Hello,
In the past, I made a mistake : using the same 10G network to run proxmox cluster AND Ceph cluster. I know, that's bad practice.
Now, I'm trying to migrate corosync (aka pvecm) on a dedicated 1G link, recently added.
All my hosts are configured on a new subnet (10.152.13.0/24) and ping each other. For reference Ceph network is on 10.152.12.0/24 subnet.
I started to follow 'Separate After Cluster Creation' on the cluster manager page on the wiki.
Finally ended up with a classical error : "nodelist or quorum.expected_votes must be configured!".
If I understand correctly, it's because the 'name' part of the corosync.conf file should be coherent with the ring0_addr for the node declaration. Right ?
But as I'm running Ceph (mon and osd) with the same hostname, if I change the name of each node in the /etc/hosts file, my Ceph cluster will probably break in the future when restarting an osd or monitor.
Does I have to change each corosync node name to point to a specific IP/host couple ? Will it break my /etc/pve/nodes/ directories ? That would be a complete mess right ?
This is a 5 nodes cluster still running PVE5 with 2 new nodes waiting to join and finally upgrade to PVE6.
Thank for your advices,
Julien
In the past, I made a mistake : using the same 10G network to run proxmox cluster AND Ceph cluster. I know, that's bad practice.
Now, I'm trying to migrate corosync (aka pvecm) on a dedicated 1G link, recently added.
All my hosts are configured on a new subnet (10.152.13.0/24) and ping each other. For reference Ceph network is on 10.152.12.0/24 subnet.
I started to follow 'Separate After Cluster Creation' on the cluster manager page on the wiki.
Finally ended up with a classical error : "nodelist or quorum.expected_votes must be configured!".
If I understand correctly, it's because the 'name' part of the corosync.conf file should be coherent with the ring0_addr for the node declaration. Right ?
But as I'm running Ceph (mon and osd) with the same hostname, if I change the name of each node in the /etc/hosts file, my Ceph cluster will probably break in the future when restarting an osd or monitor.
Does I have to change each corosync node name to point to a specific IP/host couple ? Will it break my /etc/pve/nodes/ directories ? That would be a complete mess right ?
This is a 5 nodes cluster still running PVE5 with 2 new nodes waiting to join and finally upgrade to PVE6.
Thank for your advices,
Julien