[Test] UI - Unit tests: raise global vitest timeout and remove per-test overrides

[yuneng-berri]

Summary

CI is flaky because UI unit tests mix a tight 10s global timeout with a sprinkling of per-test overrides (5s–30s) and per-waitFor overrides.

This PR:

• Raises testTimeout in ui/litellm-dashboard/vitest.config.ts from 10000ms to 30000ms.
• Removes all per-it/per-describe timeout arguments across UI test files so every test uses the single global value.
• Groups CreateUserButton and TeamInfo tests (the two most flaky suites) under nested describe blocks by area (rendering, submission, tabs, settings, etc.) to make failures easier to localize without splitting the files.

waitFor/findBy* internal timeouts were left alone — those control internal polling, not the test-level timeout.

Changes

• vitest.config.ts: testTimeout: 10000 → testTimeout: 30000.
• Removed per-test timeout overrides in CreateUserButton.test.tsx, TeamInfo.test.tsx, OldTeams.test.tsx, create_mcp_server.test.tsx, add_model_tab.test.tsx, and ModelsAndEndpointsView.test.tsx.
• Reorganized CreateUserButton.test.tsx and team/TeamInfo.test.tsx into nested describe blocks.

Testing

npm run test in ui/litellm-dashboard: 380 test files, 3750 tests, all passing.

Type

✅ Test

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
litellm	Ready	Preview, Comment	Apr 10, 2026 4:16am

[codspeed-hq]

Merging this PR will not alter performance

✅ 16 untouched benchmarks

---

_{Comparing litellm_/compassionate-shannon (c7f610c) with main (42e5583)}

[greptile-apps]

Greptile Summary

This PR reduces UI test flakiness by raising the global testTimeout in vitest.config.ts from 10 s to 30 s and removing inconsistent per-test/per-describe timeout overrides across six test files. CreateUserButton.test.tsx and TeamInfo.test.tsx are also reorganized into nested describe blocks for easier failure localization, and a few new tests are added to cover previously untested paths (modal close behavior, team-member budget display, virtual-key display).

waitFor/findBy* internal polling timeouts are deliberately left in place, which is the correct distinction to make.

Confidence Score: 5/5

Safe to merge — test-only changes that strictly improve reliability without weakening any assertions.

All changes are test housekeeping: raising a global timeout, removing inconsistent per-test overrides, and reorganizing describe blocks. No assertions are weakened, new tests add coverage, and internal waitFor timeouts are correctly preserved.

No files require special attention.

Important Files Changed

Filename	Overview
ui/litellm-dashboard/vitest.config.ts	Raised global testTimeout from 10 000 ms to 30 000 ms; no other changes.
ui/litellm-dashboard/src/components/CreateUserButton.test.tsx	Removed describe-level timeout override, reorganized tests into nested describe blocks, added 'should close modal when cancel is clicked' test — no assertions weakened.
ui/litellm-dashboard/src/components/team/TeamInfo.test.tsx	Reorganized into nested describe blocks and added new rendering/display tests; all original assertions preserved.
ui/litellm-dashboard/src/app/(dashboard)/models-and-endpoints/ModelsAndEndpointsView.test.tsx	Removed four per-test 15 000 ms overrides; internal findByText timeout (10 000 ms) correctly retained.
ui/litellm-dashboard/src/components/OldTeams.test.tsx	Removed single per-test { timeout: 30000 } override that was redundant with the new global value.
ui/litellm-dashboard/src/components/add_model/add_model_tab.test.tsx	Per-test timeout overrides removed; one waitFor internal timeout (10 000 ms) correctly left in place at line 270.
ui/litellm-dashboard/src/components/mcp_tools/create_mcp_server.test.tsx	Per-test timeout overrides removed; no assertions changed.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[Vitest global testTimeout
30 000 ms] --> B{Per-test override?}
    B -- "Yes (before PR)" --> C[Use per-test timeout
5 000–30 000 ms]
    B -- "No (after PR)" --> D[Use global timeout
30 000 ms]
    C --> E[Inconsistent,
causes flakiness]
    D --> F[Consistent across
all test suites]
    F --> G{waitFor / findBy*
internal timeout?}
    G -- "Yes (left untouched)" --> H[Internal polling bounded
e.g. 10 000 ms]
    G -- "No" --> I[Waits up to
global timeout]

Loading

_{Reviews (3): Last reviewed commit: "Merge remote-tracking branch 'origin/mai..." | Re-trigger Greptile}