[3.9.0+] Initial Server Setup Fail/Deadlock - Assumed Frontend Bug

SCM-Manager Bug Reports SCM-Manager
1

Hello, I’ve found what I think is a bug on SCM-Manager’s side:

  • Bug description (occurred issue):

    The server is run as a Docker container with the following lines:

    mkdir /home/user/scm_home
    chown 1000:1000 /home/user/scm_home
    docker run --name scm -p 8080:8080 -v /home/user//scm_home:/var/lib/scm scmmanager/scm-manager:3.11.1

    (Note: The commands are run with root permissions via sudo -s; “user” is just the name of a local user).

    I then did a bisect search by installing different versions via
    rm -r /home/user/scm_home
    docker rm scm

    and found out this way that the issue has been occuring since between 3.8.0 and 3.9.0.

    logs.zip (16.3 KB)

  • Expected result / system behavior (from 3.8.0 frontend):

    grafik
    grafik1412×725 35.9 KB

  • observed result / system behavior (from 3.11.1 frontend):

    grafik
    grafik1299×883 115 KB

    Clicking on ‘Erneute Anmeldung’ leads back to a ‘Abschluss der Initialisierung’ with an available selection. However, if you select an item, you arrive back to the page shown above. The consequence is that SCM-Manager cannot be used via the frontend. This error is not fixable with Ctrl+F5 and occurs on other devices too.

    Workaround: The error did not occur when I

    • initially started the server as a 3.8.0
    • opened the 3.8.0 page on a device until the first ‘Abschluss der Initialisierung’ dialog
    • removed the backend (docker rm, but keep the scm_home/ folder) and reran it with the same command as above as a 3.11.1
    • proceeded with the 3.8.0 page, now communicating with the rerun 3.11.1 backend. In this case, I could arrive to the login mentioned above and keeping the newer version. But it’s a non-obvious workaround.

    So my assumption is that the bug is located somewhere in the frontend.

  • SCM-Manager version and installed package:

    • 3.9.0 (earliest found occurence)
    • 3.11.1 (currently affected version)

The SCM-Manager instance runs on a Ubuntu Live server with Docker installed. The Ufw firewall has been disabled in systemctl to avoid port restrictions, and communication between the server and client is within a private network (same router).

If there is anything else you need as an info, I can try to provide it for you.

Thanks!

Till

2

Hey Till,

nice to hear from you! Just to make this clear to me:

After starting the 3.11.1 and opening http://…/scm/, you did not see this “Finish Initialization” screen but directly got to the page with the error?

image
image1139×902 108 KB

And the folder /home/user/scm_home has did not exist (or has been empty) before you start the server?

I’ve gone through this process three times now with little tweaks (running as root or as normal user, using a double / like in your pasted command :wink:), but I always “got through”.

Are there things I might miss from your description?

Keep on coding!
René

1 Like
3

Hey René, thanks!

Yes, I did the “Finish Initialization” screen. I also saw the configuration selection.

However, when I picked one, the process failed and it got stuck in a loop.

I attached a .zip with a video:

Aufzeichnung 2025-12-05 112742.zip (6.2 MB)

And these are the commands I executed beforehand:

grafik
grafik950×137 3.97 KB

For reproduction, it is important to open a new tab with the /scm link. Using one from a previous session may actually make the authorization work. I’ve encountered this myself now.

The folder /scm_home did not exist before when the error emerged.

Wish you a nice weekend,

Till

4

Hello Till, Happy New Year!

While we haven’t been able to reproduce the issue on our end yet, we are very interested in identifying the cause and finding a fix.

To help us investigate further, would you be available for a short remote session to walk us through the steps you’re seeing?
If so, please let us know your availability and we’ll get a meeting invite sent over.

Best regards
Florian