standby lan interface

[samuel fatoye]

Please can you help with my question on standy lans Two nodesc clusterh with 3 LAN interface cards. Want to make LAN 3 standy for LAN 0
LAN 1 & 0 will be used for heartbeat messages
I also two unallocated floating ip adress one for frontend network lan 0 & 3 , the other for lan 1 backend network.
Output of netstat -in
Name Network Address
lan3 10.16.13.0 10.16.13.131
lan1 10.16.14.0 10.16.14.46
lan0 10.16.13.0 10.16.13.121
lo0 127.0.0.0 127.0.0.1

My enteries for the cmquerycl -w full ascii file is
NODE_NAME NodeB
NETWORK_INTERFACE lan0
HEARTBEAT_IP 10.16.13.122
NETWORK_INTERFACE lan3
HEARTBEAT_IP 10.16.13.132
NETWORK_INTERFACE lan1
HEARTBEAT_IP 10.16.14.48


NODE_NAME NodeA
NETWORK_INTERFACE lan0
HEARTBEAT_IP 10.16.13.121
NETWORK_INTERFACE lan3
HEARTBEAT_IP 10.16.13.131
NETWORK_INTERFACE lan1
HEARTBEAT_IP 10.16.14.46

how should I amend the file to to lan3 the standy for lan0 , include the floating ipadress both the frontend and backend.
frontend ip 10.16.13.130
backend ip 10.16.14.40

Many thanks

[Marco Santerre]

if you want to have lan3 has a stanby lan card, you'll have to deconfigure lan3, cause what ServiceGuard will do, is that when it recognizes lan0 to have trouble, it will deconfigure lan0 and configure lan3 with the same IP address.

So lan3 shouldn't be configured.

after that, your ascii file should only have something like this :

NODE_NAME xxxx
#
# FRONT END
#
NETWORK_INTERFACE lan0
HEARTBEAT_IP 10.16.13.121
NETWORK_INTERFACE lan3
#
# HEART BEAT
#
NETWORK_INTERFACE lan1
HEARTBEAT_IP 10.16.14.46

that way lan3 will be recognized as the backup for your lan0.

[Michael Steele_2]

I wouldn't change a thing of the above.

Use cmscancl to get a report of what the cluster thinks is standby.

Cmquerycl determines active LANS and fills out the details automatically. LAN's that it thinks can handle the heartbeat are designated as the heartbeat LANs. Specifying -n node_name should be all you need.

As long as you've filled out your /etc/hosts file with the nodes and IPs correctly, and have used the cmquerycl -n node_name syntax correctly, then the cluster.ascii file should be filled out for you automatically.

[melvyn burnard]

step 1: deconfigure lan3 on the nodes
step 2: rerun your cmquerycl command and create the ascii file
step 3: Read the ascii file and confirm that in the comments below each nodes details that it TELLS you that lan3 is a possible standby for lan0


If it does NOT, then there is a network issue that needs to be fixed first.

[Sridhar Bhaskarla]

Hi,

Unfortunately, you cannot configure an IP address to the standby interface keeping in view the syntax of cluster configuration file You would not specify an IP address for the standby interface there.

The best way is to run cmquerycl, generate a fresh configuration file and then modify it to your needs.

cmquerycl -C cmclconfig.ascii -n node1 -n node2

Edit cmclconfig.ascii file and make changes.

-Sri

[Michael Steele_2]

Data LANS and Standby LANS are not designated during the cmquerycl operation, only Heartbeat LANS are. Call it a design error if you want but the author is providing you with knowledge that you have redundant options for carrying the Heartbeat over qualified LANS.

So by allowing all of the above to remain designated as Heartbeat LANS instead of changing them to STATIONARY_IP, which is your only other option, then becomes a naming convention decision and a knowledge of redundancy decision, for it doesn???t change the LAN???s ability to also carry the heartbeat nor data nor both simultaneously.

Specify your floating ip in the package and designate this as STATIONARY_IP in the cluster.ascii file. Use this as the final arbiter to decide which is not to be heartbeat.

BUT, to see what is currently designated as Heartbeat and Standby use the cmscancl and cmviewcl command.

Here is another link on the subject:

http://forums.itrc.hp.com/cm/QuestionAnswer/1,,0x9105670142b2d5118ff10090279cd0f9,00.html