Server error after installing twofactor_u2f

ℹ️ Support
1

I just migrated from NC 13.0.12 to 14.0.8 to 15.0.5.3 on debian jessie with Apache 2.4 and PHP 7.3. All fine so far for two other instances where I did the same steps as described below.

In my last instance, I encountered a warning when enabling the twofactor_* apps: “enabling this leads to instability” or similar (in yellow). However, that was AFTER I was installing an app through the app store within the web interface. I think it was after I installed twofactor_admin (not sure, though). After that, the server was inaccessible through web.

It gets accessible as soon as I manually delete the apps/twofactor_u2f folder via shell and it gets inaccessible (internal server error) as soon as I install twofactor_u2f, no matter whether other twofactor_* apps are installed or not.

I see the errors in NC log below but have no clue what they mean: https://seyfarth.de/nextcloud/u2f-errorlog.txt (170kB, please cite relevant parts, file will be deleted once not needed nay longer). Any hint/help/ideas what to try?

2

Getting rid of an app is not only done by deleting a directory, because the database tables of an app still exist. I think the mentioned error message is clear Message":"An exception occurred while executing 'SELECT id, user_id, key_handle, public_key, certificate, counter, name FROM oc_twofactor_u2f_registrations … Column not found: 1054 Unknown column ‘name’ in ‘field list’ …"

The app requires a database table named “oc_twofactor_u2f_registrations” which doesn’t contain the column "name"". You can try to drop the tables manually and then reinstall the app so that the tables should correctly been created again.

3

There are no tables with u2f in their name. These tables with twofactor in their name exist in my database:

  • oc_twofactor_admin_codes
  • oc_twofactor_backupcodes
  • oc_twofactor_nextcloud_notification_tokens
  • oc_twofactor_providers
  • oc_twofactor_yubikey

oc_twofactor_providers contains no rows that contain u2f in provider_id.

4

Ok, if you couldn’t find the mentioned table, you should be able to reinstall the app, if you require it.

5

If I do, my whole instance is unusable as soon as the app is installed.
Deleting it’s directory resumes to normal operation.

6

Still have this issue in NC 18.0.1 on that NC instance. Please advise on how to debug this.

7

I’ve just checked my database and found related entries in the “oc_appconfig” table. It might be worse to delete this entries too, to make sure that the app has completely been uninstalled. (Keep in mind to create a database backup :wink:):

> select * from oc_appconfig where appid like "%u2f%";
+---------------+-------------------+-------------+
| appid         | configkey         | configvalue |
+---------------+-------------------+-------------+
| twofactor_u2f | enabled           | yes         |
| twofactor_u2f | installed_version | 5.0.2       |
| twofactor_u2f | types             |             |
+---------------+-------------------+-------------+

``´