Sebastian, is this still an issue? Can you provide more details?

Steps to reproduce the issue:

1. Open https://tools.taskcluster.net/
2. Sign in.
3. Check log in state later (in this case: Login was gone when I checked 27 minutes after login).

From the failure console:
https://tools.taskcluster.net/index.e83fabe75a93ea3bffa4.v1.js

Could not renew login:
Object { error: "login_required", errorDescription: "Login required", state: "g0S4XS-KbkBQLiFl3JKqQsc80aCiSA9." }
AuthController.js:124:6
r AuthController.js:124
i es6.promise.js:75
N es6.promise.js:92
e _microtask.js:18
exports _microtask.js:37
js es6.promise.js:13
Webpack 10

Thanks! That's an error that we carry through from Auth0, suggesting that you're logged out and need to re-login (either to re-enter MFA, or to re-enter your password, or both). During that 27 minutes, were you doing anything that might affect the browser state? Are *.taskcluster.net and your IAM login in the same container? Typically errors like this have been caused by containers..

When I sign in, a new tab with tools.taskcluster.net reopens and redirects to autho0 and eventually closes. The tab has the same container color like the one from the signin action got triggered. Login data seems to be stored in Local Storage.

What about login.taskcluster.net? Is that included in the same container?

It wasn't part of the container but adding it didn't change the situation (auto--logout after 19:15).

Can you try it without using containers?

I think this is a symptom of bad interactions between federated logins (as we do with Auth0, and it in turn does with Duo and Github and other IAM-related services) and containers. I have similar problems trying to log in to Google properties using containers, for example (I somehow managed to configure it on my laptop, but not on my desktop -- containers are weird).

So it might be best to treat this as a containers bug. Alternately, the IAM team might be able to help -- they use https://discourse.mozilla.org instead of Bugzilla for their bug tracking, though.

I assume this was a containers issue -- please re-open if necessary.

I've been getting the same behavior today, 15-45 minutes or so and I get logged out...

The whole thing has been rewritten in the last few months, so this is unlikely to be the same error, but for lack of a better bug this one will do :)

This has always been a problem for me, and it used to be due to containers. I'm not sure if the cause is still the same now.

I'm hoping this will be fixed in the next deployment.