Scenario:
PPPOE-Server <—> Highsite <—> Highsite <—> Highsite with AP
PPPoe Server is X86, ROS v6.27 running in a VMware environment.
Highsite is RB mipsbe ROS v6.27 connected to PPPoe server via wireless backhaul link(s).
All access points are UBNT devices in bridge mode, WDS enabled.
Highsite interface is bridged and tunneled to the PPPoe Server
Client is mikrotik mipsbe device such as SXT or groove.
I have a problem that occurs when a client connects to one of our highsites using a mikrotik device and they have mrru=1600 set in their pppoe-client. It causes our main pppoe server to reboot as soon as some throughput occurs on the PPPoe tunnel. Workaround is to set mrru=disabled on the client but that is exploitable by client.
Seems a huge exploit hole in ROS.
Can someone provide a reason for why it occurs?
Is it related to the bridging?
Configuration on Highsite:
/interface eoip add local-address=y.y.y.y name=tunnel_ap1 remote-address=z.z.z.z tunnel-id=11
/interface bridge add name=bridge_ap1 protocol-mode=none
/interface bridge port add bridge=bridge_ap1 interface=tunnel_ap1
/interface bridge port add bridge=bridge_ap1 interface=“ethernet_ap1”
Configuration on PPPoe Server:
/interface eoip add local-address=z.z.z.z mtu=1458 name=ppp_ap1 remote-address=y.y.y.y tunnel-id=11
/interface pppoe-server server add default-profile=internet disabled=no interface=ppp_ap1 max-mru=1480 max-mtu=1480 mrru=1600
/ppp profile add change-tcp-mss=yes dns-server=a.a.a.a local-address=b.b.b.b name=inet use-compression=yes use-encryption=yes use-vj-compression=yes
Any helpfull explanation / discussion welcomed.