How to change the Proxmox IP address without locking yourself out (PVE 8 and 9)

Changing a Proxmox host's IP takes two files and one command. Getting it wrong on a remote server means a drive to the data center. Here's the safe way for a single node and for a cluster.

Share
Rack server with network cables in its ports, illustrating how to change a Proxmox host IP address safely.

Changing the IP address of a Proxmox host is a two-file job. On a box under your desk, a mistake costs you a few minutes with a keyboard and monitor. On a remote server, the same typo means you can't reach the web UI, SSH or anything else until someone gets console access.

So the real question isn't how to change the IP. It's how to change it without locking yourself out.

Short answer for a single Proxmox VE 8 or 9 node: update the bridge address in /etc/network/interfaces, update the host's line in /etc/hosts, apply with ifreload -a, and arm an automatic rollback first so a bad config undoes itself. If the node is part of a cluster, you also have to update its address in /etc/pve/corosync.conf, and that changes the order of the steps.

This guide covers both, for Proxmox VE 8 and 9 (PVE 9 is based on Debian 13 and was released in August 2025). The same steps work on PVE 7.

Before you change anything

Three checks save most of the pain:

  • Have a way back in. IPMI, iDRAC, iLO, a KVM, or a monitor and keyboard. If you have none of these and the server is remote, the rollback timer below is not optional.
  • Know the new values. IP with prefix (for example 192.168.10.20/24), gateway, and whether the new IP is on the same subnet as the old one. A new subnet usually means a new gateway too.
  • Check what depends on the old IP. NFS or SMB exports that allow only certain IPs, firewall rules and IP sets in the Proxmox firewall, monitoring, reverse proxies and DNS records. The VMs and containers keep their own addresses, so they aren't affected unless you're also moving the whole subnet.

Then back up the two files you're about to edit:

cp /etc/network/interfaces /etc/network/interfaces.bak
cp /etc/hosts /etc/hosts.bak

Where Proxmox stores the host IP

On a default install, the host's address doesn't sit on the physical NIC. It sits on the Linux bridge vmbr0, which the VMs also use. A typical /etc/network/interfaces looks like this:

auto lo
iface lo inet loopback

iface eno1 inet manual

auto vmbr0
iface vmbr0 inet static
    address 192.168.10.5/24
    gateway 192.168.10.1
    bridge-ports eno1
    bridge-stp off
    bridge-fd 0

The second place is /etc/hosts. Proxmox expects the node's hostname to resolve to its real IP, not to 127.0.1.1:

127.0.0.1 localhost.localdomain localhost
192.168.10.5 pve1.lan pve1

This line matters more than it looks. The Proxmox cluster filesystem (pve-cluster) uses it to work out the node's address, and if it still points to the old IP, services can fail to start after the change, even on a single node.

Change the IP on a single node (CLI)

I'd use this method for anything remote, because you can arm the rollback first.

Step 1: arm an automatic rollback

The command below schedules a job that restores the old network config in 5 minutes. If the new IP works, you cancel it. If you lock yourself out, it puts the old config back and you can log in again on the old IP.

systemd-run --on-active=5m --unit=net-rollback /bin/sh -c \
  'cp /etc/network/interfaces.bak /etc/network/interfaces && cp /etc/hosts.bak /etc/hosts && ifreload -a'

It runs under systemd, not in your SSH session, so it still fires when your connection drops.

Step 2: edit the bridge address

nano /etc/network/interfaces

Change address (and gateway if the subnet changes) under iface vmbr0. Only the host's own lines. Leave bridge-ports alone.

Step 3: update /etc/hosts

nano /etc/hosts

Replace the old IP on the line with your node's hostname. If you use /etc/hosts on other machines to reach this node, update those too.

Step 4: apply without rebooting

Proxmox VE 7 and newer use ifupdown2, so you don't need the reboot the old guides tell you to do:

ifreload -a

Your SSH session on the old IP will freeze at this point. That's expected.

Step 5: check the new IP and cancel the rollback

Open a new SSH session to the new address (or open https://NEW-IP:8006), then run:

ip -br addr show vmbr0
hostname --ip-address
systemctl status pve-cluster pveproxy --no-pager

hostname --ip-address should print the new IP. Both services should be active (running). If everything looks right, cancel the rollback before the 5 minutes are up:

systemctl stop net-rollback.timer

If you don't get in within 5 minutes, wait. The timer fires, restores the old files, and the host comes back on the old IP.

Change the IP from the web UI

The web UI does the same thing, just without a safety net.

  1. Go to Datacenter → your node → System → Network, select vmbr0 and click Edit.
  2. Change IPv4/CIDR and, if needed, Gateway. Click OK. Nothing is applied yet, the changes are pending.
  3. Go to System → Hosts, replace the old IP with the new one and click Save.
  4. Back in System → Network, click Apply Configuration.
  5. Open https://NEW-IP:8006 in a new tab.

I'd only use this when I'm on the same LAN or have console access. If the new address is wrong, the page just stops responding and there's no rollback.

Change the IP of a node in a Proxmox cluster

In a cluster, the nodes talk to each other over corosync, and each node's corosync address is stored in /etc/pve/corosync.conf, which is shared by the whole cluster. Change the node's IP without updating that file and the node drops out of the cluster. That's when people end up with cluster not ready - no quorum? (500), which I covered in a separate quick fix.

There's no official one-click procedure for this. A Proxmox forum reply about PVE 9 puts it plainly: there is no specific procedure, a single server is simple and a cluster is harder but possible. Here's the order that works:

  1. Move the workload. Migrate VMs and containers off the node. If you use HA, disable or pause it so the watchdog doesn't fence the node while it's unreachable.
  2. Make sure the cluster is healthy. pvecm status should show quorum. In a two-node cluster, losing one node can mean losing quorum, so plan for that before you start.
  3. Update corosync.conf on a working node. Edit a copy, not the live file, so a half-saved edit never gets synced:
cp /etc/pve/corosync.conf /etc/pve/corosync.conf.new
nano /etc/pve/corosync.conf.new

Find the node's entry under nodelist and change ring0_addr to the new IP. Then find config_version under totem and increase it by one. If you skip the version bump, the change isn't picked up. When you're done:

mv /etc/pve/corosync.conf.new /etc/pve/corosync.conf
  1. Change the network on the node itself. Same as the single-node steps: /etc/network/interfaces, /etc/hosts, then ifreload -a.
  2. Update /etc/hosts on the other nodes if they list this node by name.
  3. Restart corosync on the changed node and check:
systemctl restart corosync
pvecm status
corosync-cfgtool -s

All nodes should be listed and the cluster should be quorate.

  1. Refresh certificates and SSH keys. This updates the node certificate and the cluster's known SSH host keys:
pvecm updatecerts --force
systemctl restart pveproxy

Then re-enable HA and migrate the guests back.

If the cluster uses a separate network for corosync (which it should, in a larger setup), and you're only changing the management IP, you may not need to touch corosync.conf at all. Check which address ring0_addr actually uses before editing.

Things that still point to the old IP

After the host is back, a few things don't update themselves:

  • The console login banner. The text shown on the physical console is generated at boot, so it can still show the old IP until the next reboot. It's cosmetic.
  • Proxmox firewall. Datacenter and node rules, IP sets and aliases that use the old address.
  • Storage. NFS exports, SMB shares and iSCSI targets on a NAS that allow only the old IP.
  • Proxmox Backup Server and other clients. Anything configured by IP rather than hostname. Using DNS names for these saves you the next time.
  • Bookmarks, reverse proxies, monitoring and DNS records.

If you clone VMs on this host, a related problem is clones that come up with the same IP as the template. That one has its own fix: cloned VM with the same IP.

My rule for Proxmox network changes

Never apply a network change on a remote Proxmox host without a way back. Either have console access or arm the rollback timer. It's one extra command, and it's the difference between a 2-minute change and a 2-hour problem.

And if you're setting up a new host, put it on a DHCP reservation or a fixed address you won't need to change, and use hostnames everywhere else. The best IP change is the one you never have to make.

FAQ

Do I need to reboot Proxmox after changing the IP?

No, not on Proxmox VE 7, 8 or 9. They use ifupdown2, so ifreload -a or Apply Configuration in the web UI applies the new address live. A reboot also works, but it isn't required, and on a remote host it's riskier because you can't easily undo it.

Why can't I reach the Proxmox web UI after changing the IP?

Usually because the old address is still somewhere. Check that vmbr0 has the new IP (ip -br addr), that /etc/hosts maps the hostname to the new IP, and that pveproxy is running. Also check the gateway if you moved to a new subnet, and clear the browser's old https://OLD-IP:8006 tab.

What files do I need to change to change the Proxmox IP?

On a single node, two: /etc/network/interfaces (the address and gateway lines of vmbr0) and /etc/hosts (the line with the node's hostname). In a cluster, also /etc/pve/corosync.conf, where you change the node's ring0_addr and increase config_version.

Can I change the IP of a Proxmox cluster node without removing it from the cluster?

Yes. Update the node's ring0_addr in /etc/pve/corosync.conf and bump config_version, change the node's network config, then restart corosync and check pvecm status. Move guests off the node and pause HA first. Removing and rejoining the node is the fallback if the cluster won't accept the change.

Will changing the Proxmox host IP change my VMs' IPs?

No. VMs and containers have their own addresses, set inside the guest or by DHCP. They keep working through the bridge. They're only affected if you also change the subnet or gateway they use.