Skip to content

DHCP broken in Setup with multple bridges

Hi,

After an update from 5.4.4 to 6.1.0, DHCP stopped working for two of three bridges.

My system is a Turris Mox AEE (https://mox-configurator.turris.cz/AEE), configured with three different bridges on multiple interfaces and two WAN interfaces.

  • WAN1: IPv4 Uplink on eth0.7 (pppoe)
  • WAN2: IPv6 Uplink on eth1,lan1 (dhcp)
  • Bridge 1 - LAN spanning over eth1,lan2 - 12
  • Bridge 2 - SHIT (IoT-Devices and stuff with irregular updates) spanning over eth1.lan13 and lan 14
  • Bridge 3 - PVNET (Industrial IT) spanning over ethe1.lan15 and lan16

Each bridge has it's own ip range and runs a DHCP-Server.

In the Firewall-config LAN is allowed to reach SHIT, PVNET and WAN, SHIT and PVNET are only allowed to reach WAN. In addition, LAN, SHIT and PVNET are allowed to perform DHCP/DNS/NTP-Requests at the routers IP-Address.

After migration from 5.4.4 to 6.1.0 I figured that devices connected to the SHIT and PVNET bridges cannot be reached anymore.

I can see in /tmp/dnsmasq.log that devices from these networks send a DHCPREQUEST and the log also states they are answered with an DHCPOFFER but there are no DHCPACK messages for these networks in the log file. Clients in these networks repeat to send DHCP-Requests, and the files '/tmp/dhcp.leases' or /tmp/dhcp.leases.dynamic never liste these clients. DHCP seems to work for the LAN-Bridge, though.

dnsmasq -C /var/etc/dnsmasq.conf.<random> --test Reported a valid syntax.

  • /etc/init.d/dnsmasq restart was without success
  • rebooting the device was without success
  • changing and applying a different config to the bridge was without success

After a rollback to 5.4.4 DHCP was working normally again.

BR Dustin