Website results in 404 error

[noahtallen]

Accessing the website results in 404 error. (same issue as yesterday)

[lirbank]

Going directly to the docs works though, just the landing page seems to be gone.

This works -> https://facebook.github.io/jest/docs/en/api.html

[thymikee]

cc @ericnakagawa – we're experiencing some flakiness in website deploy, could you have a look a it?

[captbaritone]

Seems to be back up for me.

[SimenB]

Yeah, it went back up by the time @thymikee commented. Not really sure what's going on, or if it's the "fault" of docusaurus, circle ci or github pages

[chinesedfan]

It works now.

[SimenB]

Not sure what's going on - it seems like the redirect from https://facebook.github.io/jest/ to https://facebook.github.io/jest/en (notice language at the end) is down during deploys. I suspect either crowdin or docosaurus is doing something weird.

@ericnakagawa @JoelMarcey ideas?

[JoelMarcey]

@SimenB Can you provide a bit more context? The site seems functional now. Am I missing something there?

What exactly was done when you noticed that the site was down to make it come back up?

Just trying to figure out if there is something reproducible here.

[SimenB]

It's during deploy. You might be able to reproduce by triggering a deploy of the website?

[JoelMarcey]

Right. I understand that I think. But now the site is functional (right?). What was done to make the site functional again? Or are you saying that once the deploy is finished, the site is fine. i.e., in the seconds/minutes that the site is deploying, you get the funky behavior, but after that, all is good?

[SimenB]

Yes, correct. I wonder if it's related to crowdin, as it takes a few minutes and the thing that's gone is the default language redirect

[JoelMarcey]

Talking to some folks here, @ericnakagawa and @rickyvetter

Perhaps something happening when pages are being built and pages get deleted before completing. The issue reported mentioned pages "showing up" later. That sounds like something is delayed maybe something with their Github's gh-pages deployment.

almost certainly not a crowdin issue because all of the crowdin interaction happens before you git push. its either an issue with gh-pages deployment or with the browser cache.

[SimenB]

#6508 :(

[JoelMarcey]

@SimenB that issue you referenced is closed now. Is there still a problem?

Also, what version of Docusaurus are you running now? 1.2.1? That’s the latest.

_{Sent with GitHawk}

[SimenB]

Yeah, it fixes itself. But it still happens. We're in 1.0.14. Should we upgrade?

[JoelMarcey]

There have been a bunch of fixes since then. It may help. You can look at the Changelog see what has been added and fixed.

Of course, run locally first to make sure nothing breaks.

cc @yangshun @endiliey

_{Sent with GitHawk}

[yangshun]

@SimenB Do we know why it fixes itself?

Yeah, it fixes itself. But it still happens. We're in 1.0.14. Should we upgrade?

Yes, please! Actually, we'll release 1.3.0 soon and I can help to do an upgrade for Jest.

[SimenB]

Do we know why it fixes itself?

No, but I suspect (without evidence) that it's down during deploy, but comes up again after deploy is done. There was a merge-spree for PRs yesterday, same as the last time (this issue was opened when we merged PRs for the Jest 23 release).

Actually, we'll release 1.3.0 soon and I can help to do an upgrade for Jest.

Great, I'll hold off then 🙂

[endiliey]

1.3.0 is released 😊

[yangshun]

I updated Docusaurus to 1.3.0 here but CI seems to fail because of Prettier. I'm not sure if I fixed it correctly because I don't see any autofix command for scripts in package.json.

[SimenB]

Hopefully fixed