Skip to content

Self-hosted, Self-Improving, Federated Everything App — dev management, agent orchestration, persistent mind, digital twin, second brain, and creative pipeline from a single dashboard over Tailscale to port the user into the OS

test: give the watch-config fixture readiness headroom under load (#12834)

Adam Eivybc374dbaacf2
diff --git a/client/src/test/vitestTempRoot.test.js b/client/src/test/vitestTempRoot.test.js
index 2cea63651d..10558d977d 100644
--- a/client/src/test/vitestTempRoot.test.js
+++ b/client/src/test/vitestTempRoot.test.js
@@ -69,5 +69,5 @@ describe('real client runner temp lifecycle', () => {
} finally {
rmSync(host, { recursive: true, force: true });
}
- }, 90000);
+ }, 150000); // must exceed the fixture's readiness bound (WATCH_READY_MS)
});
diff --git a/scripts/lib/vitestTempRootFixture.js b/scripts/lib/vitestTempRootFixture.js
index 5d1e151eac..c49642ac70 100644
--- a/scripts/lib/vitestTempRootFixture.js
+++ b/scripts/lib/vitestTempRootFixture.js
@@ -106,7 +106,10 @@ it('controlled sibling failure', async ({ signal }) => {
`;
}
-const WATCH_READY_MS = 60000;
+// Measured: the whole watch test takes ~30s idle and ~50s under 2x CPU oversubscription
+// (a loaded full client run), so 60s left no headroom (#12834). Readiness is event-driven;
+// this only bounds a hang.
+const WATCH_READY_MS = 120000;
const FIXTURE_RUN_CEILING_MS = 120000;
// `sweepGroup` also signals after the leader exited: its detached group can