Skip to content

YC (S26) | Open Computer History | Continuously record your company computer work, map your workflows, help you find work worth automating, and power your agents' context

test(e2e): bump CI specFileRetries 1 -> 3 to absorb the distributed flake

Louis Beaumont46f500bdb5d5
diff --git a/apps/screenpipe-app-tauri/e2e/wdio.conf.ts b/apps/screenpipe-app-tauri/e2e/wdio.conf.ts
index 2b514be3d..5ebfd9eaa 100644
--- a/apps/screenpipe-app-tauri/e2e/wdio.conf.ts
+++ b/apps/screenpipe-app-tauri/e2e/wdio.conf.ts
@@ -51,10 +51,15 @@ export const config: TestrunnerConfig = {
waitforTimeout: isCi ? 20000 : 10000,
connectionRetryTimeout: isWindowsCi ? 180000 : 60000,
connectionRetryCount: isWindowsCi ? 12 : 3,
- // CI-only: retry a whole spec file once on transient WebDriver glitches
- // (e.g. "Session not found" on WebKit) before failing the run. Local runs
- // skip retries so flakes surface immediately during development.
- specFileRetries: isCi ? 1 : 0,
+ // CI-only: retry a whole spec file on transient WebDriver glitches
+ // (e.g. "Session not found" on WebKit) before failing the run. The suite has
+ // a distributed ~1-flaky-failure-per-run rate spread across different specs
+ // (focus-server, owned-browser, timeline, etc.), so a single retry wasn't
+ // enough and a different spec reddened E2E almost every run. Three retries
+ // make a genuine flake (which passes most of the time) very unlikely to
+ // survive, while a truly broken spec still fails every attempt.
+ // Local runs skip retries so flakes surface immediately during development.
+ specFileRetries: isCi ? 3 : 0,
specFileRetriesDelay: 5,
framework: 'mocha',
reporters: getReporters() as Options.Testrunner['reporters'],