First Hop Redundancy Protocol: VRRP
First hop redundancy protocols are important to any network, after all why should any carefully designed network be dependent on any single IP address or any single device for that matter? A FHRP allows us to make the default gateway for any subnet a virtual address that can float between multiple physical devices. However some of our FHRP choices are Cisco proprietary (IE: HSRP & GLBP), and recently I had to setup a FHRP between a Cisco router and Red Hat box, so VRRP was my FHRP of choice.
Now the configuration for VRRP for the Cisco side:
The VRRP 5 commands are what makes this happen. VRRP 5 is the VRRP group (and the router ID) this instance belongs to, next we have VRRP 5 IP 10.0.1.1 this where we specify the virtual IP address. This is going to the IP Address we want to ensure is available to our clients such as a default gateway. The VRRP 5 timers learn command tells this router to learn the VRRP advertisement timers from the master router. Next we have the VRRP 5 preempt delay minimum 60 this specifies a set of time in seconds before the master role is taken back.
Delaying the preempt may not appear to be inviting at first but it is useful. Let’s say the master comes back online and it resumes the role of the default gateway before it has had time to re-converge it’s routing table, well the VRRP master is now going to black hole traffic until your routing protocol re-converges.
Next we have VRRP 5 priority 150 this allows us to specify the priority of each router to control who will be the master router and who will be the backup router. The device with the highest priority is the device that will be the master router for VRRP. The final line in this configuration is the VRRP 5 track 1 decrement 100 command, this allows us to tie some type of tracking or IP SLA monitoring into the VRRP configuration. This way if the IP SLA or tracking mechanism goes down the VRRP priority will automatically be decremented ensuring this device will not be the master. After all, if the master VRRP router looses connectivity to the rest of the network due to an interface failure or link failure we don’t want it to be the master router anymore.
Let’s also go a step further an think about the preempt delay we spoke about earlier, lets say we set the preempt delay to 60 seconds and we are tracking the outside interface the preempt delay will make sure the outgoing interface is stable for at least 60 seconds before taking back its master role, now that preempt delay isn’t looking so bad now is it?
Now let’s verify our VRRP configuration using the show vrrp command:
This will show the current state of this router in the VRRP process, that state will either be Init, backup, or master. The Init state just mean VRRP is not running or is starting up. Next we see the virtual IP and the virtual MAC address, a VRRP virtual MAC address will start with 00:00:5E:00:01:XX where XX is going to be the Virtual Router ID (or VRRP process ID) Then we will see the preemption settings and our advertisement intervals. So the Show VRRP command provides us a nice summary of our VRRP configuration.
Now let’s check out some debug logs for VRRP:
The first message in there states the router received an advertisement with a priority of 150 from IP 10.0.1.92, then the VRRP event states that this priority is higher or equal to it’s own priority, the next advertisement has a priority of 0, and then the Master down timer expires, which causes the this router to become the master. The last few VRRP message show advertisements going with checksums, which is normal for VRRP operations.
I also figured I would show a VRRP Packet:
From here we can see the VRRP version currently running on the router. This will also show us the Virtual Router ID, and the IP address we are advertising along with the priority of the 10.0.1.92 host. Also check out the Layer 2 information Wireshark identifies that the MAC address is associated with VRRP and knows the Router ID is 5. Now check out the Layer 3 IP destination, it’s 224.0.0.18 so these advertisements are sent out via Multicast to all VRRP routers on this LAN.
More information on VRRP can be found in RFC 3768, note this RFC is for VRRPv2. RFC 5798 covers VRRPv3, but this post covers VRRPv2 running on Cisco routers. Cisco also has few articles to go over the configuration: VRRP & Configuring VRRP
Good writeup. I would add the RedHat was CentOS 6.2 running keepalived-1.2.2-2.el6.
Steve Clark
March 12, 2012 at 9:13 AM
[…] covered VRRP a few weeks back, which is a vendor neutral FHRP, but now let’s take a look at HSRP which is […]
Let’s configure HSRP – Hot Standby Routing Protocol. « Another networking blog
May 24, 2012 at 7:03 AM