lanfailover testing

[Sreejith Kaliyam]

hi

I am using hp11.31 bl860C servers 2node mc sericeguard cluster.
service guard version 11.19
config stationary IP on lan0
standby lan3
heartbeat IP on lan2 and standby lan1

I am testing lan failover scenario.
using ifconfig lan0 down ip switched to standby lan3.
but if I give ifconfig lan2 down it is rebooting the second node.

can any one tell me how to validate network failover in blade servers cluster.

also I noted one thing that if I disable both public lan card where I enabled IP monitoring to gateway, eventhough I am not able to ping to gateway from server the cluster is working ok.

pls help.

[Steven E. Protter]

Shalom,

Blades all get their networking via a NIC card in the blade enclosure.

A valid testing scenario would involve setting up stand by connections and then physically disconnecting the network cable from one of the network ports on the blade enclosure.

SEP

[SoorajCleris]

Hi,

The node is rebooting because NODE_FAIL FAST is enabled.

May be your hearbeat standby not working properly. Try to increase the NODE TIMEOUT parameter and do that.

Regards,
Sooraj U

[Sreejith Kaliyam]

Hi

I cant do the test of removing cable from the enclosure because this is a virtual connect configuration sharing with other servers.

I will check the node time out value. Can you please tell me if ifconfig lan* down is a valid option to test the service guard.

rgds

[SoorajCleris]

Hi,

The network port which we see is UP Link port of swithes not the connctions from the servers. So untill and unless you have a patch pathpannel (very rare) installed instead of a an ethernet switch blade, removing cable is not possible.

I dont know if we have an option of desabling downlink port of the switch from OA.

Regards,
Sooraj

[Sreejith Kaliyam]

hi

I can disable profile of the server and nic is switching, but the IP monitor is not working. I am looking for some test case from HP.

rgds

[sujit kumar singh]

Hi

can you please post the cluster config ascii file as well as the packahge config file.

make sure that in pkg config file you have

NODE_FAIL_FAST_ENABLED and
SERVICE_FAIL_FAST_ENABLED as set to NO

also


Set LOCAL_LAN_FAILOVER_ALLOWD to YES
regards
sujit

[pz7rx8]

There must be something else that has to be set up regarding local lan failover. I have a very simple setup. Serviceguard A11.19 on a BL870 blade. Lan0 is the primary and lan1 is the standby. If is stop the Enet module in interconnect bay 1 lan0 fails over fine to lan1. But if I just down the interface (ifconfig lan0 down) nothing happens. Both lan0 and lan1 remain down.