ProblemA few days ago my hAP ac (RouterBOARD 962UiGS-5HacT2HnT) spontanously rebooted when I tried to log into it via WinBox. After the reboot the DHCP server did not hand out any more leases and Radius error messages flooded the logs.
After the first panic subsided, I checked the router, rebooted it a few times, turned DHCP and Radius into debug mode but all I got were messages like these:
Oct/17/2018 08:23:47 memory dhcp error radius authentication failed for XX:XX:XX:XX:XX:XX:XX:XX: RADIUS server is not responding
and other Radius timeouts. After some more debugging, I noticed that my Usermanager interface was extremely slow as well.
The combination of DHCP, Radius and user-manager comes from my configuration where only known MAC addresses get an IP address via DHCP. For more details have a look at User-Manager Setup For Static DHCP.
Radius is the Victim, Not the Culprit
Any kind of Google search about Radius did not lead to any insights and the debug logs did not help as well.
The search for the solution took me alomost one hour, the solution just a few seconds. To get the router working properly again, only these two lines on the console were necessary:
/tool user-manager database rebuild
/tool user-manager database rebuild-log