dhcp carp and automatic sync
i am using 2 synchronised pfsense boxes with bunch of vlan interfaces and deploying machines using pxe on many different vlans.
i fail to understand how dhcp config is supposed to be replicated :
if i setup the "Failover peer IP" with the slave's ip address in the proper vlan, the synchronisation tool will setup the slave with a "Failover peer IP" corresponding to the first interface of the master which is in a different VLAN ( actually on my WAN interface )
not setting it at all actually does not create IP conflicts, but produces a disfunctionnal config since the overall setup is active-passive.
i assume it would work reasonably well if i manually set the CARP address as the gateway, though
i'm wondering if i am missing something here. is this a bug ? am-i misconfiguring something ? is there a different simple way to make sure that whichever member owns the carp address will answer to dhcp queries ?
for now, i only use the master for DHCP which is suboptimal but at least works in a predictable way.
thanks for your time
Sounds like your interfaces do not match between the two cliuster nodes. This is a requirement.
they do match
i'm using carp failover between each interface of both firewalls which definitely would fail to work otherwise. the WAN interface is the very first interface that i've been setting up and the one i'm working on is the last one.
the WAN interface is dedicated. other interfaces are all vlan interfaces built over a LAGG on each firewall. each vlan has it's own carp gateway. most vlans have their own dhcp servers which are only used for automated deployment with PXE.
i believe there is a bug in the replication code : the ip address of the slave's peer is probably automagically replaced with the first address of the master rather than the master's address for the same interface, in my lagg+vlan case at least
... and my setup is also buggy : i should definitely use the carp address as the gateway
what is your point ?
my carp setup does work. i have multiple machines in each vlan. no problem there. i can shutdown either firewall and unplug any cable without producing a mess. i had disabled pfsync in some previous tests which is why using the carp address as the gateway is required.
am i expected to configure a LAN interface as the first interface and use the LAN address as the failover peer for each of the dhcp server instances ? this would be meaningful indeed. but in that case, it may be worth to drop a line in the documentation and there is little to no point in setting the same address for each dhcp instance.