From 1a573ac3c7e92131a81aac659bb77741ae17b16d Mon Sep 17 00:00:00 2001 From: Dan Knauss Date: Mon, 22 Jun 2026 13:01:20 -0600 Subject: [PATCH] test(e2e): retry Playground smoke tests in CI to absorb boot flakiness MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The Playwright Playground smoke job intermittently fails when a single test times out in ensurePluginActivated waiting for the wp-admin Plugins heading — a transient slow-boot of the @wp-playground/cli@latest server, not a real regression (8/9 tests pass, and tests after the timeout pass). With no retries configured, one such hiccup failed the whole job, which hit main and several branches in the same window. Add retries: 2 in CI (0 locally) so flaky single-test timeouts self-heal. Tests run against an ephemeral WordPress instance, so retries are safe and idempotent. Consistent with the existing editor-not-ready race guard (d961392). Co-Authored-By: Claude Opus 4.8 --- playwright.config.js | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/playwright.config.js b/playwright.config.js index 808705d..0bc91de 100644 --- a/playwright.config.js +++ b/playwright.config.js @@ -4,6 +4,10 @@ module.exports = defineConfig({ testDir: './tests/e2e', timeout: 30_000, workers: 1, + // Retry in CI so a single transient Playground hiccup (e.g. a slow + // wp-admin boot on a loaded runner) self-heals instead of failing the + // whole smoke job. Runs against an ephemeral WordPress, so retries are safe. + retries: process.env.CI ? 2 : 0, expect: { timeout: 30_000, },