Page MenuHomePhabricator

"add link ->link inspector can be used to accept reject links and save an article" test is flaky
Open, Needs TriagePublic

Description

The following test has flaked 10 times recently:

"add link ->link inspector can be used to accept reject links and save an article"

This has caused it to appear on the flaky test dashboard.

The location of this test:

https://integration.wikimedia.org/ci/job/quibble-composer-mysql-php81-selenium/13339/

Part of: https://integration.wikimedia.org/ci/job/quibble-composer-mysql-php81-selenium/

The purpose of this ticket:

  • Investigate the flaky test and determine the cause of the flakiness.
  • Depending on the results of the investigation, implement the fix.

This is a cypress test:

https://gerrit.wikimedia.org/r/plugins/gitiles/mediawiki/extensions/GrowthExperiments/+/refs/heads/master/cypress/e2e/AddALink.cy.ts


last failure:

2026-01-01 13:49:552add link ->link inspector can be used to accept reject links and save an articlehttps://integration.wikimedia.org/ci/job/quibble-composer-mysql-php81-selenium/16874/

Event Timeline

Open questions:

Is this test being run somewhere else? If so, is it still flaky?

Open questions mk 2:

What extension is this part of?
Where is the test code?

Open questions mk 2:

What extension is this part of?

It is a part of Growth Extensions

Where is the test code?

https://gerrit.wikimedia.org/g/mediawiki/extensions/GrowthExperiments/+/9980a4060662b9c2572b667c9760de2d390bdaed/cypress/e2e/AddALink.cy.ts

Looks like it's a cypress test

Open questions:

Is this test being run somewhere else? If so, is it still flaky?

It doesn't have the @daily tag so it is not a part of the daily runs. This test has not flaked since Jan 1, but it is still on the flaky test list

It's worth reviewing anyway - there's a couple of assumptions I'd like to answer:

Is the user it's using guaranteed to be there?
Is other data it's reliant on guaranteed to be there?

If the data it's asserting on is guaranteed to be there and it hasn't flaked in a few more days, we can close this one down too.

Change #1238797 had a related patch set uploaded (by Vaughn Walters; author: Vaughn Walters):

[mediawiki/extensions/GrowthExperiments@master] [DNM] selenium: Checking AddALink.cy.ts

https://gerrit.wikimedia.org/r/1238797

This is passing 87/87 in CI ( see paste for full log: P88788 )

It also passed 29/29 (before timing out after an hour) within ci runs of the entire test suite

Is the user it's using guaranteed to be there?

Yes, it is created on each run. So as long as users can be created through the api, then the user is guaranteed to be there. And if users can no longer be created, there there are much greater problems, so I say yes, the user is guaranteed to be there.

Is other data it's reliant on guaranteed to be there?

It is guaranteed to be there as long as the task to create the data on each run is successful.
For example, in ci (as seen in the paste above)
Successfully created The Hitchhiker's Guide to the Galaxy
so "Douglas Adams", link suggestions (Hardcover, Houghton Mifflin Harcourt) exist
This worked correctly on each of the over 100 runs in CI.

Because this is passing 100 percent of the time, the data is there when needed, and this test has not failed since January 1, I believe no further action is needed on this ticket.

Change #1238797 abandoned by Vaughn Walters:

[mediawiki/extensions/GrowthExperiments@master] [DNM] selenium: Checking AddALink.cy.ts

Reason:

No longer needed. No changes necessary

https://gerrit.wikimedia.org/r/1238797

note: T415655 was merged yesterday by test platform about getting growth working with quickstart