-
Notifications
You must be signed in to change notification settings - Fork 29
New issue
Have a question about this project? # for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “#”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? # to your account
getting "invalid cipher string format" upon initial login attempt #121
Comments
Thanks for the report.
The KDF invocation is most likely not the issue here, even when using argon2, a round of pbkdf2 is used to generate the masterpasswordhash to authenticate.
This seems like a corrupted vault entry somewhere, though better logging is required to track this down. I'll add that in a bit. |
Added some debug logging on main, could you try the flatpak from: https://github.com/quexten/goldwarden/actions/runs/7984542101 and post the updated error logs? |
Launched with [INF] [12:40] [Goldwarden > Keyring] >>> Creating new memguard keyring I'm removing the dump of what seems to be my actual login data here, even if it is hashed... accept error accept unix /home/jpeeler/.var/app/com.quexten.Goldwarden/data/goldwarden.sock: use of closed network connection |
Good catch, usually I also write a note to not post credentials or password dumps when asking for logs, seems I missed writing that in this case.
So usually the flatpak should run the daemon on it's own when you launch it via the desktop entry, or flatpak run (without command). This:
Sounds like somehow the daemon is stopped. I'm a bit confused what's exactly happening here. What happens if you completely stop the flatpak:
and then run it with just: (it should still show you the agent logs) |
Is this enough?
I can't login or logout using the GUI, so at this point I'm stuck which is why I tried doing things manually previously. Is there a way to execute commands once the daemon is started the normal way? |
This sounds like another problem, where the agent path was incorrect. Should (hopefully) be fixed in 0.2.13.
If you start the daemon the normal way (that is, just starting the flatpak normally), you can run commands with flatpak run, such as: |
I've upgraded my flatpak to 0.2.13. Maybe there's a more thorough purge command, but tried: rm -rf ~/.var/app/com.quexten.Goldwarden/ Tried logging in once unsuccessfully. I assume I entered the correct password, but even after restarting from the beginning again it still reports bad password. ❯ flatpak run com.quexten.Goldwarden -- Also saw this notification once, making me think some traffic is not going to the correct place? Could not pre-login: Bad Request: {"message":"Traffic from your network looks unusual. Connect to a different network or try again later. [Error Code 6]"} |
We're slowly getting somewhere, at least this bit is fixed.
This should only happen on the official servers (bitwarden us and eu). Since you mention you are using vaultwarden, something is wrong here. Did you enter the correct vault url in the login screen again, when logging in? You can verify this with: To fix, the easiest way would be: |
Why is the base config not respected from the login dialog? I set the server and verified it via get-environment oon the CLI. After trying again on the CLI: Gtk-Message: 10:56:01.061: GtkDialog mapped without a transient parent. This is discouraged. I'm assuming that since I was able to login via the CLI earlier that there's no problem with my vaultwarden setup. |
I wonder if that port number is correct... My base url is https://:8443 |
I also get this issue, tested on latest version 0.2.15 built from aur. Got it before on earlier versions, checked back a bit later to see if this ticket would change status, but see that its been idle. Also vaultwarden, currently 1.30.5, no issues from rbw, or bitwarden official browser plugins or bw-cli. No custom port, normal 443, not using the flatpak, no connection refused.
Alongside the error on the CLI:
|
@Roukoswarf could you post the output of: |
I could try making a new account in case its an artifact of an old vault? The vault is using argon2id with the currently recommended settings by upstream. 3 KDF iterations, 64MB memory, 4 KDF parallelism. I tried logging in with api keys, no change, i see my vault contents of some sort in the log, where i cut it in the last line of the log given before, login does appear to succeed, the failure is after. It asks for my fido2 in the logs, and on screen, but always hangs for a while then fails and asks for OTP and succeeds, but thats probably a different issue, not sure why fido2 is failing, works via browser/other clients as well. |
Oh this is interesting. You seem to have a cipher of type '0', that is AesCbc256_B64 without mac. I really thought this was already force migrated by the official clients some time ago. I wonder if it actually has this type, or if it's just the default because parsing is failing somewhere.. |
How would I migrate this manually? I use the official client which doesn't complain about anything, or convert it either I guess. I recently (6 months ago) change from pkbdf to argon, which id have expected to re-key everything, would a fresh master password force a whole recalc? Cant be vaultwarden specific, server doesn't do the crypt. |
Rotating encryption keys doesnt seem to actually work, apps get logged out, but the browser console log is full of "mac required" errors. I'm willing to blame vaultwarden-web for this, but google digs up nothing on a workaround. A fresh account logs in successfully via goldwarden, and successfully rotates account encryption. Can safely say that this is not a goldwarden issue, though some messaging would be good to add, not sure of a better workaround than recreation, clients behave fine, but you cannot rotate encryption key. Wonder if purge would reset encryption, or if it just deletes everything without rotation. Edit: export -> purge -> import does resolve the issue without a new vault entirely. |
Very weird. Might actually be a bug in vaultwarden where it kept old ciphers around despite rotation.
In the latest release, the offending ciphers should get ignored, and the rest will get synced, with warnings printed to console. |
FWIW, I was able to get everything working now. I assume some changes were made in the latest release. Will leave it up to either of you as to whether or not this should be closed now. One small oddity I encountered was when setting the AUTH_SSH_SOCK using a tilda gave me: Using the full expanded path works fine though. |
I have a hosted vaultwarden instance that I'm using. I've also installed goldwarden via Flatpak.
I executed the custom domain config options as well as
goldwarden config set-vault-url
. After attempting to login I get this error message:Login failed: Could not sync vault: could not sync: invalid cipher string format
Is there an assumption being made for the KDF algorithm in use? (Perhaps here?) Wasn't sure if that's what is being described in #24 or not.
On the vaultwarden side I see this:
[2024-02-19 20:48:53.442][vaultwarden::api::identity][INFO] User x@abc.com logged in successfully. IP: 10.0.2.100
The text was updated successfully, but these errors were encountered: