Testing Meteor 3 apps

meteortesting:mocha is the canonical Meteor test driver. It boots the app in test mode (no app code runs except the test files) and reports results in the server console.

If an existing test fails, hangs, or flakes and the failing layer is unknown, use meteor-debugging to isolate the cause. Return here when evidence points to test setup, design, driver behavior, fixtures, or regression coverage.

Decision flow

1. Unknown failure in an existing test? Use meteor-debugging to classify application, data, harness, browser, environment, or shared-state causes.
2. Pure logic, no Meteor APIs? Plain Mocha or any test runner; nothing Meteor-specific.
3. Method or publication? Integration test it with meteortesting:mocha using Meteor.server.method_handlers[name].apply(ctx, args) or publish_handlers.
4. Need to drive a real DDP client (cross-process or end-to-end DDP semantics)? Use --full-app mode and DDP.connect.
5. UI flow / browser interaction? Start the normal app with deterministic test settings and run Playwright or Cypress against it. Use meteor test --full-app only when the browser flow must load app-test modules.

Setup

meteor add meteortesting:mocha
meteor npm install --save-dev @types/mocha

// package.json
{
  "scripts": {
    "test": "TEST_WATCH=1 meteor test --driver-package meteortesting:mocha",
    "test:ci": "meteor test --once --driver-package meteortesting:mocha",
    "test:full": "meteor test --full-app --driver-package meteortesting:mocha"
  }
}

Test-mode conventions:

• Normal test mode eagerly loads *.test[s].* and *.spec[s].* files, while ordinary application code loads only when imported by a test.
• Filename discovery ignores tests/ directories. A configured meteor.testModule instead loads its explicit entry module and imports.
• --full-app eagerly loads the normal app plus *.app-test[s].* and *.app-spec[s].* files, and sets Meteor.isAppTest.

For Meteor 3.5.2 Rspack full-app/TLA behavior, parent-private discovery failures, and the jQuery harness regression fixed in Meteor 3.6-beta.0, read release-specific test setup. Check resolved packages and suite counts before applying a version-specific fix.

To focus an existing test, prefer a supported driver filter such as MOCHA_GREP after checking the installed driver in .meteor/versions; use .only only temporarily. These select registered tests but do not prevent other test modules from evaluating. If loading itself causes interference, inspect meteor.testModule and read references/focused-runs.md before narrowing entrypoint imports. Restore every temporary focus or import change, then run the normal affected suite. Read the configured entrypoint paths from the project; do not invent paths, scripts, ports, or helpers.

For 3.6-beta.3 Rspack ignore exceptions or separate legacy test entries, use the architecture and ignore boundaries. Check expected suites on each selected program; a modern zero-exit run is not proof of legacy coverage.

Method test (server-side)

import assert from "node:assert/strict";
import { Meteor } from "meteor/meteor";
import { Items } from "/imports/api/items";

if (Meteor.isServer) {
  describe("items.add", function () {
    beforeEach(async function () {
      await Items.removeAsync({});
    });

    it("inserts when authed", async function () {
      const _id = await Meteor.server.method_handlers["items.add"].apply(
        { userId: "u1" },
        [{ title: "x", qty: 1 }],
      );
      const doc = await Items.findOneAsync(_id);
      assert.equal(doc.title, "x");
    });

    it("rejects when unauthed", async function () {
      await assert.rejects(() =>
        Meteor.server.method_handlers["items.add"].apply({ userId: null }, [{}]),
      );
    });
  });
}

.apply(context, argsArray) is the documented form; Meteor.server.method_handlers[name] is the registered method implementation. Pass a stub this with userId (and unblock, connection, etc. when needed).

Publication test (server-side)

if (Meteor.isServer) {
  describe("items.mine publication", function () {
    it("only ships the subscribing user's items", async function () {
      await Items.removeAsync({});
      await Items.insertAsync({ _id: "a", ownerId: "u1" });
      await Items.insertAsync({ _id: "b", ownerId: "u2" });

      const cursor = await Meteor.server.publish_handlers["items.mine"]
        .apply({ userId: "u1" }, []);
      const docs = await cursor.fetchAsync();
      assert.deepEqual(docs.map((d) => d._id), ["a"]);
    });
  });
}

For publications using the low-level this.added / this.changed API, test with a real DDP client via --full-app mode (next section).

Always await the direct publish handler. Conventional handlers return the cursor immediately; async handlers return a Promise of the cursor.

End-to-end DDP test (--full-app mode)

import { DDP } from "meteor/ddp-client";
import { Mongo } from "meteor/mongo";
import { Tracker } from "meteor/tracker";

const conn = DDP.connect(Meteor.absoluteUrl());
const RemoteItems = new Mongo.Collection("items", { connection: conn });

function ready(sub) {
  return new Promise((resolve) => {
    const computation = Tracker.autorun((c) => {
      if (sub.ready()) {
        c.stop();
        resolve();
      }
    });
  });
}

const sub = conn.subscribe("items.mine");
await ready(sub);
const docs = RemoteItems.find().fetch();

Run with meteor test --full-app --driver-package meteortesting:mocha.

Client unit test (Minimongo)

if (Meteor.isClient) {
  describe("Items minimongo", function () {
    beforeEach(function () { Items.remove({}); });

    it("filters by ownerId", function () {
      Items.insert({ _id: "a", ownerId: "u1" });
      Items.insert({ _id: "b", ownerId: "u2" });
      assert.equal(Items.find({ ownerId: "u1" }).count(), 1);
    });
  });
}

Client tests run in the browser the driver spawns. With TEST_BROWSER_DRIVER=puppeteer, the browser is headless Chromium and results flow back to the server console.

Do not treat a server-only green run as complete client coverage. The driver prints a message when no browser is connected; configure TEST_BROWSER_DRIVER in CI and confirm client test counts are present.

Browser E2E

Run Playwright or Cypress against the normal application unless the test needs Meteor's app-test modules. The server command must work from a clean clone with a tracked, nonsecret settings fixture or fully documented test environment variables. Do not point it at an ignored developer settings file.

Cover at least one complete mutation flow and one rejected server action. Basic page-title checks do not exercise async method stubs, authorization, or publication readiness.

Anti-patterns

• Reach for Jest. Jest does not understand Meteor's build system. Use meteortesting:mocha.
• Mock Mongo. Run the real driver against a clean DB.
• Forget if (Meteor.isServer) { ... } guards. Server tests crash if they run in the browser harness.
• Skip beforeEach cleanup. Tests leak state across runs.
• Catch and log a beforeEach failure without rethrowing. The test then runs against unknown state and can report a misleading assertion.
• Assert only the database postcondition for a rejected method. Also require the Promise to reject with the expected Meteor.Error.
• Replace Meteor.userId or another global without restoring it in afterEach.
• Use Meteor.call with callbacks in async test code. Use Meteor.callAsync or unwrap method_handlers directly.

See also

• references/mocha-setup.md
• references/focused-runs.md
• references/ddp-test-helpers.md
• references/eval-cases.md