Network failover behavior... !?

[Warren_9]

Hi ALL,

I would like to clarify on the network failover behavior on the MC/SG.

I've got the following configure....

NODE_NAME NODE1
NETWORK_INTERFACE lan0
HEARTBEAT_IP 192.168.1.1
NETWORK_INTERFACE lan1
HEARTBEAT_IP 192.168.2.1
NETWORK_INTERFACE lan2
NETWORK_INTERFACE lan3

linkloop lan0 --> lan3 OK
linkloop lan1 --> lan2 OK

I've got problem when both lan0 and lan1 is down.... the ip binded to incorrect interface.
lan0 --> lan2
lan1 --> lan3

any idea?
do i need to change the configure as follow?

NODE_NAME NODE1
NETWORK_INTERFACE lan0
HEARTBEAT_IP 192.168.1.1
NETWORK_INTERFACE lan3
NETWORK_INTERFACE lan1
HEARTBEAT_IP 192.168.2.1
NETWORK_INTERFACE lan2

Thanks,
Warren.

[Mridul Shrivastava · Accepted solution]

Yes thats the way it shoul be configured.

Standby LAN interface should have no IP address assigned and this way lan3 will become standby interface for lan0 only and lan2 for lan1 only.

In the previous configuration lan2 and lan3 were standby interfaces for both the lan cards so at the time of failover the first available interface will be assigned the IP. This is the reason lan0 binded to lan2 since that was the first available interface.

I think the last one is better anyways other two interface can not communicate to each other so there is point in keeping lan2 and lan3 as standy for both the interface. Keep them separated one for one.

[Warren_9]

Thanks for the reply .... but i'm still very confuse on the cluster configure.

I use the cmscancl to check on the configure and all the 4 interfaces are in the same "bridged net ID" ...... i think that is the root cause of the incorrect failover, right??

I've got another cluster which having the same network setting and cluster configure, but the 4 network interfaces are in 2 difference "bridged net ID".

thanks & regards,
Warren

[Mridul Shrivastava]

Yes, you are right...

That is the reason it says that standby lan interfaces should be from same net ID.

Here cluster sees all 4 interfaces on same net ID hence in case of first setup it takes both standby interfaces for both the static IP addresses and assign them on availability at the time of failover.

If only one interface is available on that net ID then other standby interface would be ignored since that violates the criteria of standby interface.

[Warren_9]

will log call the HP to check why the Net ID are all the same....

[Mridul Shrivastava]

It has to do with the physical connections of the NIC cards. You need to make some changes in the physical connections so they are on different bridged net ID.

[Warren_9]

thanks for your information.

the cluster is build by HP and i think he did something wrong during the configuration.

When I using the cmscancl to check the cluster configure, the 4 interfaces are in 1 brigded net.

But if I use cmquerycl to check again, it show 2 brigded networks.