lltstat is unable to resolve the mac address of the llt interface due to a network / switch configuration problem.
systemA:~# lltstat -nvv|more
LLT node information:
Node State Link Status Address
* 0 systemA OPEN
ce4 UP 00:14:4F:44:88:51
ce10 UP 00:14:4F:67:6E:BB
1 systemB OPEN
ce4 UP 00:14:4F:67:5B:ED
ce10 DOWN 00:14:4F:44:88:1F
systemB:~# lltstat -nvv|more >>>>>> the MAC address is not showing here
LLT node information:
Node State Link Status Address
0 systemA OPEN
ce4 UP 00:14:4F:44:88:51
ce10 DOWN
* 1 systemB OPEN
ce4 UP 00:14:4F:67:5B:ED
ce10 UP 00:14:4F:44:88:1F
Specify the MAC address of the LLT interface directly in the /etc/llttab file.
Here are the procedures to do that without affecting the running applications. Let us assume this is a two node cluster (systemA and systemB) running a failover type of application.
1. Find out what is the MAC address of each of the LLT interface, by running the following commands on each node:
# /opt/VRTSllt/getmac /dev/ce:4 >>>> this will give you the mac address of interface ce4
# /opt/VRTSllt/getmac /dev/ce:10 >>>> this will give you the mac address of interface ce10
2. Once the MAC address is known, then modify the /etc/llttab on each node and add the last 6 lines as follows:
* on node systemB (Node 1)
set-node systemB
set-cluster 88
link ce4 /dev/ce:4 - ether - -
link ce10 /dev/ce:10 - ether - -
set-bcasthb 0
set-arp 0
set-addr 0 ce4 00:14:4F:44:88:51
set-addr 0 ce10 00:14:4F:67:6E:BB
!!Warning: Only set address for peer node, and should not set local node address manually. local node address will be automatically discovered when starting LLT. (Need to add set-verbose 1 to see the message for finding local mac address)
* on node systemA (Node 0)
set-node systemA
set-cluster 88
link ce4 /dev/ce:4 - ether - -
link ce10 /dev/ce:10 - ether - -
set-bcasthb 0
set-arp 0
set-addr 1 ce4 00:14:4F:67:5B:ED
set-addr 1 ce10 00:14:4F:44:88:1F
3. Freeze all the service groups in the cluster:
# haconf -makerw
# hagrp -freeze
# haconf -dump -makero
4. Stop 'had' and unload gab and llt modules:
In any node, run the following:
# hastop -all -force
Once 'had' has stopped in both nodes run the following in both nodes:
# gabconfig -U >>>>> capital case 'U'
Once gab has been unconfigured in both nodes run the following in both nodes:
# lltconfig -U >>>>> capital case 'U'
5. Restart llt, gab and 'had':
Once llt has been unconfigured in both nodes then, in both nodes run the following to start llt again:
# lltconfig -c
and in both nodes in the cluster, run the following to start gab again:
# gabconfig -c -n2
and finally, restart the 'had' again...
Run the following on each of the nodes in the cluster:
# hastart
6. Finally unfreeze the service groups above:
# haconf -makerw
# hagrp -unfreeze
# haconf -dump -makero