Troubleshooting

The container won’t start at all

Under production settings, RedScribe deliberately refuses to start rather than run in a half-safe state. Two things cause this:

• Debug mode is on. Production settings won’t start with debug mode enabled, since that would show full error details, including source code and settings, to anyone who triggers an error. If you genuinely need debug mode temporarily, there’s an explicit opt-in for that; leave it off otherwise.
• There’s no root key yet. Production settings require the instance’s encryption key to exist before they’ll start at all. Development settings only warn about this and keep running, production settings hard stop. See First run to generate one.

It starts, but warns about running in “dev” mode

This means DJANGO_SETTINGS_MODULE is still pointed at development settings, which is the default in .env.example. That’s fine for trying things out locally, but a real deployment needs production settings instead, since development mode skips HTTPS enforcement, secure cookies, and strict transport security entirely. See Configuration for how to switch.

It starts, but warns about no root key

Everything works except anything that touches encrypted content, findings, comments, checklist results, and so on, all of which need the instance’s key before they can be read or written. This warning is expected the very first time you start RedScribe, before you’ve generated one. See First run.

Redirect loop (constant HTTPS redirects)

Covered in Configuration under “Common mistake: redirect loops”. Short version: production settings trust a header your reverse proxy is supposed to set, and if nothing is actually acting as that proxy in front of the app, every request forces an HTTPS redirect it can never satisfy.

/setup/ doesn’t work, it redirects to the login page

This page only ever works before any admin account exists on the instance. If it’s redirecting you, someone (maybe an earlier attempt, maybe a teammate) has already created one. Ask whoever holds that account to log in and create yours from User management instead, or use the command line method if you have shell access to the server.

TLS certificate never appears (Tailscale method)

This method needs the host to already be joined to a Tailscale network with HTTPS certificates turned on for that network, before you start RedScribe. If that wasn’t already true, certificate issuing silently has nothing to work with. Check tailscale status on the host, confirm HTTPS certificates are enabled for your network in the Tailscale admin console, then restart the stack.

TLS certificate never appears (bring your own certificate)

The web server waits for both certs/fullchain.pem and certs/privkey.pem to exist, under those exact filenames, before it’ll start serving over HTTPS at all. Double check both files actually landed where expected. A common slip is copying only one of the two, or copying them to a different path.

The command line setup asks for a password and rejects mine

Local accounts need a password of 14 characters or more. Anything shorter gets rejected with no partial account created, just try the command again with a longer one.

Tip

Still stuck? Most first-run problems come down to one of the two things above, dev settings where prod was needed, or a missing root key. Confirm those two first before digging further.