Log of unstable tests

[phillipj]

We're having some issues with unstable and flaky test runs on Travis. To list incidents and not forget about them, please add a comment with the name of the test of you experience it when creating PRs / merging to master.

[phillipj]

anchorMarkdownHeadings/makes pretty slugs when opening PR with unrelated changes #231. Fixed after two Travis restarts.

[mrkiffie]

Do we have any more details of the anchorMarkdownHeadings/makes pretty slugs test failure or steps to replicate?

[phillipj]

The short answer is no. Wasnt able to re-produce it locally, nor see anything unusual in the Travis logs.

[fhemberger]

Can we increase the timeout for those tests (t.timeoutAfter(ms)) and see if the situation improves? Default seems to be 500ms, maybe Travis takes sometimes just a tad longer?

[phillipj]

TBH I would like to see timeouts occur, before trying to increase the test timeouts. The regex code related to making these slugs should really not use more than 500ms.

Very sorry I didnt copy the Travis output when it was failing 😞

@fhemberger I've seen you commenting some PRs with "ignore the errors, Travis is drunk". Do you remember which tests were failing in those cases?

[stevemao]

I have some experience with unknown travis fails with some bigger project (conventional-changelog/conventional-changelog@40ab546 and conventional-changelog/conventional-changelog@104b3aa). I believe it might be travis own problem. Timeout might not be the issue (I've increased twice). The worse part is the error message doesn't tell much and it's hard to replicate :(

[phillipj]

Closing this for now, lets open it again if unstable tests starts showing again.