Repository navigation
test: resolve potentially flaky playwright tests - #66026
Merged
Merged
Conversation
System tags are server-global, but the systemtags specs cleaned up by deleting every tag on the instance (`clearTags()` in `afterAll`, and a mass delete in `beforeEach` of the admin settings spec, which also used the fixed names `foo` / `bar`). A spec running in parallel lost its tags mid-test, and the non-forced delete threw when a tag vanished underneath it. Generate tag names through a `tagName` fixture that deletes exactly the tags created under those names after each test. Assisted-by: ClaudeCode:claude-opus-5-5 Signed-off-by: Ferdinand Thiessen <opensource@fthiessen.de>
`force_language` / `force_locale` were deleted in `beforeAll` and restored in `afterAll` of the whole personal info spec. With `fullyParallel`, every worker that runs a test of the file runs those hooks, so one worker restoring the enforcement made the language and locale section read-only for another worker still testing it. The per-test `beforeEach` only narrowed that window. Toggle the enforcement around the language and locale tests alone and run them serially, so a single worker owns that state. The restored values are the defaults of the test server. Assisted-by: ClaudeCode:claude-opus-5-5 Signed-off-by: Ferdinand Thiessen <opensource@fthiessen.de>
`networkidle` is timing-fragile with background polling. After picking a locale it could also resolve on the page that was about to reload, before the reload had even started. Wait for the reload's `load` event instead, and rely on the retrying assertion after opening the profile page. Assisted-by: ClaudeCode:claude-opus-5-5 Signed-off-by: Ferdinand Thiessen <opensource@fthiessen.de>
The session and account fixtures retried a failed login or `createRandomUser()` exactly once after a fixed 800 ms. The transient failures are real (logins answered with HTTP 500 twice in a row, user creation exiting non-zero), and a single fixed-delay retry is not enough to ride them out. Share a `retryWithBackoff` helper that tries three times with an exponential delay, and use it in every fixture that had its own copy of the single retry. Assisted-by: ClaudeCode:claude-opus-5-5 Signed-off-by: Ferdinand Thiessen <opensource@fthiessen.de>
`ContactsMenuPage.search()` promised to wait for the server response but only filled the debounced input, so callers asserted against whatever the list showed before the search request was even sent. Wait for the contacts request carrying the search term. Assisted-by: ClaudeCode:claude-opus-5-5 Signed-off-by: Ferdinand Thiessen <opensource@fthiessen.de>
`isFullyInViewport()` first asserted that the locator resolves to one element. For a row the virtual list has dropped from the DOM, which is the strongest form of "not in the viewport", that assertion threw, and since `expect.poll()` does not catch errors thrown by its callback, the scrolling specs polling for `false` failed instead of passing. Report an element that is not rendered as not in the viewport. Assisted-by: ClaudeCode:claude-opus-5-5 Signed-off-by: Ferdinand Thiessen <opensource@fthiessen.de>
`selectAll()` / `deselectAll()` force-clicked the master checkbox without checking the outcome. A click landing mid-render could be a no-op, and the caller only failed later on an unrelated assertion. Use `check()` / `uncheck()` instead, which assert the resulting checked state. The data attribute sits on the `<input>` itself. Assisted-by: ClaudeCode:claude-opus-5-5 Signed-off-by: Ferdinand Thiessen <opensource@fthiessen.de>
`appRow()` filtered table rows by a substring of their text and silently took the first match in DOM order, so "Files" also matched "Files sharing". The enable and disable button helpers inherited that. Match the row by the app name its details link starts with. Assisted-by: ClaudeCode:claude-opus-5-5 Signed-off-by: Ferdinand Thiessen <opensource@fthiessen.de>
The app management tests enable `testing` and disable `updatenotification` instance-wide, but only reset that state before each of their own tests. The last test's changes leaked into every spec running afterwards on the same server. Restore both apps to their default state once the spec is done. Assisted-by: ClaudeCode:claude-opus-5-5 Signed-off-by: Ferdinand Thiessen <opensource@fthiessen.de>
The group share test had the sharee delete the file through the shared folder right after creating the share. Shares mount into the recipient's tree asynchronously, so the DAV `DELETE` could hit a 404 before the mount was there. Poll for the shared file with `waitForShare()` first, as the other sharing specs do. Assisted-by: ClaudeCode:claude-opus-5-5 Signed-off-by: Ferdinand Thiessen <opensource@fthiessen.de>
The rename input selects the basename in a `$nextTick` after it is rendered, but the test read `selectionStart` / `selectionEnd` once as soon as the input was visible, which could be before the selection was applied. Poll the selection range until it matches. Assisted-by: ClaudeCode:claude-opus-5-5 Signed-off-by: Ferdinand Thiessen <opensource@fthiessen.de>
susnux
marked this pull request as ready for review
October 29, 2026 23:20
susnux
requested review from
DerDreschner,
kristian-zendato,
pringelmann and
skjnldsv
and removed request for
a team
October 29, 2026 23:20
skjnldsv
approved these changes
Oct 30, 2026
pringelmann
approved these changes
Oct 30, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
resolve the remaining potential flaky playwright test issues.
Checklist
3. to review, feature component)stable32)AI (if applicable)