* cleanup(webhooks): remove instant meeting trigger support
* remove dead code
* remove instant meeting from prisma enums and related cleanup
* fix(webhooks): make instant meeting enum migration atomic
The /org/[orgSlug]/[user] and /team/[slug] routes were removed in the Cal.diy fork (#28903/#28904) but their SSR helpers and Team screen component were left behind. No route, component, or test still imports them (verified with ripgrep). Deleting them removes dead code and makes it clearer that team/org pages are gone.
* docs: add self-hosting troubleshooting guide
Add a dedicated troubleshooting page (docs/self-hosting/troubleshooting.mdx)
covering the most common self-hosting issues:
- 500 error during onboarding caused by missing STRIPE_PRIVATE_KEY (#25993)
- Redirect to localhost after deployment (NEXTAUTH_URL / NEXT_PUBLIC_WEBAPP_URL) (#21921)
- API v2 service not starting in Docker (missing REDIS_URL, JWT_SECRET, WEB_APP_URL)
- CLIENT_FETCH_ERROR in Docker logs
- SSL issues behind a reverse proxy
- Prisma user creation failure on first setup
Also adds the new page to the "Getting Started" navigation group in docs/docs.json.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
* fix(docs): provide safer SSL troubleshooting alternatives
Replace the blanket NODE_TLS_REJECT_UNAUTHORIZED=0 recommendation
with three options in order of preference:
1. Use HTTP internally with proper header forwarding
2. Add internal CA to NODE_EXTRA_CA_CERTS
3. Disable TLS verification (last resort with security warning)
The previous guidance could expose users to MITM attacks on all
external API calls (Stripe, Google, etc.).
* fix(docs): correct NEXTAUTH_URL guidance to prevent OAuth breakage
The previous guidance recommended setting NEXTAUTH_URL to localhost
for SSL/DNS issues, which breaks OAuth callbacks since external
providers would redirect to localhost instead of the public domain.
- Replace localhost workaround with extra_hosts in docker-compose
- Add nginx proxy header configuration example
- Add warnings explaining why localhost breaks OAuth
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
* fix(docs): correct troubleshooting guide based on codebase verification
- Stripe section: Rewrite to accurately reflect current behavior (app
gracefully handles missing keys by marking Stripe as 'not installed'
rather than crashing). Clarify that NEXT_PUBLIC_STRIPE_PUBLIC_KEY
belongs in .env.appStore.
- API v2 section: Add missing required vars (STRIPE_API_KEY,
STRIPE_WEBHOOK_SECRET, NEXTAUTH_SECRET) that crash the service if
absent. Move WEB_APP_URL to optional (it has a fallback default).
- CLIENT_FETCH_ERROR section: Add caveat about HTTPS URLs failing
with the extra_hosts approach when app listens on port 3000.
- Vercel note: Clarify that NEXTAUTH_URL is auto-inferred via
VERCEL_URL, not just 'left empty'.
- Database section: Replace unverifiable metadata/id advice with
actionable migration and setup guidance matching actual code in
apps/web/app/api/auth/setup/route.ts.
Co-Authored-By: romitgabani1 <romitgabani1.work@gmail.com>
* docs: fix inaccuracies in self-hosting troubleshooting guide
- Add missing CALENDSO_ENCRYPTION_KEY to API v2 required variables
- Fix setup endpoint path from /api/auth/setup to /auth/setup
- Add note about NEXTAUTH_URL auto-derivation from NEXT_PUBLIC_WEBAPP_URL
* fix(docs): correct NEXTAUTH_URL derivation mechanism description
NextAuth infers the base URL from the request's Host header when
NEXTAUTH_URL is not set, not from NEXT_PUBLIC_WEBAPP_URL directly.
Co-Authored-By: romitgabani1 <romitgabani1.work@gmail.com>
---------
Co-authored-by: shockzM1 <shockz@dsn.so>
Co-authored-by: Paperclip <noreply@paperclip.ing>
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Both booking-sheet-keyboard.e2e.ts and bookings-list.e2e.ts intermittently
timeout in CI when clicking the booking item button before the DOM has
finished rendering. Adding explicit waitFor({ state: 'visible' }) on the
role=button element after the parent booking item is visible ensures the
click target is fully ready.
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* fix: mock @calcom/lib/i18n in EventManager.test.ts to prevent vitest worker shutdown flake
TranslationService transitively imports @calcom/lib/i18n which triggers slow
module resolution via vite's RPC. When the vitest worker shuts down before it
completes, it causes 'Closing rpc while fetch was pending' errors.
Mocking the module prevents the actual module resolution during test loading.
Co-Authored-By: romitgabani1 <romitgabani1.work@gmail.com>
* chore: remove explanatory comments per review feedback
Co-Authored-By: romitgabani1 <romitgabani1.work@gmail.com>
---------
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
- Add await to unawaited bookingSeatsRepositoryFixture.create calls
- Clean up leftover selected slots before seated event tests
The flakiness was caused by two issues:
1. Missing await on bookingSeatsRepositoryFixture.create() - the HTTP
request to fetch slots could execute before the booking seat record
was written to the database, leading to incorrect seat counts.
2. Leftover SelectedSlots records leaking between test groups - the
availability calculation fetches all unexpired reserved slots by
userId (not eventTypeId), so reserved slots from earlier tests
appeared as busy times when computing slots for seated event types.
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* fix: stabilize flaky Team filter E2E tests in bookings-list
Backport fix from calcom/cal: replace fragile expect.poll().toBe(1)
with explicit toBeHidden() wait for filtered-out booking item followed
by toHaveCount(1). This prevents race conditions where the DOM hasn't
updated yet after the team filter API response returns.
Fixes all three Team filter tests:
- Team filter shows bookings for direct team event types
- Team filter shows bookings for managed event types (child events)
- Team filter excludes bookings from other teams
Co-Authored-By: romitgabani1 <romitgabani1.work@gmail.com>
* fix: adjust direct team event types test assertion
In cal.com, the team filter includes personal bookings of team members,
so instead of asserting the personal booking is hidden, verify that the
team booking is visible and present. The managed event types and
cross-team tests correctly use toBeHidden since those filters do
exclude the expected items.
Co-Authored-By: romitgabani1 <romitgabani1.work@gmail.com>
* fix: strengthen direct team event types assertion with filter-active check
Add assertion that the teamId filter popover trigger is visible in the UI,
proving the filter was applied before checking the team booking is present.
This addresses the concern that toBeVisible alone would pass even without
the filter being active.
Co-Authored-By: romitgabani1 <romitgabani1.work@gmail.com>
* chore: remove comments per review feedback
Co-Authored-By: romitgabani1 <romitgabani1.work@gmail.com>
---------
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* fix: resolve flaky team-management E2E test
Fix two failure modes in the 'Can create teams via Wizard' test:
1. Strict mode violation: locator('[data-testid=new-team-btn]') resolves to
2 elements during Next.js streaming/hydration. Fixed by using .first().
2. Race condition in disband assertion: raw .count() check doesn't wait for
UI to update after team deletion. Replaced with Playwright's auto-retrying
toBeHidden() assertion with a 10s timeout.
Co-Authored-By: romitgabani1 <romitgabani1.work@gmail.com>
* chore: remove explanatory comments per review feedback
Co-Authored-By: romitgabani1 <romitgabani1.work@gmail.com>
---------
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* fix: harden seed script org settings upsert and P2002 error handling
Co-Authored-By: romitgabani1 <romitgabani1.work@gmail.com>
* fix: remove PII from P2002 log and recover existing user instead of returning null
- Replace username interpolation in log message with generic text (Cubic violation #2, confidence 9/10)
- On P2002, fetch the existing user from DB and return it with membership data instead of returning null, which was dropping users from org setup on retries (Cubic violation #3, confidence 9/10)
Co-Authored-By: bot_apk <apk@cognition.ai>
---------
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Co-authored-by: bot_apk <apk@cognition.ai>
Co-authored-by: Alex van Andel <me@alexvanandel.com>
* fix: add missing vi.mock() calls to prevent vitest worker shutdown flakiness
Add vi.mock() calls for modules that trigger background network requests
or database connections during import. These transitive imports can cause
the vitest worker RPC to shut down while pending fetch/network operations
are still in flight, resulting in flaky test failures with:
Error: [vitest-worker]: Closing rpc while "fetch" was pending
The primary modules mocked are:
- @calcom/app-store/delegationCredential (triggers credential lookups)
- @calcom/prisma (triggers database initialization)
- @calcom/features/calendars/lib/CalendarManager (triggers calendar API calls)
- @calcom/features/auth/lib/verifyEmail (triggers email service)
- @calcom/lib/domainManager/organization (triggers domain lookups)
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* fix: remove conflicting empty prisma mocks from files with prismock/prismaMock setups
- Remove vi.mock('@calcom/prisma', () => ({ default: {}, prisma: {} })) from 28 files
that already have prismock/prismaMock test doubles. Vitest hoists all vi.mock() calls
and the last one wins, so these empty mocks were overriding the functional test doubles.
- Fix CalendarSubscriptionService.test.ts to reuse the shared mock from
__mocks__/delegationCredential instead of creating a new unconfigured vi.fn()
- Remove DelegationCredentialRepository.test.ts empty prisma mock (different pattern)
- Remove vi.mock from inside beforeEach in intentToCreateOrg.handler.test.ts
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* fix: add comprehensive delegationCredential mock exports to prevent CI test failures
The vi.mock blocks for @calcom/app-store/delegationCredential were missing
exports that the code under test transitively imports (e.g.
enrichUsersWithDelegationCredentials, enrichUserWithDelegationCredentialsIncludeServiceAccountKey,
buildAllCredentials, getFirstDelegationConferencingCredentialAppLocation).
Added all exports with passthrough implementations so the booking flow
works correctly without triggering real network requests.
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* fix: correct credential mock return shapes to match real module API
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* fix: revert unintended yarn.lock changes
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
---------
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* fix: resolve duplicate new-team-btn selector in teams E2E test
Remove TeamsCTA from the loading skeleton to prevent duplicate
data-testid='new-team-btn' elements in the DOM during Next.js
streaming SSR. The skeleton's Suspense fallback and the resolved
page content can briefly coexist, causing Playwright's strict
mode to fail when both contain the same test ID.
Also use .first() in the E2E test as a defensive measure.
Co-Authored-By: romitgabani1 <romitgabani1.work@gmail.com>
* fix: assert single new-team-btn element instead of using .first()
Address Cubic AI review feedback: use toHaveCount(1) assertion
to ensure exactly one new-team-btn exists, rather than .first()
which could mask duplicate-element regressions.
Co-Authored-By: romitgabani1 <romitgabani1.work@gmail.com>
* fix: revert test changes, only remove CTA from skeleton
Per user feedback: only change needed is removing TeamsCTA from
the loading skeleton to prevent users without a team plan from
bypassing the upgrade banner via the skeleton CTA.
Co-Authored-By: romitgabani1 <romitgabani1.work@gmail.com>
---------
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* perf: batch DB operations in performance test setup to avoid CI timeout
Replace ~5200 sequential DB inserts with batch createMany() operations
in createHugeAttributesOfTypeSingleSelect. The test was timing out at
60s on 2-vCPU CI runners because the setup phase alone (creating 400
users, 800 memberships, 4000 attribute assignments one-by-one) exceeded
the timeout before the actual logic under test even ran.
Co-Authored-By: romitgabani1 <romitgabani1.work@gmail.com>
* perf: optimize createAttributesScenario with batch DB operations
Move batch DB operations (createMany) into the shared createAttributesScenario
helper so all tests benefit, not just the performance test. Simplify
createHugeAttributesOfTypeSingleSelect to delegate to the optimized helper.
Co-Authored-By: romitgabani1 <romitgabani1.work@gmail.com>
* Revert "perf: optimize createAttributesScenario with batch DB operations"
This reverts commit c739edd3718796d61f6510ea3a9155065b69a1ce.
* Revert "perf: batch DB operations in performance test setup to avoid CI timeout"
This reverts commit ff6642ae49190c42dd8673a12ee11c8c971370b4.
* perf: optimize createAttributesScenario with batch DB operations
Replace sequential prisma.*.create() calls in the shared createAttributesScenario
helper with batch createMany() calls, reducing ~5,200 sequential DB operations to
~10 batch operations. Simplify createHugeAttributesOfTypeSingleSelect to delegate
to the optimized helper instead of duplicating batch logic.
Co-Authored-By: romitgabani1 <romitgabani1.work@gmail.com>
* fix: preserve deterministic user ordering in createAttributesScenario
Map users by email after findMany to ensure the index-based attribute
assignment matches the original userDataList order, since findMany does
not guarantee return order.
Co-Authored-By: romitgabani1 <romitgabani1.work@gmail.com>
* revert: make test timeout back to 1 min
* refactor: extract batch helpers createTestUsers, createTestMembershipsForUsers, createTestAttributeAssignments
Organize batch DB operations from createAttributesScenario into
reusable helper functions mirroring the existing single-record helpers.
Co-Authored-By: romitgabani1 <romitgabani1.work@gmail.com>
* cleanup: remove unused single-record helpers and JSDoc comments
Remove createTestUser, createTestMemberships, createTestAttributeAssignment
(now replaced by batch versions). Remove added comments.
Co-Authored-By: romitgabani1 <romitgabani1.work@gmail.com>
* chore
---------
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
The locale E2E tests (de->ar and de->pt-BR) were flaky because they used
a brittle CSS selector (.bg-default > div > div:nth-child(2)) to click
the language dropdown. This selector was susceptible to react-select's
internal Input wrapper element intercepting pointer events.
Changes:
- Add data-testid='locale-select' to the language Select in general settings
- Replace CSS path selectors with getByTestId('locale-select') in tests
- Replace text selector for pt-BR with getByTestId('select-option-pt-BR')
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
The 'should render the /teams page' test was failing because PR #27650
introduced a showHeader flag that hides the 'Teams' heading for users
without teams. Other tests in the file were updated to create users
with hasTeam: true, but this test was missed.
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* fixed team event form ui
* edge cases for some mobiles
* revert with removed padding
* chore
---------
Co-authored-by: Devanshu Sharma <devanshusharma658@gmail.com>
* feat - Restrict same email to create more than 'n' active bookings at a time
* updated checkbookerbookinglimit function
* type fix
* minor change
* import fix
* minor fixes
* back to null on disable
* back
* type check
* managed edge cases
* chore: name changes
* name changes
* fix
* minor change
* changed name
* use default value for maxactivebookingsperbooker, and some minor changes
* disabling bookerbooking limit for recurring event
* disabling bookerbooking limit for recurring event
* type fix
* ui fix and backend eventtype update check
* removed unused code
* minor changes
* update validation
* chore
* Do not check booking limits if rescheduling
* minor fix
* Refactor eventType update errors
* Typo fix
---------
Co-authored-by: Joe Au-Yeung <j.auyeung419@gmail.com>
* fix: hiding And connect with specific team members option for non org teams
* removed console logs
* added comment
---------
Co-authored-by: Tushar Bhatt <95581504+TusharBhatt1@users.noreply.github.com>
* restricting platform user from creating another regular org
* changed as suggested, additional check of membership is accepted
---------
Co-authored-by: Hariom Balhara <hariombalhara@gmail.com>
* made text dependent on host's refund policy by conditionally rendering refund message
* add more conditions in case of no refund message
* made refund message more appropriate
* Update apps/web/public/static/locales/en/common.json
* Update apps/web/public/static/locales/en/common.json
* Update apps/web/modules/bookings/views/bookings-single-view.tsx
---------
Co-authored-by: romit <romitgabani1@gmail.com>
Co-authored-by: Tushar Bhatt <95581504+TusharBhatt1@users.noreply.github.com>
Romit