Network failed in cluster

[kaltsi]

This is my first proxmox cluster. I looking for the best system for my company.

A built a 3 nodes cluster. Make few VM and add to HA.
The 2th node network cable unplugged.

Where can I confuguration in this situation that, all vm migrate to the other nodes which was runing the 2th node?

thx

 

[dietmar]

Where can I confuguration in this situation that, all vm migrate to the other nodes which was runing the 2th node?

You can define failover domains (see 'man rgmanager')

 

[kaltsi]

You can define failover domains (see 'man rgmanager')

Thank you for your prompt reply

  <rm>
    <failoverdomains>
        <failoverdomain name="appvirt-migrate" nofailback="0" ordered="1" restricted="0">
            <failoverdomainnode name="appvirt-01" priority="10"/>
            <failoverdomainnode name="appvirt-02" priority="20"/>
            <failoverdomainnode name="appvirt-03" priority="30"/>
        </failoverdomain>
    </failoverdomains>
    <pvevm autostart="1" vmid="100" domain="appvirt-migrate"/>
  </rm>
</cluster>

Must use "domain" variable in the "pvevm" tag or not necessary?

 

[dietmar]

Must use "domain" variable in the "pvevm" tag or not necessary?

Only if you want to use the failover domain.

 

[kaltsi]

Some mistake in setting.

When the node stops, will be not available the vm. (Error Connection error 595: No route to host)
But the VM not migrated to another node.
If this node is available again, migrating the vm.

Which part I make a mistake?

Here is the complete cluster file:

<?xml version="1.0"?>
<cluster config_version="8" name="appvirt-cluster">
  <cman expected_votes="3" keyfile="/var/lib/pve-cluster/corosync.authkey"/>
  <quorumd allow_kill="0" interval="3" label="proxmox_qdisk" tko="10">
    <heuristic interval="3" program="ping 10.1.11.150 -c1 -w1" score="1" tko="4"/>
  </quorumd>
  <totem token="54000"/>
  <clusternodes>
    <clusternode name="appvirt-01" nodeid="1" votes="1"/>
    <clusternode name="appvirt-02" nodeid="2" votes="1"/>
    <clusternode name="appvirt-03" nodeid="3" votes="1"/>
  </clusternodes>
  <rm>
    <failoverdomains>
        <failoverdomain name="appvirt-migrate" nofailback="0" ordered="1" restricted="0">
            <failoverdomainnode name="appvirt-01" priority="10"/>
            <failoverdomainnode name="appvirt-02" priority="20"/>
            <failoverdomainnode name="appvirt-03" priority="30"/>
        </failoverdomain>
    </failoverdomains>
    <pvevm autostart="1" vmid="100" domain="appvirt-migrate" />
  </rm>
</cluster>

 

[kaltsi]

I configured fencing device, and work fine.

Thx

 

[dietmar]

btw, I consider quorumd with heuristics quite dangerous.

 

[kaltsi]

btw, I consider quorumd with heuristics quite dangerous.

thx

My last question

Why are the light of servers red, in "Storage view"?
Because in all another view are green.

 

[tom]

your see a minor bug here. as long as the server view is green, everything is ok.