create cluster faild

[hamada omara]

hi
I have 2 real server and the first one have old vm and I create cluster in it .
and new one try to add it to the cluster and I get error and told me run this code to solve the error



can any one help me Please .

thank you .

 

[t.lamprecht]

Can you post the steps with which you wanted to add a new node?

It seems your corosync config is empty/wrong.

Post the output from

cat /etc/pve/corosync.conf

from the node where you created the cluster and

pvecm status

from both.

And remember that you have to add the new node to the cluster through the new node, e.g.:

node-A: pcevm create clustername
node-B: pvecm add <IP-from-node-A>

 

[hamada omara]

I try and try with my self and try to solve this but no way
this server is online server

the main cluster





***------------------------------------------------------------------------------------------------------------------------
the add node





root@Diskcodevm-2 ~ # pvecm status
Cannot initialize CMAP service
root@Diskcodevm-2 ~ #





and I can login for 5 s only see that

 

[t.lamprecht]

I try and try with my self and try to solve this but no way
this server is online server

Hetzner? Do you have reliable multicast working between your two nodes? You can check with: https://pve.proxmox.com/wiki/Troubl...luster_issues#Diagnosis_from_first_principles
And with a two node Cluster you could also switch to unicast with no loss of performance/reliability.

[EDIT] This section contained wrong information of a corosync.conf setting regarding a cluster identification ID, sorry for that, accused the wrong thing.[/EDIT]

Apart from this you want to run a IPv6 multicast cluster, have you also configured the /etc/hosts files accordingly? Can you post its content?

Also how could you even add a node with this? Did you used "--force"?
As of the missing /etc/pve/authkey.pub thats another problem, maybe a result of the above, hard to tell.

Call pveum once over CLI as it should regenerate this key if not existing.

The output from:

systemctl status pve-cluster

would also be neat.