02-13-2017 12:50 PM - edited 02-13-2017 12:56 PM
I'm playing with Aruba OS 8.X in the lab.
I've succesfully configured a Mobility Master, but I'm having issues creating a cluster of Mobility Controllers.
I'm trying to follow the example on the Aruba Solutions Exchange, but my controllers keep reporting that they are L3 connected.
I would also like to setup a VRRP between the controllers in the cluster.
Here is a dwg:
Here are the commands I tried to use:
no paging #Group-level configuration cd /md/ns-lab/cluster-nh configure terminal lc-cluster group-profile cluster-nh write memory cd /md/ns-lab/cluster-nh/00:1a:1e:00:da:90 lc-cluster group-membership cluster-nh write memory /md/ns-lab/cluster-nh/00:1a:1e:00:90:40 lc-cluster group-membership cluster-nh write memory #Group-level configure cd /md/ns-lab/cluster-nh lc-cluster group-profile cluster-nh controller 172.30.97.10 controller 172.30.97.11 redundancy write memory
However I get the following error:
(mm-nh-p) [md] #logon 172.30.97.10 Last login: Mon Feb 13 09:49:05 2017 from 126.96.36.199 (ctrl-nh-1) [MDC] *#show lc-cluster group-membership Cluster Enabled, Profile Name = "nh-cluster" Redundancy Mode On Active Client Rebalance Threshold = 50% Standby Client Rebalance Threshold = 75% Unbalance Threshold = 5% Cluster Info Table ------------------ Type IPv4 Address Priority Connection-Type STATUS ---- --------------- -------- --------------- ------ self 172.30.97.10 150 N/A CONNECTED (Leader) peer 172.30.97.11 100 L3-Connected CONNECTED (Member, last HBT_RSP 39ms ago, RTD = 0.260 ms) (ctrl-nh-1) [MDC] *#
Also If I try to configure a VRRP IP I get
lc-cluster group-profile nh-cluster (mm-nh-p) [cluster-nh] (Classic Controller Cluster Profile "nh-cluster") #controller 172.30.97.10 priority 150 vrrp-ip 172.30.97.18 vrrp-vlan 376 (mm-nh-p) [cluster-nh] (Classic Controller Cluster Profile "nh-cluster") #controller 172.30.97.11 priority 100 vrrp-ip 172.30.97.18 vrrp-vlan 376 Error: vrrp-ip 172.30.97.18 matches with vrrp-ip for controller ip 172.30.97.10
This seems like this should work. Also, after I perform the configuration if I try to show the configuration profile I get:
(mm-nh-p) [cluster-nh] #show lc-cluster group-profile nh-cluster Classic Controller Cluster Profile "nh-cluster" undefined. (mm-nh-p) [cluster-nh] #
Any help would be appreciated.
Solved! Go to Solution.
03-14-2017 12:25 AM
I saw the same problem and was able to solve it. Reason was that there was a dummy vlan configured on both MCs but the vlan was not connected thru my network between both MCs. After I removed this vlan cluster was L2 connected.
This is the role:
As a follow-up to the cluster members in a full mesh state, the cluster members connection type become known as L3-Connected.
Once all cluster peers are in L3-Connected state, a mechanism called VLAN Probing is launched on each cluster member.
The Vlan Probing scheme on each cluster member consists of sending a specially crafted L2 broadcast packet on each configured vlan on the controller.
This vlan probing scheme is bidirectional and is performed between every pair of cluster members
04-05-2017 01:48 AM
If you want to use the VRRP IP address in a cluster which is a good idea if your controller fails and you still want the cluster to be reached by external radius servers (CoA packet by instance), you need two VRP IP addresses :
controller 172.30.97.11 priority 100 vrrp-ip 172.30.97.19 vrrp-vlan 376
172.30.87.18 is master on 172.30.97.10 and back-up on 172.30.97.11 and 172.30.87.19 is master on 172.30.97.11 and back-up on 172.30.97.10
a month ago
This remains the same logic, configure 3 VRRP IP addresses, configure each controller to be master for one of these VIP (CTRL1=VIP1, CTRL2=VIP2, CTRL3=VIP3) and for each VRRP group configure a standby. The configuration is made in the lc-cluster group profile not to confuse with vrrp config for VLAN IP interfaces. T Just to clarify, this is optional to provide a virtual IP address for radius servers (NAS-IP). AP however will use another mechanism to terminate active and standby GRE tunnels to 2 controllers within the cluster . There is no need to use a VRRP IP for LMS in 8.x, redundancy is implemented within the cluster.