Automatic Failover when Power Supply off

dropped

New Member
Aug 30, 2014
2
0
1
Helo Guys!
I hope you can help my.
My configuration looks like a proxmox cluster with two nodes and enabled HA and fencing.
Live Migration over the webinterface works absolutly fine.
Migration by a controlled reboot via shell works fine, too.
All the VM's from Server1 will be migrate to Server2 by rgmanager :)

This is my problem:
how i can configure, that the migration do when the power supply or network cable pulled?
This is my current cluster.conf:

Code:
<?xml version="1.0"?>
<cluster config_version="8" name="cluster">
  <cman expected_votes="1" keyfile="/var/lib/pve-cluster/corosync.authkey" two_node="1"/>
  <fencedevices>
    <fencedevice agent="fence_ilo" ipaddr="192.168.129.129" login="root" name="fenceA" passwd="CHANGE-ME"/>
    <fencedevice agent="fence_ilo" ipaddr="192.168.129.130" login="root" name="fenceB" passwd="CHANGE-ME"/>
  </fencedevices>
  <clusternodes>
    <clusternode name="ha1" nodeid="1" votes="1">
      <fence>
        <method name="1">
          <device action="reboot" name="fenceA"/>
        </method>
      </fence>
    </clusternode>
    <clusternode name="ha2" nodeid="2" votes="1">
      <fence>
        <method name="1">
          <device action="reboot" name="fenceB"/>
        </method>
      </fence>
    </clusternode>
  </clusternodes>
  <rm>
    <pvevm autostart="1" vmid="100"/>
    <pvevm autostart="1" vmid="101"/>
    <pvevm autostart="1" vmid="102"/>
  </rm>
</cluster>

--
dropped
 
You use a special setup => "two_node = 1". That mean that both nodes assume they have
quorum if they loose connection to the other node.

This is an arbitrary assumption, which is dangerous in most cases, and lead to the observed behavior.
It allows a node to continue providing service when the other node dies.

As tom suggested, it is always better to use at least 3 nodes.
 
This is my problem:
how i can configure, that the migration do when the power supply or network cable pulled?

For above reasons, it is a bad idea to pull cables with such config, because both nodes assume
they have quorum. So please test by pulling power supply (or reboot node).

But you use fence_ilo, which badly fails when the other node has no power. I also fails
when you pull the network cable! Basically your fencing never works with the error scenarious
you are testing.

IMHO, your setup is totally unusable for HA.