Docusaurus: Site becomes stale (nav stops working) after idle for extended period of time

[jonnyasmar]

🐛 Bug Report

After leaving a page idle for some extended period of time, navigation seems to stop working. Clicking any SPA link on the page triggers the loading bar which progresses indefinitely and never loads the page. A refresh resolves the issue immediately.

Have you read the Contributing Guidelines on issues?

Yes

To Reproduce

(Write your steps here:)

1. Load a page
2. Leave it idle for a while (not sure how long 🤷‍♂️)
3. Click any SPA link on the page
4. Observe the indefinite loading bar

Expected behavior

I thought it would navigate to the page I clicked.

Actual Behavior

It doesn't navigate to the page I clicked 😥

Your Environment

• Docusaurus version used: 2.0.0-alpha.54
• Environment name and version (e.g. Chrome 78.0.3904.108, Node.js 10.17.0): Chrome 84.0.4147.89; Node 12.16.1
• Operating system and version (desktop or mobile): Mac Catalina; Desktop

Reproducible Demo

Can't share anything at present, but I believe I may have even witnessed this on https://v2.docusaurus.io

[slorber]

Hi,

Do you have an opensource site with link to source code and live version?

Idle for a while, how long was it last time it happened?

What is your production host?
Has the site been redeployed in the meantime?

I suspect this may be related to rebuilds during Idle time, and according to deployment setup, older files required to navigate may not be available anymore.

If this happens again, please don't refresh, but try to inspect console logs and network tab.

Could also be somehow related to this issue, as js output files seem to change too unnecessarily: https://github.com/facebook/docusaurus/issues/3383

[jonnyasmar]

This is a lot of excellent stuff to go on @slorber , thanks!

I don't have anything open source that I can share right now, but I will consider all of this feedback and see if I can't create a repro deployment or at least provide more info, here.

Idle time: not sure, I usually notice it in the morning after logging off of work the night before.
Prod host: AWS
Redeployed: possibly, but I can't confirm at present that this has happened every time I've witnessed this issue

Thanks!

[lex111]

In order to reduce staled issues, I am closing this issue, but feel free to open it again if needed.