mesh-lab/test/integration/built-store-cross-node.test.ts

/**
 * THE NO-FAKE MULTI-NODE GATE: two machines, everything built by the mesh itself.
 *
 * The rewrite-based multi-node beds pre-stock runtime images and rewrite each manifest's
 * placeholder digest to whatever they stocked — bed code doing the builder's job, which means the
 * builder's job was never under test. This bed removes the shortcut. The scenario names NO images;
 * the committed manifests are registered VERBATIM; the only thing that ever turns
 * `mesh-runtime-<m>@sha256:0000…` (or an `artifact:` reference) into a real digest is the mesh's
 * own builder, exactly as in production.
 *
 * What it proves, end to end:
 *   1. The installer raises `anchor` into a mesh of one — building the control plane (ADR 0073),
 *      standing up the registry and the builder.
 *   2. The mesh BUILDS the shared base, then postgres and lavinmq, from the forge — and adopts the
 *      foundation's own store and broker as those modules (ADR 0078), the store's genesis superuser
 *      carried in through `secret accept` (the same act as hq phase3 deliverSuperuser).
 *   3. `node2` joins, and its consumers are BUILT and delivered the same way: `amqp-ping` opens the
 *      one broker and `letta` gets a database on the one store — both across the overlay, admitted
 *      by the firewall the nftables module derives (issues 055/056/057 in one bed).
 *
 *   MESH_LAB_HOST_BINARY=.../mesh-host   MESH_LAB_BOOTSTRAP_BINARY=.../mesh-bootstrap
 *   MESH_LAB_BUNDLE=.../examples/foundation-first-node.lock  (the TEMPLATE)
 *   MESH_LAB_CATALOG=.../mesh-catalog/modules
 *   MESH_LAB_SOURCE=git://<forge>/mesh-controller.git   MESH_LAB_SOURCE_REF=<commit>
 *   MESH_LAB_BUILD_REF=<branch or commit for module builds, default main>
 */
import { test, before, after } from "node:test";
import assert from "node:assert/strict";
import { existsSync } from "node:fs";
import { resolve } from "node:path";
import { loadScenario } from "../../src/declaration/parse.ts";
import { raise } from "../../src/lifecycle/raise.ts";
import { destroy, exec, push } from "../../src/lifecycle/operate.ts";
import { bootstrapBinaryPath, hostBinaryPath, HOST_PATH } from "../../src/lifecycle/place.ts";
import { labIsUsable, destroyAll, foundationBundle } from "./harness.ts";
import { genesis, type GenesisResult } from "./genesis.ts";

const SCENARIO = "built-store-cross-node";
const CONTROL = "anchor";
const NODE = "node2";
const ANCHOR = "192.0.2.10";
const REGISTRY = `${ANCHOR}:5000`;
const BASE = { module: "mesh-tools", repo: "mesh-tools", path: "" };

const capability = await labIsUsable();
const binary = hostBinaryPath();
const installer = bootstrapBinaryPath();
const bundle = process.env["MESH_LAB_BUNDLE"] ?? "";
const catalogDir = process.env["MESH_LAB_CATALOG"] ?? "";
const source = process.env["MESH_LAB_SOURCE"] ?? "";
const sourceRef = process.env["MESH_LAB_SOURCE_REF"] ?? "";
const KEEP = !!process.env["MESH_LAB_KEEP"];

function forgeUrl(repo: string): string {
  const override = process.env[`MESH_LAB_SOURCE_${repo.toUpperCase().replaceAll("-", "_")}`];
  if (override) return override;
  return source.replace(/[^/]+\.git$/, `${repo}.git`);
}
function refFor(repo: string): string {
  const override = process.env[`MESH_LAB_BUILD_REF_${repo.toUpperCase().replaceAll("-", "_")}`];
  return override ?? process.env["MESH_LAB_BUILD_REF"] ?? "main";
}
const baseManifest = process.env["MESH_LAB_BASE_MANIFEST"] ??
  resolve(catalogDir, "..", "..", BASE.repo, "module.json");

const skip =
  !capability.usable ? capability.why :
  !binary ? "MESH_LAB_HOST_BINARY is not set to a built mesh-host" :
  !installer ? "MESH_LAB_BOOTSTRAP_BINARY is not set to a built mesh-bootstrap" :
  !source ? "MESH_LAB_SOURCE is not set to the repository the control plane is built from" :
  !sourceRef ? "MESH_LAB_SOURCE_REF is not set to the commit to build" :
  !bundle || !existsSync(bundle) ? "MESH_LAB_BUNDLE is not set to a foundation template" :
  !catalogDir || !existsSync(catalogDir) ? "MESH_LAB_CATALOG is not set to mesh-catalog/modules" :
  false;

let instanceId = "";
let raised: GenesisResult;

function quote(s: string): string { return `'${s.replaceAll("'", `'\\''`)}'`; }

async function on(machine: string, 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(machine: string, command: string, timeoutMs?: number): Promise<string> {
  const { out, ok } = await on(machine, command, timeoutMs);
  if (!ok) throw new Error(`${machine}: ${command}\n${out}`);
  return out;
}
/** The control plane — retried through the brief recreate window a spec change causes. */
async function mesh(command: string, timeoutMs?: number): Promise<string> {
  const deadline = Date.now() + (timeoutMs ?? 120_000);
  for (;;) {
    const got = await on(CONTROL, `docker exec mesh-controller /mesh-controller ${command}`, timeoutMs);
    if (got.ok) return got.out;
    if (!/is not running|No such container/.test(got.out) || Date.now() > deadline) {
      throw new Error(`${CONTROL}: mesh ${command}\n${got.out}`);
    }
    await new Promise((r) => setTimeout(r, 5_000));
  }
}
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;
}

/** Register a committed manifest VERBATIM — the point of this bed: no rewriting, ever. */
async function registerModule(module: string, manifest: string): Promise<string> {
  assert.ok(existsSync(manifest), `no manifest for ${module} at ${manifest}`);
  const onMachine = `/tmp/${module}.json`;
  await push(instanceId, CONTROL, manifest, onMachine);
  await must(CONTROL, `docker cp ${onMachine} mesh-controller:/${module}.json`);
  return mesh(`module add /${module}.json`);
}

/** Build a module with the mesh's own builder, issue its account, and assign it to a node. */
async function buildAndAssign(module: string, node: string, opts?: { build?: boolean }): Promise<void> {
  await registerModule(module, resolve(catalogDir, module, "module.json"));
  if (opts?.build !== false) {
    const built = await mesh(
      `build ${forgeUrl("mesh-catalog")} --path modules/${module} --ref ${refFor("mesh-catalog")} --wait 1200s`,
      1_500_000);
    assert.doesNotMatch(built, /failed/i, built);
  }
  const manifest = (await on(CONTROL, `docker exec mesh-controller cat /${module}.json`)).out;
  if (manifest.includes("MESH_BROKER_FILE")) {
    const issued = await mesh(`module issue ${module} --node ${node}`);
    assert.match(issued, /scoped to what it (emits and consumes|consumes and emits)/,
      `the broker account for ${module} was not issued:\n${issued}`);
  }
  await mesh(`assign ${node} ${module}`);
  await mesh(`push ${node}`, 600_000);
}

async function waitForContainer(node: string, container: string, seconds = 300): Promise<string> {
  const deadline = Date.now() + seconds * 1_000;
  let last = "";
  while (Date.now() < deadline) {
    const ps = (await on(node, `docker ps -a --format '{{.Names}}\t{{.Status}}'`)).out;
    last = ps;
    const line = ps.split("\n").find((l) => l.split("\t")[0]?.trim() === container);
    if (line && /^Up /.test(line.split("\t")[1]?.trim() ?? "")) return ps;
    await new Promise((r) => setTimeout(r, 5_000));
  }
  const logs = (await on(node, `docker logs --tail 20 ${container} 2>&1`)).out;
  throw new Error(`${container} never came up on ${node}:\n${last}\n--- logs ---\n${logs}`);
}

before(async () => {
  if (skip) return;
  const bed = await raise(loadScenario(`scenarios/${SCENARIO}.yml`), {
    onProgress: (m) => console.log(`raise: ${m}`),
  });
  instanceId = bed.instanceId;
  console.log(`INSTANCE ${instanceId}${KEEP ? " (KEEP)" : ""}`);
  console.log("NOTHING WAS LOADED: this scenario names no images; the mesh builds or pulls everything.");

  // Genesis: the installer raises anchor into a mesh of one, BUILDING the control plane, and
  // leaves the anchor enrolled with an agent running (hostService).
  raised = await genesis({
    instanceId,
    node: CONTROL,
    installer: installer as string,
    catalogDir,
    bundleTemplate: foundationBundle(bundle, []),
    registry: REGISTRY,
    source,
    sourceRef,
    toolsSource: forgeUrl(BASE.repo),
    toolsRef: refFor(BASE.repo),
    catalogSource: forgeUrl("mesh-catalog"),
    catalogRef: refFor("mesh-catalog"),
    sdkSource: forgeUrl("mesh-sdk"),
    sdkRef: refFor("mesh-sdk"),
    site: "hosting",
    hostService: true,
    hostBinary: binary,
    log: (m) => console.log(m),
  });
  if (!raised.ok) throw new Error(`${raised.step || "genesis"}: ${raised.why}\n\n${raised.report.join("\n")}`);

  // node2 joins the mesh: a token against the anchor's public broker endpoint, then an agent.
  await mesh(`node add ${NODE}`);
  const token = tokenFrom(await mesh(`token issue --node ${NODE}`));
  const said = await must(NODE, `${HOST_PATH} enrol --token ${quote(token)}`);
  assert.match(said, new RegExp(`enrolled as ${NODE}`), said);
  await must(NODE, `nohup ${HOST_PATH} run > /var/log/mesh-host.log 2>&1 & sleep 3`);
}, { timeout: 3_000_000 });

after(async () => {
  if (KEEP) { console.log(`MESH_LAB_KEEP set — leaving ${instanceId} standing`); return; }
  if (instanceId) await destroy(instanceId);
  await destroyAll(`${SCENARIO}-`);
}, { timeout: 900_000 });

test("a joined node's consumers open the store and broker the mesh built and adopted, over the overlay", {
  skip, timeout: 3_600_000,
}, async () => {
  // The overlay, so bindings carry `.internal` names; the firewall, so from:mesh is what admits them.
  await mesh(`overlay place ${CONTROL} --hub --endpoint ${ANCHOR}:51820 --site lab`);
  await mesh(`overlay place ${NODE} --site lab`);
  await mesh(`assign ${CONTROL} networking`);
  await mesh(`assign ${NODE} networking`);

  // The shared base first — every module with code of its own stands on it.
  await registerModule(BASE.module, baseManifest);
  const base = await mesh(`build ${forgeUrl(BASE.repo)} --ref ${refFor(BASE.repo)} --wait 1200s`, 1_500_000);
  assert.doesNotMatch(base, /failed/i, base);

  // Genesis already adopted the store as the postgres module (superuser accepted), and installed
  // nftables and the catalogue — verified rather than redone. The broker is the one foundation
  // half genesis leaves to the mesh proper: adopt it here, BUILT by the mesh's own builder.
  await waitForContainer(CONTROL, "mesh-postgres", 120);
  await buildAndAssign("lavinmq", CONTROL);
  await waitForContainer(CONTROL, "mesh-lavinmq", 600);

  // The consumers on the joined node — built by the mesh, delivered from its registry.
  await buildAndAssign("amqp-ping", NODE);
  await buildAndAssign("letta", NODE);
  await mesh(`push ${NODE}`, 900_000);

  // 057: the provider node is composed again so its provisioners learn of the remote consumers.
  await mesh(`push ${CONTROL}`, 600_000);

  // THE BROKER, cross-node: amqp-ping's binding names the control-node's overlay name, its vhost is
  // minted on the one broker, and it holds the connection.
  await waitForContainer(NODE, "amqp-ping", 600);
  const amqpBound = (await on(NODE, `cat /var/lib/amqp-ping/amqp.json 2>&1`)).out;
  assert.match(amqpBound, new RegExp(`${CONTROL}\\.internal`),
    `the amqp grant does not point at the control-node over the overlay:\n${amqpBound}`);
  {
    const deadline = Date.now() + 240_000;
    let vhosts = "";
    while (Date.now() < deadline) {
      vhosts = (await on(CONTROL, `docker exec mesh-broker lavinmqctl list_vhosts 2>&1`)).out;
      if (new RegExp(`${NODE}|amqp-ping`).test(vhosts)) break;
      await new Promise((r) => setTimeout(r, 5_000));
    }
    assert.match(vhosts, new RegExp(`${NODE}|amqp-ping`),
      `no vhost was minted on the one broker for the joined consumer:\n${vhosts}\n` +
      `--- provisioner ---\n${(await on(CONTROL, `docker logs mesh-lavinmq 2>&1 | tail -20`)).out}\n` +
      `--- consumer ---\n${(await on(NODE, `docker logs amqp-ping 2>&1 | tail -20`)).out}`);
  }

  // THE STORE, cross-node: letta's binding names the control-node, and the login the mesh derived
  // authenticates on the one store with the password the provisioner minted. (letta's own app needs
  // pgvector and is not the claim here — the credential reaching a real database is.)
  const bound = await (async () => {
    const deadline = Date.now() + 240_000;
    let raw = "";
    while (Date.now() < deadline) {
      const got = await on(NODE, `cat /var/lib/letta/database.json 2>/dev/null`);
      if (got.ok && /"as"/.test(got.out)) { raw = got.out; break; }
      await new Promise((r) => setTimeout(r, 5_000));
    }
    assert.match(raw, /"as"/,
      `the mesh never wrote letta's database binding:\n${raw}\n` +
      `--- store provisioner ---\n${(await on(CONTROL, `docker logs mesh-postgres 2>&1 | tail -20`)).out}`);
    return JSON.parse(raw) as { as: string; at: string; provision: string };
  })();
  assert.equal(bound.provision, "postgres-database", `letta was bound the wrong provision: ${bound.provision}`);
  assert.match(bound.at, new RegExp(`${CONTROL}\\.internal`),
    `letta's database binding does not point at the control-node over the overlay: ${bound.at}`);
  const pw = (await must(NODE, `cat /var/lib/letta/database.secret`)).trim();
  assert.ok(bound.as && pw, `letta's login or password was empty (as=${bound.as})`);
  {
    const conn = `postgresql://${bound.as}:${encodeURIComponent(pw)}@127.0.0.1:5432/${bound.as}?sslmode=disable`;
    let pg = { out: "", ok: false };
    const deadline = Date.now() + 120_000;
    while (Date.now() < deadline) {
      pg = await on(CONTROL, `docker exec mesh-postgres psql ${quote(conn)} -tAc 'select 1' 2>&1`);
      if (pg.ok && /^1$/m.test(pg.out)) break;
      if (/authentication failed/i.test(pg.out)) break;
      await new Promise((r) => setTimeout(r, 5_000));
    }
    assert.doesNotMatch(pg.out, /authentication failed/i,
      `the store does not know the password the mesh delivered letta:\n${pg.out}`);
    assert.match(pg.out, /^1$/m, `letta's login could not open its database on the one store:\n${pg.out}`);
  }

  return `amqp: ${amqpBound.trim().slice(0, 120)}…\ndb: as=${bound.as} at=${bound.at}`;
});