We were unable to reproduce this problem locally. However, we think we have tracked down the problem and it fixed it in 3.11.0. This issue can be re-opened there are any confirmed cases in 3.11.x or later.

Hi everyone,

We have put together a list of things that you could capture that would help us debug this issue. These will be useful if you see it while using a version of Stash 3.9.1 or greater:

1. Capture the entire page and its JS/CSS resources using your browser's "Save Page" function (Ctrl+S usually).
2. Capture the source of the page as Stash sent it down, right click on the page and choose "View page source" and save that content. (This is because Save Page often rewrites URLs on the page.)
3. The headers your browser is using to make requests (you can find these on this page http://pgl.yoyo.org/http/browser-headers.php)
4. The language setting in your Stash user profile
5. The default language setting for your Stash instance (you may have to ask your administrator for this information)

You will need to do the first two steps without refreshing the broken page as when you refresh the page might become correct again.

If you manage to capture this data can you raise a support ticket at https://support.atlassian.com so you can secuerly send it to us, please reference this ticket so its properly escalated.

Thank you
Scott.

Another possible instance of this for Stash 3.8.0: https://answers.atlassian.com/questions/17437288/stash-3.8.0-mac-users-see-ui-language-randomly-change-to-german---has-anyone-heard-of-this

Hi,
we are on 3.9.2 and still have this bug:

Hello everyone,

We found a bug in the way the front end calculates the language that could have resulted in some problems. That bug has been fixed in Stash 3.9.1. Please let us know if you continue to see problems after upgrading to Stash 3.9.1.

Thank you,

– Brent

When this happens for me a forced reload of the page reverts it to my chosen language.
Also, I usually get pages in French and I know we have a lot of French users. I imagine there is some caching occurring somewhere from a previous user displaying the same page.

We are seeing this issue after upgrading to 3.8.0. We are not clustering yet, and the users are seeing the German language briefly.

Hi guys,

We just started experiencing this problem after upgrading our Stash from 3.2.5 to 3.7.0 about a week ago (we are single-server, not Stash Data Center).
In our case, most of our users' computers have Japanese set at the OS and browser levels, but English set in Stash.
What we're experiencing is that the English setting in Stash is sometimes being ignored and Stash is displaying Japanese (presumably due to either the OS or browser language setting).

After receiving a user report of this issue, we were able to reproduce it ourselves and found that...

• It is reproducible on both Chrome and Firefox
• Setting our account preference to English in Stash, but the OS and browser to Japanese, results in English being displayed sometimes and Japanese displayed other times
  • If you keep pressing F5 to refresh a page in Stash, we found that sometimes it would display English and sometimes Japanese
• Setting English higher than Japanese at the browser-level still resulted in Stash displaying Japanese sometimes
  • Perhaps it's looking at the OS setting? Or perhaps it had cached the browser language setting from when Japanese was higher?

Note: We have Stash's global default set to "English (United States)" in the Administration area, so the only possible place Stash could be picking up the Japanese setting is from either the browser or OS.

Since we can reliably reproduce this problem, please let us know if there's any data you need for your investigation.
Thanks.

Yes you're right. We are using a single standalone Stash server (but connected to JIRA and Jenkins).

Thanks for the additional input Alex and Martin. We've been investigating this issue and have only been able to replicate easily on a clustered Stash Data Center instance. However, given that the issue was reported before Stash Data Center was released, I'm assuming nobody reporting this issue is running in a clustered environment (let me know if that's wrong).

We'll keep trying to track this one down.