6.46 update seems to broke my radius user manager.

RouterOS The User Manager
1

Hello,

I have upgraded my system to 6.46 stable. After that I have changed the usb device which keeps usermanager database and log files. I see multiple disk drives even though currently there is only one. After installing new usb I gave the command /tool user-manager database set db-path=disk1/usermanager to relocate the userman database. After this command everything seems fine, but after reboots it is gone. The configuration does not persist.

Any help much appreciated.

2

We have reproduced this issue, thanks for reporting it.

3

Hello,

Thank you for the response, unfortunately I had to spend hours and then I finally gave up and reinstalled the router with netinstall, then the problem has gone. But the backups, I took did not work, I have lost my capsman, all router configuration. I have started from zero. I used to have two device which were controller by capsman, radius server with mac and dhcp authentication… I had to quit this configuration because I could not redo my previous configuration which had been done three years ago. I had to move to dhcp to opnsense, quit using for a while the radius based authentication, and quit all together mac based authentication and content with less security. I also had to quit capsman because second device never showed up in the capsman controlled list.

I am stupid for upgrading. So this issue has totally destroyed my mikrotik infrastructure and gave me a sleeplest night, and back pain without any gain.

Anyway I am still pleased to know it is a bug. I would like to hear a solution if possible.

4

I had the same problem. Writing a schedule solved the problem somewhat, but it was repeated.
/tool user-manager database set db-path=disk1/user-manager/
At first, I ran this schedule in startup but after a few hours the configuration does not persist.

Next, I run the schedule every two hours to fix the problem temporarily. This error is displayed to me

“internal error: faild to listen on 127.0.0.1: entry already exist”

What this means and how it will be resolved. I didn’t find anything on the internet!

In addition, this problem is fixed by updating the routerOS to 6.42.2?

5

Hello

I have been having a similar problem since last year.

UM Db path is not set after reboot of router. Causes “Authentication Failed - radius timeout” for PPPoE users.

Path has to be set via terminal again and/or rebuild database.

If this does not work, PPPoE users need to be removed and re-added manually in UM. FRUSTRATING.

I have tried contacting support two times, to no avail.

Any help please?

Thanks in advance.

6

UPDATE

I have rolled back FW to 6.44.5 and the problem has disappeared.

I am not sure on exactly which FW the issue first presented for me but I have a strong belief it was from 6.46 onwards.

For anyone that this may help and hoping MT Support notices this.

ZSam

7

I have exactly same problem, sometimes I need to reformat external unit to use again “disk1”
I can copy and remove file inside, but for usermanager the database is like readonly.

I suggest you what is worked for me for months:

  1. Reall all first, and try to understand all first…
  2. Backup EVERYTHING: export file .rsc / make backup / export user manager database. /export certificates / export files inside files / etc.
  3. copy on one directory the actual user-manager files sqldb and logsqldb (not the backups) NO any “JOURNAL” FILES!!!(***)
  4. put sqlite3 from https://www.sqlite.org/2021/sqlite-tools-win32-x86-3350500.zip on same directory and on windows prompt go on same directory do this commands:
sqlite3 logsqldb VACUUM
sqlite3 sqldb VACUUM
  1. Netinstall the device (I suggest use 6.48.2 netinstall) with the version routeros 6.47.9 + user-manager without keep any previous config
  2. Update routerboot to 6.47.9, disable unneded system packages but keep absolutly dhcp and security active (it’s not the time to explain), and reboot
  3. Reimport CERTIFICATES first, if any.
  4. Reimport any file needed
  5. import on disk1 or directly on root the user-manager directory with the files “vacuumed”, replace existing files if needed
  6. reimport .rsc with notepad (or similar) on terminal section-by-section and read first what you set. SKIP USER-MANAGER PARTS
  7. When done set user-manager database path accordingly with new directory TWICE
    do that command on winbox console (copy and paste):
/tool user data
clear-log
y
.. sess rem [find]
rebuild
y
save
justnow

Done.

Note:
***: In other cases do not delete for any reason the journal files if exist on a running user-manager

8

Thank you @rextended

I stayed on old firmware to avoid the problem… A few days ago I needed to replace my MT and the oldest firmware for it was one that reproduced this problem.

After digging and digging I noticed that the problem was that when a PPPoE user lost connection due to reboot or power failure etc, their dial in remained “active” in userman sessions. Thus causing a new radius connection being unable to authenticate…

I finally found this solution posted by a fellow user which solved my problem after SO long!

“Remove false active sessions via script”

http://forum.mikrotik.com/t/remove-false-active-sessions-via-script/47999/10

:local LastSessionUpdate;
:local SessionTimeout;
:foreach i in=[/tool user-manager session find where active=yes] do={
 
# When was the last Update of the session-informations
:set [/tool user-manager session get $i till-time]
 
# SessionTimeout is a value that tells me how many seconds ago the last update of this session was
:set ([system clock get time] - [:pick $LastSessionUpdate ([:find $LastSessionUpdate " "]+1) [:len $LastSessionUpdate]]-[/system clock get gmt-offset])
 
# if last update is more then Timeout seconds ago then close session and log it
:if ($SessionTimeout > $Timeout) do={
/tool user-manager session remove  numbers=$i
:log warning (" Removed false active session by Zaib - Username is  " . [/tool user-manager session get $i user]);
}
}