José Almeida
21cc4304a6
The Docker e2e Playwright project was the natural home for any test that needs a real Docker daemon on the host — but its name implied it was only about the Docker executor. SSH e2e tests (incoming in follow-up commits) belong here too, since they use a Dockerised sshd container as the SSH executor's remote target. - Rename project: docker -> containers, widen testMatch to also include tests/ssh/*.spec.ts. - Rename env flag: KANDEV_E2E_DOCKER -> KANDEV_E2E_CONTAINERS, with the old name kept as a deprecated alias for one release so local scripts and CI configs don't break overnight. - Rename CI job: e2e-docker -> e2e-containers, switch its blob report name and the merge-job dependency. - Add apps/web/e2e/README.md explaining the project layout, what the containers project covers (Docker executor + SSH executor), why it was renamed, and how to run it locally. - Add docs/specs/ssh-executor/e2e-plan.md tracking the SSH e2e coverage we'll implement on top of this rename (real sshd container, no mocks, 17 spec files covering ~75 cases). Includes a "Deferred" section that flags arm64 as something to revisit when agentctl-linux-arm64 lands in the build pipeline.
96 lines
3.9 KiB
TypeScript
96 lines
3.9 KiB
TypeScript
import { defineConfig, devices } from "@playwright/test";
|
|
|
|
const CI = !!process.env.CI;
|
|
|
|
export default defineConfig({
|
|
testDir: "./tests",
|
|
// Single worker per process — CI uses --shard to split tests across matrix
|
|
// runners (each gets its own 4 vCPUs). Tests run serially within the worker
|
|
// because the testPage fixture does e2eReset on a shared worker-scoped
|
|
// backend before each test.
|
|
//
|
|
// Isolation strategy: office-routing-* specs are gathered into their own
|
|
// Playwright project (see below) so the worker-scoped backend env
|
|
// (KANDEV_MOCK_PROVIDERS, KANDEV_PROVIDER_FAILURES) that those specs
|
|
// restart with cannot leak into specs that count agents or read the
|
|
// topbar agent name. Each routing spec restarts the backend back to
|
|
// baseline in `afterAll` (see backend.restart() — no args = revert to
|
|
// the fixture's baseline env snapshot).
|
|
fullyParallel: false,
|
|
forbidOnly: CI,
|
|
retries: CI ? 2 : 0,
|
|
workers: 1,
|
|
timeout: 60_000,
|
|
// CI uses blob reporter for cross-shard merge-reports; local uses list.
|
|
reporter: CI ? [["blob", { outputDir: "./blob-report" }]] : "list",
|
|
outputDir: "./test-results",
|
|
|
|
use: {
|
|
trace: "on-first-retry",
|
|
screenshot: "only-on-failure",
|
|
video: "on-first-retry",
|
|
},
|
|
|
|
projects: [
|
|
{
|
|
// The office-routing-* specs call `backend.restart()` with
|
|
// KANDEV_MOCK_PROVIDERS, which permanently mutates the backend's
|
|
// env for the lifetime of the worker (and registers extra
|
|
// canonical providers). Run them in their own project so that
|
|
// pollution can never leak into specs that count agents or read
|
|
// the topbar agent name — see CLAUDE.md's note on
|
|
// KANDEV_MOCK_PROVIDERS for the underlying invariant.
|
|
name: "routing",
|
|
testMatch: /office-routing-.*\.spec\.ts/,
|
|
use: { ...devices["Desktop Chrome"] },
|
|
},
|
|
{
|
|
name: "chromium",
|
|
testIgnore: [
|
|
/mobile-.*\.spec\.ts/,
|
|
// Container-backed tests (Docker executor, SSH executor) live in the
|
|
// `containers` project and skip when Docker is not available locally.
|
|
// See apps/web/e2e/README.md for what runs there.
|
|
/docker\/.*\.spec\.ts/,
|
|
/ssh\/.*\.spec\.ts/,
|
|
/office-routing-.*\.spec\.ts/,
|
|
],
|
|
use: { ...devices["Desktop Chrome"] },
|
|
},
|
|
{
|
|
name: "mobile-chrome",
|
|
testMatch: /mobile-.*\.spec\.ts/,
|
|
use: { ...devices["Pixel 5"] },
|
|
},
|
|
{
|
|
// Real-container E2E. Opt-in: run with `playwright test --project=containers`.
|
|
// Spawns the backend with KANDEV_E2E_CONTAINERS=1 (KANDEV_E2E_DOCKER=1
|
|
// is honored as a deprecated alias for one release). Builds the
|
|
// kandev-agent:e2e and kandev-sshd:e2e images and skips entirely on
|
|
// hosts without a Docker daemon — Docker is used as the runtime for
|
|
// both the Docker executor's own containers AND the sshd target the
|
|
// SSH executor connects to. Container-bound tests are slow (~10-30s
|
|
// each) so they live in their own project to keep the default CI fast.
|
|
//
|
|
// See apps/web/e2e/README.md for context and how to run locally.
|
|
name: "containers",
|
|
testMatch: [/docker\/.*\.spec\.ts/, /ssh\/.*\.spec\.ts/],
|
|
use: { ...devices["Desktop Chrome"] },
|
|
timeout: 180_000,
|
|
// Per-test sharding: CI runs `--shard=N/6` to split this project's
|
|
// tests across runners. Playwright only shards at the test level
|
|
// when fullyParallel is true (otherwise sharding is by file, and
|
|
// since this project has a single spec file every test would land
|
|
// in shard 1). Each shard is its own process with its own backend,
|
|
// and workers:1 still serializes tests within a shard, so this is
|
|
// safe.
|
|
fullyParallel: true,
|
|
},
|
|
],
|
|
|
|
// No webServer — each Playwright worker spawns its own frontend
|
|
// via the backend fixture (see fixtures/backend.ts)
|
|
|
|
globalSetup: "./global-setup.ts",
|
|
});
|