Lan Failover

[Gene Kornacki_2]

When I issue a ifconfig lan0 down I was under the impression SG would fail over to the second adapter. This is not happeneing and when I run cmviewcl -v the cluster thinks lan0 is still up. This is out of my config file.


NODE_NAME smelly
NETWORK_INTERFACE lan0
STATIONARY_IP xxx.xxx.xxx.xxx
NETWORK_INTERFACE lan2
NETWORK_INTERFACE lan1
HEARTBEAT_IP xxx.xxx.xxx.xxx

Each lan card is plugged into the same switch (for now), that shouldn't have anything to do with it?

What am I missing?

[John Poff · Accepted solution]

Hi,

Your configuration looks right. If you are trying to test the LAN failover I'd suggest unplugging the cable to the LAN card for lan0 and monitoring syslog (tail -f). You should see MC/SG switch to lan2, and then switch back when you plug lan0 in again.

JP

[Darrell Allen]

I don't ever remember trying to test failover by downing the interface. I always pull the switch.

You could also disable the port on your switch for lan0 to test a failover.

Being plugged into the same switch will not be the problem (and it sounds like you understand you should use separate switches if possible - single point of failure thing).

Darrell

[Darrell Allen]

Excuse me, I meant pull the cable, not switch.

Darrell

[Unix Administrator_5]

Hey Gene,
Dont know the answer to your question -- i'm still trying to figure out autoport aggregation

[Ashwani Kashyap]

since link level communication is still up , there will be no lan failover . To do so either pull out the LAN cable from the primary NIC card or temporary disable the port on the switch where your primary NIC card goes . This should initiate a local LAN failover . Check your package log file or syslog.los file to verify .

Use netstat -rn to actually verify that your package is now bounf on the standby NIC card .

[Gene Kornacki_2]

It's working. Thanks everyone. I'm enjoying this product the closer and closer I get it to production.

Gene