Hi, cluster in Community Edition moves interfaces and starts the processes in the other node at the moment there is an issue in the MASTER but for a unique NIC card by default, first of all, you should ensure your farm is configured in a virtual interface in eth0 or ethX where the cluster is configured. For example:

eth0: NIC on node1

eth0 NIC on node2

eth0:VIP1 Virtual IP on the cluster where farm should be running.

Node1 runs eth0:VIP1 so MAC for eth0:VIP1 should be the same as eth0 NIC for node1. Now Node1 fails, the SKUDONET cluster will move the eth0:VIP1 to eth0 NIC on node2 and the SKUDONET service will start with the following command :

/etc/init.d/skudonet start 

Now you should ping eth0:VIP1 and you should see the MAC of eth0: IP on node2. Take into account that this ping should be done in the same network without passing through any gateway or the MAC for eth0:VIP1 will be the gateway MAC. 

Confirm your config is properly done and do the tests to confirm the MAC change is done and the SKUDONET service starts in node2 if node1 fails. 

Regards!

Hi @emiliocm, thanks for your recommendation. I have configured the VIP and use the VIP on the farm as well. When the Master is down, the Backup is taking over, but for the VIP I can't ping, and I can't access the web too. Maybe there is another way to keep the web up? Is it possible to have two virtual IP in the cluster?

Regards,

Hi @dzoelfi, you can configure as many virtual IPs as you want for the cluster (eth0) the service will start them on the other node as soon as it takes control. 

Can you extend further information about "the VIP you can't ping"? You detected that eth0:VIP is configured and up in the other node as soon as it takes the control. that is important because if you confirm it is, the issue seems more from the network. 

Can you share the cluster configuration file here? Please follow this article:

https://www.skudonet.com/knowledge-base/howtos/how-to-configure-a-cluster-in-skudonet-community-edition-v-7/

I can confirm it works because a community user recommended a few changes weeks ago for a better understanding. 

Thanks!

@dzoelfi any update? did you see my latest comment?

Hi @emiliocm, thank you for your explanation and sorry for my late reply. I will follow that documentation and try to reproduce again on different environment to see if my issue because of the environment network/limitation or something else.

Regards,

Hi @dzoelfi, thanks for the update, you have to identify if the issue is the load balancer or in your network.

To check if the issue is the load balancer.

-Please configure your cluster following my shared article, later stop the master node, the backup node should take the service configuring as UP the virtual IP. Please execute the command:

ip a show

It should show you the virtual IP in the new master node, previously the backup and you should see the switch action in the logs /var/log/syslog

To check if the issue is the network.

If the previous point is successful then please ping the virtual IP from the same network as the load balancer and you should see the new MASTER replies you, if not then your networking devices are not updating the new MAC for the virtual IP in the new MASTER node. 

Regards!