RHEL7 cluster not failing over when the network link is down (cable unplugged)
 

Hi guys,

I created a 2 node cluster where in they share one vip. it is active standby setup. if I use 'ipdown' on the network interface card; the VIP and other clustered services will be moved to the other. This is what we want.

But we faced a problem when the NIC is up but we yank the network cable out or the network link is dropped for the active node; the VIP is not switched to the other and remains. Note that there is a 2nd network interface used as a dedicated heartbeat network between the two (10.0.0.1 and 10.0.0.2).

I found this page from rhel support but i wasn't able to get it to work:
https://access.redhat.com/solutions/2139231

Can you send me to a direction on where I can find a solution?

So I found a guide online:

https://access.redhat.com/solutions/2049063

and it suggested to create a ethmonitor on both the nodes:

# pcs resource create eth1-monitor ethmonitor interface=eth1 --clone

and then create a constraint. I did this. but then all the other resources will not start...

Had you implement fencing ?
Without fencing, cluster would not work well.

Hi,

we found the solution: (ens33 is the data network interface)

# pcs resource create ens33-monitor ethmonitor interface=ens33 --clone
# pcs constraint location <cluster_group_name> rule score=-INFINITY ethmonitor-ens33 ne 1

we use the following command to simulate a link is down (not an interface is down).
ip link set ens33 down

Hi,

we found the solution: (ens33 is the data network interface)

# pcs resource create ens33-monitor ethmonitor interface=ens33 --clone
# pcs constraint location <cluster_group_name> rule score=-INFINITY ethmonitor-ens33 ne 1

we use the following command to simulate a link is down (not an interface is down).
ip link set ens33 down