MikroTik CHR: use a routed /29 from your ISP, step by step
Some internet providers sell you a small block of public addresses, often a /29: eight addresses, of which six are usable. This page shows how it works and how to set it up on a MikroTik router. We built it in a lab and ran every step, including the step that fails if you skip the provider's part.
Tested on
Date: 16 November 2026 · Router: MikroTik RouterOS 7.12.1 (CHR, a virtual machine) · Provider and customer sides: Debian 12
Host: Proxmox VE 8.4.21. The router, the "provider" and the "customer" ran as throwaway test machines on private bridges with no outside connection. All addresses are documentation examples: 198.51.100.0/30 for the link and 203.0.113.0/29 for your block, not real ones.
What a routed /29 is
A routed block is different from a bridged one. With a routed block, the provider picks one address on your router, the uplink address, and sends the whole block to it. Nothing is "plugged into" the block at the provider's end. Your router has the block on its LAN port and hands the addresses to your machines.
That gives you three things to remember:
- The block has 8 addresses: the first is the network address, the last is the broadcast address, and one is your router's own LAN address. That leaves five for your machines in a /29 when the router takes the first usable address.
- Your router needs two addresses that have nothing to do with each other: the uplink address on the provider side (often from DHCP or assigned by the provider), and the first usable address of the block on your LAN side.
- The provider must have a route that says "this block is reached through your uplink address." Without it, nothing works, and you cannot fix that part yourself.
The lab
provider 198.51.100.1/30 --- ether1 [CHR] ether2 --- customer 203.0.113.2/29
uplink 198.51.100.2 (DHCP) LAN 203.0.113.1/29
The "provider" gave the router 198.51.100.2 by DHCP on its uplink port, which is how many ISPs hand out the uplink address. The "customer" machine ran in its own network namespace so its traffic had to cross the router.
Step 1: put the block on the router's LAN port
On the router, give the LAN port the first usable address of the block with the block's prefix length:
[admin@MikroTik] > /ip address add address=203.0.113.1/29 interface=ether2 comment=routed-29
[admin@MikroTik] > /ip address print
Columns: ADDRESS, NETWORK, INTERFACE
0 D 198.51.100.2/30 198.51.100.0 ether1
1 203.0.113.1/29 203.0.113.0 ether2
Use your own block and your own LAN port. The D on the first line means that address came from DHCP.
Step 2: set up a machine in the block
On the customer machine we used 203.0.113.2/29 with the router's 203.0.113.1 as its gateway:
$ ip route
default via 203.0.113.1 dev ens20
203.0.113.0/29 dev ens20 proto kernel scope link src 203.0.113.2
At this point the machine reaches its router, but not the provider:
$ ping -c2 203.0.113.1
2 packets transmitted, 2 received, 0% packet loss
$ ping -c2 198.51.100.1
2 packets transmitted, 0 received, 100% packet loss
This is the part that confuses people. The provider's machine received the ping fine, but it has no way to send the reply back to 203.0.113.2, because nobody told it where that block lives.
Step 3: the provider's route is the missing piece
From the provider side, before the route exists, the block is unreachable:
$ ip route get 203.0.113.2
203.0.113.2 via <default gateway> dev eth0
$ ping -c2 203.0.113.2
2 packets transmitted, 0 received, 100% packet loss
The provider machine has no route for the block, so it sends the packet out its default route (the address of our lab's default gateway is left out of the output above). It does not go to your router, so nothing comes back.
Step 4: the provider adds the route
This is what your ISP does when it turns on your block. In the lab the "provider" ran:
$ ip route add 203.0.113.0/29 via 198.51.100.2
$ ip route get 203.0.113.2
203.0.113.2 via 198.51.100.2 dev ens19 src 198.51.100.1
$ ping -c3 203.0.113.2
3 packets transmitted, 3 received, 0% packet loss
The reply has ttl=63: one hop, through the router. A traceroute shows the same:
$ traceroute -n 203.0.113.2
1 198.51.100.2 0.622 ms
2 203.0.113.2 0.623 ms
With a real provider you don't run this command; you ask them, or find the setting in their portal, and give them your uplink address. If you don't have a static uplink address, ask how they handle a routed block with a DHCP or PPPoE uplink. Some will only route a block to a fixed address.
Step 5: check the other direction, and more addresses
$ ping -c2 198.51.100.1 # from the customer machine, now works
2 packets transmitted, 2 received, 0% packet loss
$ ip addr add 203.0.113.3/29 dev ens20 # a second address from the block
$ ping -c2 203.0.113.3 # from the provider side, no further change
2 packets transmitted, 2 received, 0% packet loss
Any other address in the block, .3 to .6, works the same way with the router as its gateway.
What went wrong in our lab
- We locked ourselves out of the router. Our first attempt removed the DHCP client from the router's uplink to "set a fixed address". That removed the only address the router had, so we could no longer reach it, and we had to rebuild the test router. Add the static address first, and only then remove the DHCP client, or leave DHCP on if your provider gives the uplink address that way, as we did the second time.
- The first "ping fails" was the provider, not our router. The router and LAN side were correct from the start. The only thing wrong was the missing route on the provider's side. Check the order: LAN side, then router, then whether the provider has the route.
- A new CHR has an empty admin password. We left it that way inside a lab with no outside connection. On a real router, set a password before connecting it to anything.
Before you use this on a real network
This lab had no firewall and no address translation, so it only shows how the block is routed. On a real network, add your own firewall rules for the router itself and for traffic into your block. We did not test firewall rules for this guide, so we are not giving any here.