$scope.configureBackup is undefined

[mehdi]

Hello!

Trying to migrate my cloudron to a new server, I am running into an issue when trying to import the backup configuration into the new server : on the front end, it seems to load without end. Opening the dev-tools, I see that no request is made, and that there is an error in the console $scope.configureBackup is undefined.

Additionally, it seems I can only get the web interface to work with Firefox and Safari (which gives me a slightly different error for the backup thing : Error: undefined is not an object (evaluating '$scope.configureBackup.backupFolder'), which actually may give more information).

Chrome gives me a NET::ERR_CERT_INVALID error, and it is weirdly not bypassable: (sorry, the explanation is in french, with a machine translation following)

Un chiffrement est normalement utilisé sur le site XX.XX.XX.XX pour protéger vos informations. Lors de la dernière tentative de 
connexion de Chrome au site XX.XX.XX.XX, des identifiants 
inhabituels et incorrects ont été retournés. Il est possible qu'un 
individu malveillant tente de se faire passer pour XX.XX.XX.XX ou 
qu'un écran de connexion Wi-Fi ait interrompu la connexion. Vos 
informations restent sécurisées, car nous avons arrêté la 
connexion avant l'échange des données.

Vous ne pouvez pas consulter le site XX.XX.XX.XX pour le 
moment, car il a envoyé des identifiants brouillés qui ne peuvent 
être traités par Chrome. Les erreurs réseau et les attaques sont 
généralement temporaires. Vous devriez donc pouvoir accéder à 
cette page ultérieurement.

Encryption is normally used on XX.XX.XX.XX to protect your 
information. The last time Chrome tried to connect to 
XX.XX.XX.XX, unusual and incorrect credentials were returned. It 
is possible that a malicious individual is attempting to impersonate 
XX.XX.XX.XX or that a Wi-Fi connection screen has interrupted 
the connection. Your information is still secure because we 
stopped the connection before the data was exchanged.

You cannot view XX.XX.XX.XX at this time, as it has sent 
scrambled credentials that cannot be processed by Chrome. 
Network errors and attacks are usually temporary. So you should 
be able to access this page later.

This is on 7.0.4, with filesystem backup (and yes, my backup folder is copied onto the new server, at the same location, and owned by the yellowtent user). Any way I can trick the backup into finishing? Willing to do it with a curl command or something similar if necessary.

[girish]

@mehdi Are you getting this error, when you go to https://ip/restore.html ? (i.e this is not an app restore, correct?) It seems to me that some resource is not loading correctly, because of the cert issues. Does curl -k https://ip/restore.html work OK?

[girish]

@mehdi said in $scope.configureBackup is undefined:

trying to import the backup configuration into the new server

FWIW, you can just fill that form manually instead of importing from the json file and see if that helps. If you open up the json, you can just fill up those values in the form.

[girish]

I found this curious trick in https://dblazeski.medium.com/chrome-bypass-net-err-cert-invalid-for-development-daefae43eb12 . It says if you type “badidea” or “thisisunsafe” , it will let you proceed!

[mehdi]

On Firefox, I see no error either on the console tab nor on the network tab before the "main" error when I am trying to restore. Yes, this is indeed on the https://IP/restore.html page, not for an app restore

[mehdi]

@girish Same error when filling the form manually

[mehdi]

@girish Nice trick. Just tried it, i managed to get the page in Chrome. However, I still get the same error : TypeError: Cannot read properties of undefined (reading 'backupFolder') is the exact wording in Chrome

[mehdi]

@girish said in $scope.configureBackup is undefined:

Does curl -k https://ip/restore.html work OK?

Yes, it does work OK.

[girish]

@mehdi OK, just to rule out the obvious, I tried to backup/restore 7.0.4 install myself and it seems to work. I used Vultr Object Storage.

Not sure what the problem is, do you think you can send the IP to support@cloudron.io ? And maybe the json as well (it doesn't have the secrets, so should be safe).

[mehdi]

@girish maybe it's specific to the filesystem backup restore process? I can send you the IP, sure, but to be safe (as I had not seen the new generate-setup-token parameter in the doc ^^) I have systemctl stop box for the time being. Can I enable generate-setup-token after the initial install, with the CLI ?

[mehdi]

I have managed to add a SETUP_TOKEN after the fact (for anyone interested, simply run openssl rand -hex 10 > /etc/cloudron/SETUP_TOKEN as root). I have sent you the IP and token by email.

BTW, I have also tried other backup providers on the interface (with arbitrary data in the form), and they seem to indeed send a web request. So I believe the issue is on the front-end side, and specific to the filesystem backup provider.

[mehdi]

@girish Found the error : https://git.cloudron.io/cloudron/dashboard/-/blob/master/src/js/restore.js#L292

backupConfig.backupFolder = $scope.configureBackup.backupFolder;

should be

backupConfig.backupFolder = $scope.backupFolder;

[mehdi]

Now getting an error 400 setup token required. The setup token is in the URL, and in the network tab i do see it being sent in the request...

[mehdi]

Managed to get the restore working, by fixing the /home/yellowtent/box/dashboard/dist/js/restore.js file manually on the server, with the fix explained above, and removing the SETUP_TOKEN file. I did not go to the length of looking for the root cause of the setup token thing not working, probably the variable being sent inside an object instead of at the root of the JSON request or something.

[nebulon]

Great investigation! I have fixed up that copy and paste error in the restore.js now

[mehdi]

@nebulon Thanks ! I'll let you look for the precise fix for the SETUP_TOKEN problem by yourselves though ^^

[girish]

@mehdi this was indeed a bug in the restore code path. the setupToken, as you guessed, should be sent in the top level object and not as part of the backupConfig . Fixed now.