1st hardware upgrade of a pve node

[mimizone]

Hello,

I need just confirmation before I do our first hardware upgrade of one of our PVE node in the cluster.
My understanding is that removing a node from a cluster is something that should be done carefully.
I prefer to ask again before I create explosions in our infrastructure

The node needs to be fully decommissioned because we need to install new IO devices in the server and change RAID controller settings.

So I was looking at doing:
- move all workloads out of the node (done)
- remove replication jobs (none)
- stop the node
- remove the node from the pve cluster : pvecm delnode XXX
- reinstall proxmox on the new hardware
- rejoin the cluster

note that PVE is version 7.2.3

 

[mimizone]

I'll add one question: where can I download the older 7.2.3 in case...

 

[jsterr]

I would update to latest pve BEFORE. Also make sure to readd the host under a new name + new ip. For the rest check the docs:

1. https://pve.proxmox.com/pve-docs/pve-admin-guide.html#_remove_a_cluster_node
2. https://pve.proxmox.com/pve-docs/pve-admin-guide.html#pvecm_join_node_to_cluster

Its also useful to backup /etc/network/interfaces

Edit: If you reinstall with same name and hostname checkout this note from https://pve.proxmox.com/wiki/Cluster_Manager#_remove_a_cluster_node

After removal of the node, its SSH fingerprint will still reside in the known_hosts of the other nodes. If you receive an SSH error after rejoining a node with the same IP or hostname, run pvecm updatecerts once on the re-added node to update its fingerprint cluster wide.

 

[mimizone]

Thanks for confirming the steps.

Seems like I'll have to upgrade to a more recent version anyway cause I can't find 7.2.3 anymore anywhere.

I am planning on using the same IP/hostname and understand I'll need to refresh the SSH fingerprint/certs everywhere.
That's the only caveat right?

This point about not reusing the hostname/ip has always been scary to me because I don't understand what else may be broken under the hood if we do reuse them. It's a bit cumbersome to have to rename, and eventually change IP too. Hardware names are often "static", like based on their position in the rack for instance. IP is a bit more flexible, but could be also limited in a rack depending on how IP addressing is done.

My guess is that the point to be careful about is to make sure the node is fully removed from the cluster, reinstall and refresh the cert/keys so they all can talk/vote again?

 

[jsterr]

Thanks for confirming the steps.

Seems like I'll have to upgrade to a more recent version anyway cause I can't find 7.2.3 anymore anywhere.

I am planning on using the same IP/hostname and understand I'll need to refresh the SSH fingerprint/certs everywhere.
That's the only caveat right?

This point about not reusing the hostname/ip has always been scary to me because I don't understand what else may be broken under the hood if we do reuse them. It's a bit cumbersome to have to rename, and eventually change IP too. Hardware names are often "static", like based on their position in the rack for instance. IP is a bit more flexible, but could be also limited in a rack depending on how IP addressing is done.

My guess is that the point to be careful about is to make sure the node is fully removed from the cluster, reinstall and refresh the cert/keys so they all can talk/vote again?

Yeah its definetly doable to reinstall it with the same nodename and IP. Before cluster join I would refresh the certs. Then THIS should NOT be needed: https://www.oreilly.com/library/vie...05/30851255-6ac6-447d-bc87-19cb50e1c1a6.xhtml