Hi,
I have a 2011UAS-2HnD on firmware 3.0, running webfig 5.26.
My unit is connected via PPPoE to my ISP.
I was configuring an OpenVPN Server, and have verified it has been working fine… HOWEVER… I was tweaking an OpenVPN client config file on a linux host (the client connecting to the 2011UAS), and I was tweaking some settings and lo and behold, I lost connection to the router.
The router was inaccessible for 24 hours until I returned home to take a look at it.
It seems that my linux openvpn conf which excluded auth-user-pass somehow triggered the PPPoE connection to reset and enter a redial loop.
01:45:45 ovpn,info TCP connection established from 172.16.5.16
01:45:45 ovpn,info : dialing…
01:45:47 ovpn,info : using encoding - AES-256-CBC/SHA1
01:45:48 pppoe,ppp,info pppoe-ncf: initializing…
01:45:48 pppoe,ppp,info pppoe-ncf: dialing…
01:45:49 pppoe,ppp,info pppoe-ncf: authenticated
01:45:49 pppoe,ppp,info pppoe-ncf: connected
01:45:49 system,info dns changed
01:45:49 pppoe,ppp,error could not add address: already have such address (6)
01:45:49 pppoe,ppp,info pppoe-ncf: terminating… - could not add address: already have such address (6)
01:45:49 pppoe,ppp,info pppoe-ncf: disconnected
01:45:49 pppoe,ppp,info pppoe-ncf: initializing…
01:45:49 pppoe,ppp,info pppoe-ncf: dialing…
01:45:50 pppoe,ppp,info pppoe-ncf: authenticated
01:45:50 pppoe,ppp,info pppoe-ncf: connected
01:45:50 pppoe,ppp,error could not add address: already have such address (6)
01:45:50 system,info dns changed
01:45:50 pppoe,ppp,info pppoe-ncf: terminating… - could not add address: already have such address (6)
01:45:50 pppoe,ppp,info pppoe-ncf: disconnected
01:45:51 pppoe,ppp,info pppoe-ncf: initializing…
01:45:51 pppoe,ppp,info pppoe-ncf: dialing…
If I check /ip address, I can see my pppoe-ncf IP is still in there, but in disabled (invalid?) state:
2 I 206.XX.YY.ZZ/32 67.XX.YY.ZZ pppoe-ncf
so the pppoe client will never restore the connection without manual intervention.. either a reboot, or by manually deleting that stale address.
Also /ip route gets filled with stale default routes.
IF I add auth-user-pass to my linux openvpn client script (I was missing it by mistake), then this problem doesn’t occur.
It seems somehow the OpenVPN server connect code is corrupting the PPPoE client state and kicking off this infinite loop of sadness 
I wonder if anyone can suggest a workaround in case I inadvertently trigger this condition again (I will not neglect to use auth-user-pass, as I understand it is required by the mikrotik openvpn server port – I was working towards using client certs and thought the password might not be required, but some reading showed it was).
Additionally, this is kind of a serious DoS vector, isn’t it? I can effectively knock my router off the internet by simply initiating an OpenVPN session with it.
Please let me know what would be required to debug.