I made a mistake when I created a 5-node cluster. I added all nodes by IP to the cluster and `pvecm status` now shows:
The cluster is operational, the problem is that the IPs on the subnet 10.15.15.0/24 don't actually point to the host names of the nodes. They point to a different host name assigned to each host. This is because I want to separate the cluster traffic and the admin network traffic - they are on different physical interfaces.
The IPs and host names look like the table below.
Host names for the cluster/corosync IPs:
"Real" host names for the admin access network (10.15.2.0/24):
The latter is what `hostname -f` shows on the nodes. Consequently, Proxmox uses this hostname and the associated 10.15.2.0/24 network for any cluster operation aside from corosync, but this is not good, as this is not intended for cluster operations like migrations, replication, etc., and is a gigabit network instead of the 10G network for the cluster.
Now, my question is: can I simply change the host names so that pveX.whatever.local (hostnames without the trailing 'c' in their name) point to the cluster IPs as shown in `pvecm status`? Would that cause any issues? There is also a CEPH cluster configured on the nodes on the correct networks dedicated to it, but CEPH has its own separate network setting, so I don't hink it would be affected. Production VMs are already running on the cluster, unfortunately. Worst case I could schedule a full downtime.
Code:
Cluster information
-------------------
Name: cluster2
Config Version: 5
Transport: knet
Secure auth: on
Quorum information
------------------
Date: Mon Sep 25 14:42:15 2023
Quorum provider: corosync_votequorum
Nodes: 5
Node ID: 0x00000001
Ring ID: 1.4d
Quorate: Yes
Votequorum information
----------------------
Expected votes: 5
Highest expected: 5
Total votes: 5
Quorum: 3
Flags: Quorate
Membership information
----------------------
Nodeid Votes Name
0x00000001 1 10.15.15.11 (local)
0x00000002 1 10.15.15.12
0x00000003 1 10.15.15.13
0x00000004 1 10.15.15.15
0x00000005 1 10.15.15.16The cluster is operational, the problem is that the IPs on the subnet 10.15.15.0/24 don't actually point to the host names of the nodes. They point to a different host name assigned to each host. This is because I want to separate the cluster traffic and the admin network traffic - they are on different physical interfaces.
The IPs and host names look like the table below.
Host names for the cluster/corosync IPs:
Code:
pve1c.whatever.local -> 10.15.15.11
pve2c.whatever.local -> 10.15.15.12
etc."Real" host names for the admin access network (10.15.2.0/24):
Code:
pve1.whatever.local -> 10.15.2.11
pve2.whatever.local -> 10.15.2.12
etc.The latter is what `hostname -f` shows on the nodes. Consequently, Proxmox uses this hostname and the associated 10.15.2.0/24 network for any cluster operation aside from corosync, but this is not good, as this is not intended for cluster operations like migrations, replication, etc., and is a gigabit network instead of the 10G network for the cluster.
Now, my question is: can I simply change the host names so that pveX.whatever.local (hostnames without the trailing 'c' in their name) point to the cluster IPs as shown in `pvecm status`? Would that cause any issues? There is also a CEPH cluster configured on the nodes on the correct networks dedicated to it, but CEPH has its own separate network setting, so I don't hink it would be affected. Production VMs are already running on the cluster, unfortunately. Worst case I could schedule a full downtime.
