Node IP changes within a cluster

[d3dl3g]

I'm planning to re-jig my cluster tomorrow, grouping similar pieces of homelab kit into different adjacent IPs. so for this particular job id be looking at moving 2 of my nodes from IPs 0.2 and 0.6 to 0.110 and 0.111
I'd be looking to modify

/etc/network/interfaces
/etc/hosts

my resolve.conf only has `search home` and the IPs of my DNS Sinkholes so i assume unless im changing sinkhole IPs i wont need to touch it

Would it be easier/best practice to recreate cluster after changing PVE IPs or modify a file to the new IPs?

if modify, which file(s) should i be looking at?

 

[abobakr]

Hello,
My suggestion would be to add the new IPs as [1] redundant links to the existing cluster. Then verify the setup as described in the documentation by temporarily disconnecting the old link on one node and ensuring that its status remains online while disconnected. In any case, it's recommended to use a separate network for Corosync [2][3], as it requires consistent low latency. A dedicated 1 Gbit NIC is sufficient in most situations.

[4] As the final step, plan for the removal of the old IPs if necessary. Depending on your setup, the IPs will need to be changed in the following locations:
/etc/network/interfaces (on each node)
/etc/hosts (on each node)
/etc/pve/corosync.conf (only needs to be updated on one node)

Be sure to create a backup before making any changes, so you can easily revert in case something goes wrong.

How many nodes you have in the cluster? 2 node? if so I would suggest to setup a QDevice before performing this changes.
Without a QDevice you risk loosing network quorum, Once a node loses quorum, the cluster filesystem at /etc/pve will become read-only and will prevent you from doing most operations in Proxmox VE. Please take a look at the documentation for more information about QDevice [5].

(If you are using HA, it's better to stop it during this operation. The reason is that when a node hosting HA services loses Corosync quorum for around 60 seconds, it will fence itself so that its HA resources can be recovered on another node that still has quorum after 120 seconds.
Please refer to [0] for more information about fencing).

[0] https://pve.proxmox.com/pve-docs/pve-admin-guide.html#ha_manager_fencing
[1] https://pve.proxmox.com/pve-docs/chapter-pvecm.html#_adding_redundant_links_to_an_existing_cluster
[2] https://pve.proxmox.com/pve-docs/chapter-pvecm.html#pvecm_cluster_requirements
[3] https://pve.proxmox.com/pve-docs/chapter-pvecm.html#_separate_cluster_network
[4] https://forum.proxmox.com/threads/change-cluster-nodes-ip-addresses.33406/
[5] https://pve.proxmox.com/pve-docs/pve-admin-guide.html#_corosync_external_vote_support

 

[abobakr]

I just realize it's been more than a month since this thread started and its mentioned that you would do it within a day, @d3dl3g any observation from your side?