A question about heart beating lancard in Serviceguard

[Wang,MinJie]

Hi all
I have a cluster and I want to test how heart_beat lancard work
The following is the configuration of the both lancard:
system A:192.1.1.1
system B:192.1.1.2
And I run a pkg on system A and I "ifconfig lancard down" to disable the heartbeat lancard on system A
but there is no response
Technically SystemA should restart and system B will take over the pkg
What's wrong?

[sajeer_2]

Have you configured all the interfaces to carry Heart Beat?.In that case when you disable primary HB card,the serviceguard might be using the other active interfaces to carry HB.
check the how many interfaces are configured with HEARTBEAT_IP in cluster ASCII file.

sajeer

[RAC_1]

Is package switching enabled? do you have heartbeat timeout set correctly? Does show any error messages/information in syslog.log?

[freddy_21]

you must check about ur cluster ascii file.
soem time configuration using redudant for heart beat
example ascii file :

NODE_NAME db
NETWORK_INTERFACE lan0
HEARTBEAT_IP 192.168.4.24
NETWORK_INTERFACE lan4
NETWORK_INTERFACE lan3
HEARTBEAT_IP 16.178.0.2
FIRST_CLUSTER_LOCK_PV /dev/dsk/c12t0d2

it's means you have redundancy heartbeat. if your heartbeat lan problem, your cluster will use primary lan for heartbeat. So your package will not failed over. you can try to unplug primary lan and heart beat lan. your package will be failed over.

thanks
freddy

[melvyn burnard]

Using the ifconfig down command is NOT a valid test for Serviceguard, as the link is still up and traffic is still seen on the LAN, you need to physically remove the cable, or turn off the switch port that it is connected to for a valid test

[Michael Steele_2]

Right to not using ifconfig for testing heartbeat failover. Just pull the lan cable. Use 'cmscancl' to get more information about what's failing. Also note that the heartbeat is MAC address dependent and not IP address. Are you using a null modem cable / direct connection or going through a switch?