diff --git a/scenarios/catalogue-mqtt.yml b/scenarios/catalogue-mqtt.yml new file mode 100644 index 0000000..ef47164 --- /dev/null +++ b/scenarios/catalogue-mqtt.yml @@ -0,0 +1,46 @@ +# One machine that becomes a mesh and then assigns itself mosquitto — the bed that proves the +# run-once primitive (novox/hq ADR 0052) end to end. +# +# mosquitto is the sharp case ADR 0052 was written for: its Dynamic Security plugin loads at broker +# start and refuses to come up unless `dynamic-security.json` already holds an admin client. That +# file is not state a reconcile loop can carry — it is a step that must run once, after the data +# directory exists and BEFORE the broker container starts. mosquitto's manifest declares that step +# as a `run-once: true` init container, placed before the `server` (broker) container: the same +# runtime image, carrying mosquitto's own `bootstrap/` entrypoint and `mosquitto_ctrl`, run to +# completion under the module's own account. The host runs it, requires exit 0, and only then reaches +# the broker — so "the broker came up" is itself the proof the seed ran, because an unseeded store +# crash-loops the broker. +# +# MESH_LAB_HOST_BINARY=.../mesh-host MESH_LAB_BUNDLE=.../examples/substrate-first-node.lock +# scripts/build-module-runtime.sh mosquitto builds mesh-runtime-mosquitto:development (carrying +# mosquitto_ctrl and the compiled bootstrap entrypoint) into the local daemon, which this scenario +# stocks and serves by digest from its own registry. eclipse-mosquitto:2 must be in the local +# daemon to be stocked. +scenario: catalogue-mqtt + +segments: + hosting: + kind: public + cidr: [192.0.2.0/24] + +machines: + anchor: + at: { segment: hosting, address: [192.0.2.10] } + inbound: allow + memory: 3GiB + cpus: 2 + +images: + # The first-node substrate: store, broker, control. + - postgres:17-alpine + - cloudamqp/lavinmq:latest + - mesh-control:development + # mosquitto's broker (the service image) and its runtime, the latter built by + # scripts/build-module-runtime.sh mosquitto into the local daemon and stocked into the scenario's + # own registry, which is where the host pulls it from. The runtime image is reused for the + # run-once bootstrap step and for the serve container. + - eclipse-mosquitto:2 + - mesh-runtime-mosquitto:development + +place: + all: [host, runtime] diff --git a/scripts/build-module-runtime.sh b/scripts/build-module-runtime.sh index af653d2..ecc87d0 100755 --- a/scripts/build-module-runtime.sh +++ b/scripts/build-module-runtime.sh @@ -21,7 +21,7 @@ BASE="${RUNTIME_BASE:-node:22-bookworm-slim}" ( cd "$MESH_SDK" && npm run build >/dev/null ) ( cd "$MESH_TOOLS" && npm run build >/dev/null ) # Compile whichever of the module's entrypoints exist. -SRCS=(); for f in client.ts index.ts tools/index.ts provisioner/index.ts; do [ -f "$MOD/$f" ] && SRCS+=("$f"); done +SRCS=(); for f in client.ts index.ts tools/index.ts provisioner/index.ts bootstrap/index.ts; do [ -f "$MOD/$f" ] && SRCS+=("$f"); done TSC="$MESH_SDK/node_modules/.bin/tsc"; ( cd "$MOD" && "$TSC" "${SRCS[@]}" --module NodeNext --moduleResolution NodeNext --target ES2022 --outDir dist >/dev/null ) STAGE="$(mktemp -d)"; trap 'rm -rf "$STAGE"' EXIT @@ -41,6 +41,11 @@ EXTRA="" case "$MODULE" in postgres) EXTRA='RUN apt-get update && apt-get install -y --no-install-recommends postgresql-client && rm -rf /var/lib/apt/lists/*' ;; minio) EXTRA='COPY --from=minio/mc:latest /usr/bin/mc /usr/bin/mc' ;; + # mosquitto drives its dynsec admin — and its run-once bootstrap seeds the store — through + # `mosquitto_ctrl`. It is not in `mosquitto-clients` on bookworm; the `mosquitto` package carries + # it (with its shared libraries), and installing from apt keeps them together — copying the binary + # out of the (musl) eclipse-mosquitto image into this (glibc) base would not load. + mosquitto) EXTRA='RUN apt-get update && apt-get install -y --no-install-recommends mosquitto && rm -rf /var/lib/apt/lists/*' ;; # mongodb's client shells out to `mongosh`. Install it from MongoDB's own apt repo so its shared # libraries come with it — copying just the binary out of the mongo image leaves it unable to load. mongodb) EXTRA='RUN apt-get update && apt-get install -y --no-install-recommends gnupg curl ca-certificates && curl -fsSL https://pgp.mongodb.com/server-7.0.asc | gpg --dearmor -o /usr/share/keyrings/mongodb.gpg && echo "deb [signed-by=/usr/share/keyrings/mongodb.gpg] https://repo.mongodb.org/apt/debian bookworm/mongodb-org/7.0 main" > /etc/apt/sources.list.d/mongodb.list && apt-get update && apt-get install -y --no-install-recommends mongodb-mongosh && rm -rf /var/lib/apt/lists/*' ;; diff --git a/test/integration/assigned-catalogue-mqtt.test.ts b/test/integration/assigned-catalogue-mqtt.test.ts new file mode 100644 index 0000000..fb02003 --- /dev/null +++ b/test/integration/assigned-catalogue-mqtt.test.ts @@ -0,0 +1,368 @@ +/** + * The mesh assigns mosquitto, and a run-once step seeds the Dynamic Security store BEFORE the broker + * starts — the end-to-end proof of the run-once primitive (novox/hq ADR 0052). + * + * mosquitto is the case ADR 0052 was written for. Its dynsec plugin loads at broker start and refuses + * to bring the broker up unless `dynamic-security.json` already holds an admin client — a file that is + * not state a reconcile loop can carry, but a step that must run once, after the data directory + * exists and before the broker container starts. The manifest declares that step as a `run-once: true` + * init container placed before the `server` (broker) container: the module's own runtime image, + * carrying mosquitto's `bootstrap/` entrypoint and `mosquitto_ctrl`, run to completion under the + * module's own account (ADR 0047). The host runs it, requires exit 0, and — because a failed step + * gates the apply — only then reaches the broker. + * + * So the proof is layered, and the centre of it is that the broker came up at all: + * - the run-once step ran to completion and was not left running (a step, not a service); + * - the seed file is on disk, owned by the broker's uid so the broker can read it AND persist to it + * (the ownership question ADR 0052 left for the lab to settle — fixed in mosquitto's bootstrap); + * - the broker is up and STABLE — which it cannot be against an unseeded store, so a broker that + * stays up is itself the proof the seed ran at the right phase; + * - the node reached `current`, which the apply reaches only past a completed run-once step. + * And, as a consumer-facing proof on top: the seeded admin authenticates over MQTT (the plugin loaded + * the seed), and mosquitto's provisioner — running in the assigned runtime — creates a scoped client + * for a contribution the mesh delivered, which then authenticates with the password the mesh minted. + * + * MESH_LAB_HOST_BINARY=.../mesh-host MESH_LAB_BUNDLE=.../examples/substrate-first-node.lock + * scripts/build-module-runtime.sh mosquitto builds mesh-runtime-mosquitto:development (carrying + * mosquitto_ctrl and the compiled bootstrap entrypoint) into the local daemon, which + * scenarios/catalogue-mqtt.yml stocks. eclipse-mosquitto:2 must be in the local daemon to be + * stocked; the host pulls both from the scenario's own registry by digest. + */ + +import { test, before, after } from "node:test"; +import assert from "node:assert/strict"; +import { existsSync, readFileSync } from "node:fs"; +import { loadScenario } from "../../src/declaration/parse.ts"; +import { raise } from "../../src/lifecycle/raise.ts"; +import { destroy, exec } from "../../src/lifecycle/operate.ts"; +import { hostBinaryPath, HOST_PATH } from "../../src/lifecycle/place.ts"; +import { labIsUsable, destroyAll } from "./harness.ts"; + +const capability = await labIsUsable(); +const binary = hostBinaryPath(); +const bundle = process.env["MESH_LAB_BUNDLE"] ?? ""; + +const skip = !capability.usable + ? `lab not usable: ${capability.why}` + : !binary || !existsSync(binary) + ? "MESH_LAB_HOST_BINARY is not set to a built mesh-host" + : !bundle || !existsSync(bundle) + ? "MESH_LAB_BUNDLE is not set to a substrate bundle (mesh-host examples/)" + : false; + +const SCENARIO = "catalogue-mqtt"; +const MACHINE = "anchor"; + +let instanceId = ""; +let stocked: string[] = []; + +function quote(s: string): string { + return `'${s.replaceAll("'", `'\\''`)}'`; +} + +async function on(command: string, timeoutMs?: number): Promise<{ out: string; ok: boolean }> { + const { stdout } = await exec(instanceId, MACHINE, [ + "sh", "-c", `exec 2>&1\n${command}\necho "__exit=$?"`, + ], timeoutMs); + const marker = stdout.lastIndexOf("__exit="); + if (marker < 0) return { out: stdout, ok: false }; + return { out: stdout.slice(0, marker), ok: stdout.slice(marker + 7).trim() === "0" }; +} + +async function must(command: string, timeoutMs?: number): Promise { + const { out, ok } = await on(command, timeoutMs); + if (!ok) throw new Error(`${MACHINE}: ${command}\n${out}`); + return out; +} + +/** The control plane, a container on the node. */ +async function mesh(command: string, timeoutMs?: number): Promise { + return must(`docker exec mesh-control /mesh-control ${command}`, timeoutMs); +} + +/** The pinned reference for one of the scenario's images, by repository. */ +function pinned(repository: string): string { + const found = stocked.find((r) => r.slice(r.indexOf("/") + 1, r.indexOf("@")) === repository); + assert.ok(found, `the scenario stocks no ${repository}; it serves ${stocked.join(", ")}`); + return found; +} + +/** The substrate bundle, its image references pointed at this scenario's own registry. */ +function bundleFor(images: string[]): string { + let text = readFileSync(bundle, "utf8"); + for (const ref of images) { + const repository = ref.slice(ref.indexOf("/") + 1, ref.indexOf("@")); + const escaped = repository.replaceAll("/", "\\/").replaceAll(".", "\\."); + text = text.replaceAll(new RegExp(`[A-Za-z0-9_.:-]+\\/${escaped}@sha256:[0-9a-f]+`, "g"), ref); + } + return text; +} + +function tokenFrom(said: string): string { + const found = said.split("\n").map((l) => l.trim()).find((l) => l.length > 100 && !l.includes(" ")); + assert.ok(found, `no token in:\n${said}`); + return found; +} + +async function settled(withinMs = 600_000): Promise { + const until = Date.now() + withinMs; + let last = ""; + while (Date.now() < until) { + const asked = await on(`docker exec mesh-control /mesh-control status --json`); + if (asked.ok) { + try { + const state = JSON.parse(asked.out) as { + wrong: { node: string; outcome: string }[]; + waiting: { node: string }[]; + reported: { node: string; outcome: string; current: boolean }[]; + }; + const bad = state.wrong.find((w) => w.node === MACHINE); + if (bad) throw new Error(`${MACHINE} did not apply what it was sent: ${bad.outcome}\n${asked.out}`); + const word = state.reported.find((r) => r.node === MACHINE); + if (!state.waiting.some((w) => w.node === MACHINE) && word?.outcome === "applied" && word.current) return; + last = asked.out; + } catch (err) { + if (err instanceof Error && err.message.includes("did not apply")) throw err; + last = asked.out; + } + } + await new Promise((r) => setTimeout(r, 5000)); + } + throw new Error(`${MACHINE} never caught up within ${Math.round(withinMs / 1000)}s. Last:\n${last}`); +} + +before(async () => { + if (skip) return; + + const raised = await raise(loadScenario(`scenarios/${SCENARIO}.yml`), { + onProgress: (m) => console.log(`raise: ${m}`), + }); + instanceId = raised.instanceId; + stocked = raised.images; + + // Raise the substrate — store, broker, control — from the bundle. + await must(`cat > /tmp/substrate.lock <<'MESHBUNDLE'\n${bundleFor(raised.images)}\nMESHBUNDLE`); + await must(`${HOST_PATH} apply /tmp/substrate.lock`, 600_000); + const up = await must(`docker ps --format '{{.Names}}'`); + for (const c of ["mesh-store", "mesh-broker", "mesh-control"]) { + assert.match(up, new RegExp(c), `the substrate did not raise ${c}:\n${up}`); + } + + // The node joins its own mesh and starts the host so it applies what it is pushed. + await mesh(`node add ${MACHINE}`); + const token = tokenFrom(await mesh(`token issue --node ${MACHINE}`)); + await must(`${HOST_PATH} enrol --token ${quote(token)}`); + await must(`nohup ${HOST_PATH} run > /var/log/mesh-host.log 2>&1 & sleep 3`); +}, { timeout: 1_800_000 }); + +after(async () => { + if (instanceId) await destroy(instanceId); + await destroyAll(`${SCENARIO}-`); +}, { timeout: 600_000 }); + +test("the mesh assigns mosquitto: a run-once step seeds dynsec before the broker, and the broker comes up", { + skip, timeout: 1_500_000, +}, async () => { + // mosquitto's dynsec config: the plugin refuses to start unless dynamic-security.json holds an + // admin client, so the store MUST be seeded first. The `run-once` bootstrap container is declared + // BEFORE `server` (the broker) and reuses the module's runtime image; the host runs it to + // completion, then starts the broker. The runtime `server`/`runtime` shape mirrors the committed + // manifest, with images pinned to what this scenario serves by digest. + const mosquittoConf = + "persistence true\n" + + "persistence_location /mosquitto/data\n\n" + + "log_dest stdout\n" + + "log_type warning\n" + + "log_type error\n" + + "log_type notice\n\n" + + "# Every client authenticates; identities and their per-topic ACLs are managed\n" + + "# at runtime by the dynamic security plugin, whose store the plugin itself owns.\n" + + "allow_anonymous false\n" + + "plugin /usr/lib/mosquitto_dynamic_security.so\n" + + "plugin_opt_config_file /mosquitto/data/dynamic-security.json\n\n" + + "# MQTT listener\n" + + "listener 1883\n\n" + + "# MQTT-over-WebSockets listener\n" + + "listener 8081\n" + + "protocol websockets\n"; + + const manifest = JSON.stringify({ + module: "mosquitto", + version: "1", + provides: [{ name: "mqtt-topic", scope: "mesh" }], + serves: { "mqtt-topic": {} }, + emits: ["module.mosquitto.topic.provisioned", "module.mosquitto.topic.deprovisioned"], + // The events entrypoint subscribes to its own lifecycle events (an audit log), so it consumes + // them too — declared, or the substrate never makes the queue the runtime binds (ADR 0046). + consumes: ["module.mosquitto.topic.provisioned", "module.mosquitto.topic.deprovisioned"], + receives: { "mqtt-topic": "/var/lib/mosquitto-module/grants/mesh.json" }, + grants: { "mqtt-topic": "/var/lib/mosquitto-module/grants" }, + "own-secrets": { + admin: "/var/lib/mosquitto-module/admin.secret", + broker: "/var/lib/mesh/mosquitto/broker", + }, + resources: [ + { id: "mesh-state", type: "directory", path: "/var/lib/mesh/mosquitto", mode: "0700" }, + { id: "state", type: "directory", path: "/var/lib/mosquitto-module", mode: "0700" }, + { id: "grants-dir", type: "directory", path: "/var/lib/mosquitto-module/grants", mode: "0700" }, + // The broker runs as uid 1883, so the shared data directory it seeds into and persists to is + // its own. + { id: "data", type: "directory", path: "/services/mosquitto/data", mode: "0700", owner: "1883:1883" }, + { + id: "server-conf", type: "file", path: "/var/lib/mosquitto-module/mosquitto.conf", + mode: "0600", owner: "1883:1883", content: mosquittoConf, + }, + { id: "net", type: "network", name: "mosquitto" }, + // THE run-once step: seed dynsec offline, once, before the broker. It reuses the runtime image + // (`mesh-tools run ` imports mosquitto's bootstrap entrypoint, which writes the + // admin client into dynamic-security.json and chowns it to the broker's uid, then exits). It is + // declared BEFORE `server`; the host runs it to completion and requires exit 0 before starting + // the broker. + { + id: "bootstrap", type: "container", name: "mosquitto-bootstrap", + image: pinned("mesh-runtime-mosquitto"), "run-once": true, + volumes: [ + "/services/mosquitto/data:/mosquitto/data", + "/var/lib/mosquitto-module/admin.secret:/run/secrets/admin:ro", + ], + env: { + MESH_PROVISION_MQTT: "mosquitto:1883", + MESH_PROVISION_ADMIN_USER: "mesh-admin", + MESH_PROVISION_PASSWORD_FILE: "/run/secrets/admin", + MESH_DYNSEC_FILE: "/mosquitto/data/dynamic-security.json", + }, + args: ["run", "/app/modules/mosquitto/dist/bootstrap/index.js"], + }, + { + id: "server", type: "container", name: "mosquitto", image: pinned("eclipse-mosquitto"), + network: "mosquitto", ports: ["1883", "8081"], + volumes: [ + "/services/mosquitto/data:/mosquitto/data", + "/var/lib/mosquitto-module/mosquitto.conf:/mosquitto/config/mosquitto.conf:ro", + ], + }, + { + id: "runtime", type: "container", name: "mesh-mosquitto", + image: pinned("mesh-runtime-mosquitto"), network: "mosquitto", + volumes: [ + "/var/lib/mesh/mosquitto/broker:/run/secrets/broker:ro", + "/var/lib/mosquitto-module/grants:/var/lib/mosquitto-module/grants:ro", + "/var/lib/mosquitto-module/admin.secret:/run/secrets/admin:ro", + ], + env: { + MESH_BROKER_FILE: "/run/secrets/broker", + MESH_RECEIVES: "/var/lib/mosquitto-module/grants/mesh.json", + MESH_PROVISION_MQTT: "mosquitto:1883", + MESH_PROVISION_ADMIN_USER: "mesh-admin", + MESH_PROVISION_PASSWORD_FILE: "/run/secrets/admin", + }, + }, + ], + }); + + await must(`printf %s ${quote(manifest)} > /tmp/mosquitto.json && docker cp /tmp/mosquitto.json mesh-control:/mosquitto.json`); + await mesh("module add /mosquitto.json"); + const issued = await mesh(`module issue mosquitto --node ${MACHINE}`); + assert.match(issued, /scoped to what it emits and consumes/, issued); + await mesh(`assign ${MACHINE} mosquitto`); + + // ONE push, ONE convergence — and reaching `current` requires the run-once step to have completed, + // because the apply is gated on it and stops if it fails. + await mesh(`push ${MACHINE}`); + await settled(); + + // --- the run-once step ran to completion, and is NOT left running ------------------------------- + // A step is run in the foreground and removed once it exits 0; a service is left running. So the + // bootstrap container must be gone from `docker ps -a`, not lingering. + const allContainers = await must(`docker ps -a --format '{{.Names}}'`); + assert.doesNotMatch(allContainers, /(^|\n)mosquitto-bootstrap(\n|$)/, + `the run-once bootstrap was left as a container instead of run to completion:\n${allContainers}`); + + // --- the seed is on disk, owned by the broker's uid so the broker can read AND persist it -------- + const seedPath = "/services/mosquitto/data/dynamic-security.json"; + const seedExists = await on(`test -f ${seedPath}`); + assert.ok(seedExists.ok, + `the run-once step did not seed ${seedPath}:\n${(await on(`tail -40 /var/log/mesh-host.log`)).out}`); + const owner = (await must(`stat -c '%u:%g' ${seedPath}`)).trim(); + assert.equal(owner, "1883:1883", + `the seeded store is not owned by the broker's uid, so the broker cannot read/persist it: ${owner}`); + const seed = await must(`cat ${seedPath}`); + assert.match(seed, /mesh-admin/, `the seed does not hold the dynsec admin client:\n${seed}`); + + // --- the broker is up and STABLE — the proof the seed was loaded -------------------------------- + // An unseeded store crash-loops the broker: dynsec aborts start. So a broker that is up now, and + // still up after a moment with no restart, is the end-to-end proof the store was seeded before it + // started. + const running = await must(`docker ps --format '{{.Names}}'`); + for (const name of ["mosquitto", "mesh-mosquitto"]) { + assert.match(running, new RegExp(`(^|\\n)${name}(\\n|$)`), + `${name} is not running after the push:\n${running}\n---host log---\n${(await on(`tail -40 /var/log/mesh-host.log`)).out}`); + } + await new Promise((r) => setTimeout(r, 8000)); + const brokerState = (await must(`docker inspect -f '{{.State.Running}} {{.RestartCount}}' mosquitto`)).trim(); + assert.equal(brokerState, "true 0", + `the broker is not up and stable — it crash-loops against an unseeded store:\n${brokerState}\n` + + `${(await on(`docker logs mosquitto 2>&1 | tail -30`)).out}`); + const brokerLog = (await on(`docker logs mosquitto 2>&1`)).out; + assert.doesNotMatch(brokerLog, /Error found at|Unable to load|unable to load Dynamic security/i, + `the broker reported a dynsec load error against the seeded store:\n${brokerLog}`); + + // --- the scoped broker account exists ------------------------------------------------------------ + const users = await must(`docker exec mesh-broker lavinmqctl list_users 2>&1`); + assert.match(users, /anchor-mosquitto/, `the scoped account is not on the broker:\n${users}`); + + // --- the seeded admin authenticates over MQTT: the plugin loaded the seed ------------------------ + const adminPw = (await must(`cat /var/lib/mosquitto-module/admin.secret`)).trim(); + let adminClients = { out: "", ok: false }; + const untilAdmin = Date.now() + 60_000; + while (Date.now() < untilAdmin) { + adminClients = await on( + `docker exec mosquitto mosquitto_ctrl -h 127.0.0.1 -p 1883 -u mesh-admin -P ${quote(adminPw)} dynsec listClients 2>&1`, + ); + if (adminClients.ok && /mesh-admin/.test(adminClients.out)) break; + await new Promise((r) => setTimeout(r, 3000)); + } + assert.match(adminClients.out, /mesh-admin/, + `the seeded admin could not authenticate / dynsec did not load the seed:\n${adminClients.out}`); + + // --- the provider path: a consumer the mesh delivered gets a scoped client that authenticates ---- + // mosquitto's provisioner reconciles the contributions file the mesh writes at MESH_RECEIVES: for + // each consumer it reads the login the mesh derived and the mesh-minted password the host unsealed, + // and creates exactly that dynsec client, scoped to its own topic subtree (ADR 0048). A + // hand-written contributions file + secret stand in for the control plane's write; their SHAPE is + // what mesh-control produces. The proof is authentication as the consumer with the mesh's password + // — a provisioner that invented its own would refuse the connection. + const consumerPw = "mesh-minted-mqtt-7b2e1a"; + await must(`printf %s ${quote(consumerPw)} > /var/lib/mosquitto-module/grants/app.secret`); + const contributions = JSON.stringify({ + contributions: 1, + requirement: "mqtt-topic", + generated: "by the mesh — do not edit", + given: [ + { as: "app-one", node: "app-node", at: "192.0.2.20:1883", secret: "/var/lib/mosquitto-module/grants/app.secret", values: {} }, + ], + }); + await must(`printf %s ${quote(contributions)} > /var/lib/mosquitto-module/grants/mesh.json`); + + let created = { out: "", ok: false }; + const untilCreated = Date.now() + 90_000; + while (Date.now() < untilCreated) { + created = await on( + `docker exec mosquitto mosquitto_ctrl -h 127.0.0.1 -p 1883 -u mesh-admin -P ${quote(adminPw)} dynsec listClients 2>&1`, + ); + if (created.ok && /app-one/.test(created.out)) break; + await new Promise((r) => setTimeout(r, 3000)); + } + assert.match(created.out, /app-one/, + `mosquitto's provisioner never created the consumer's scoped client from the delivered contribution:\n` + + `${(await on(`docker logs mesh-mosquitto 2>&1 | tail -30`)).out}\n---\n${created.out}`); + + // Authenticate as that consumer with the password the MESH minted, and publish to its own subtree. + // A successful connect+publish (exit 0) means the provisioner created the client with exactly that + // password; a wrong password would refuse the connection. + const published = await on( + `docker exec mosquitto mosquitto_pub -h 127.0.0.1 -p 1883 -u app-one -P ${quote(consumerPw)} -q 1 -t app-one/hello -m hi 2>&1`, + ); + assert.ok(published.ok, + `the consumer could not authenticate as app-one with the mesh-minted password and publish to its subtree:\n${published.out}`); +});