
- mesh-runtime-lavinmq:development
|
|
- mesh-runtime-amqp-ping:development
|
|
|
|
place:
|
|
all: [host, runtime]
|