General-Foo linux, technology and more

Migrate resources depending on ocf:pacemaker:ping RA with Pacemaker

The job we want to accomplish is quite simple. Let's just say you have two nodes nodeA and nodeB on two different locations which both have a individual gateways. Now let's imagine the following scenario, gatewayA goes down, and resulting in this there is no more internet connectivity on nodeA, so ideally you would want the cluster to switch to the node which still has internet connectivity. Using the resource agent ping you can create such a rule in pacemaker.

There are 3 different resource agents to accomplish this tasks, but we are going to use just one specific.

--> ocf:heartbeat:pingd (Deprecated you shouldn't use this one anymore)
--> ocf:pacemaker:pingd
--> ocf:pacemaker:ping

Configure

First let's create the ping primitive in crm and create a clone of it, so it runs on both nodes. We are going to ping ip addresses outside of the network, so internet ip addresses. For each successful ping result the p_ping primitive will score 1 point which is going to be multiplied by 100. But we'll come back on scoring later on.

# crm
crm(live)# primitive p_ping ocf:pacemaker:ping params host_list="8.8.8.8 4.2.2.2" multiplier="100" dampen="5s" op monitor interval="60" timeout="60"  op start interval="0" timeout="60" op stop interval="0" timeout="60"
crm(live)# clone c_ping p_ping
crm(live)# commit

host_list the ip addresses that are going to be pinged separated by a space e.g "x.x.x.x y.y.y.y"

multiplier each successful ping will result in a score value of 1 which is then multiplied by the multiplier value, usually you should use 100 here.

If you look at the cluster now you should see something similar to this.

Online: [ nodeB nodeA ]

 Master/Slave Set: ms_drbd [p_drbd]
     Masters: [ nodeB ]
     Slaves: [ nodeA ]
 Resource Group: g_cluster
     p_fs       (ocf::heartbeat:Filesystem):    Started nodeB
     p_ip       (ocf::heartbeat:IPaddr2):       Started nodeB
 Clone Set: c_ping [p_ping]
     Started: [ nodeB nodeA ]