Secondary DHCP server

I have two Balance 20X’s on my network, and each is configured as a DHCP server. (Usually when I mention that, there pushback that this should not be two servers on the network, but the RFC specifically calls this out as a good thing for reliability.)

However, it would be more convenient if one of the units served the requests most of the time. It seems this is a concept of having primary and secondary servers. The secondary servers delay responding, or do not respond unless they don’t see a response from a primary server.

I don’t see support for this in the UI. Am I missing something?

Do I need to make a feature request for this? It would seem to be pretty simple to add a delay if configured as secondary, or not respond to the first request seen.

No. this is not supported. Easiest way is to set one DHCP server to dish out the first half of the range and the other to do the 2nd half.

Yes you do.

Moved it to Feature requests for you.

Thanks, I had completely missed that that forum existed.

Easiest way is to set one DHCP server to dish out the first half of the range and the other to do the 2nd half.

That’s more or less what I do, but I also have reservations for most devices, for which the CSV option makes it easy to copy the lists between the two.

Barry: I’m curious as to the use case for two DHCP servers. I’m also wondering if those same routers placed in HA mode might not get you where you want to go. HA is supported with that model and works well.

1 Like

See, pushback.

High availability is a completely different can of worms, which doesn’t work for my use case, See: Peplink | Pepwave - Forum

High availability only works with a primary unit, and the backup unit becomes useless, until called into action by VRRP.

My primary goal is a network which continues to function if any one component goes down, the network may be unattended for long periods (weeks), so this has to happen automatically, or remotely, preferably automatically. I’m also looking to have two fully functional routers, each which have their own usual jobs, not least as WiFi access points, until the other goes down. It would be a perfect use case, if Peplink did high availability in what seems like a sensible manner to me. (See Sample Configuration 2 in the RFC.)

In the absence of high availability, I’m using the multiple DHCP server approach. Each serves a common pool of allocations for most devices, and from non overlapping pools of addresses for devices I never added to the allocations. Each specifies itself as the router, so if the other goes it’s clients still have somewhere to send their traffic. I’ve using a couple of VWANs to route outbound traffic via the other connection, if necessary via the other Balance unit.

In the absence of high availability, any devices unfortunate enough to be attached to the wrong router when it goes down have to wait for their lease to expire, then when looking for a new lease, they’ll find the other Balance unit, and start again. The lease is 2 hours, so there will be downtime, but things will eventually recover (automatically).

The only device I’ve found which can’t live with that is a Google Chromecast with Google TV. That’s very disappointing that the Android platform seems to have such a problem.

For reference RFC 2131 about DHCP has this to say in section 1.6 Design goals:

A DHCP client must be prepared to receive multiple responses
to a request for configuration parameters. Some installations
may include multiple, overlapping DHCP servers to enhance
reliability and increase performance.

So I have no idea why people recoil in horror whenever I mention this concept.

I even got told, not for the first time,

Please note that there can only be one DHCP server (one router) on the network,

In my latest support ticket. It gets tiresome.