Skip to main content
RunBook Academy

← All break/fix scenarios in Proxmox VE

beginnerNetworking~15 min

Network between PVE and PBS becomes unreachable

Reported symptoms

  • ●vzdump jobs targeting pbs-main fail with connection refused
  • ●pvesm status reports pbs-main as inactive
  • ●From the PVE host, ping pbs.example.com returns destination unreachable
  • ●Backup jobs from other clusters to the same PBS work normally

Evidence

  • · ip route on the PVE host shows a missing default route
  • · /etc/network/interfaces on the PVE host has no `gateway` directive
  • · ip addr shows the management interface has only a link-local address
  • · Recent file edits to /etc/network/interfaces are timestamped today
Diagnosis and resolutionclick to reveal

Root cause

A recent edit to /etc/network/interfaces removed the gateway line as part of an unrelated change. Without a default route, the host can reach other hosts on its own subnet but cannot reach PBS on a different subnet. This is the typical outcome when someone edits network config without testing cross-subnet reachability afterwards.

Remediation

1. Add the gateway line back to /etc/network/interfaces: ``` auto vmbr0 iface vmbr0 inet static address 10.0.0.11/24 gateway 10.0.0.1 ``` 2. Reload the network: `ifreload -a` (or `systemctl restart networking` if ifreload is not installed). 3. Verify: `ip route show default` should now show the gateway. 4. Test cross-subnet: `ping 10.10.0.5` (the PBS host on the backup subnet) should succeed. 5. Re-test vzdump to PBS.

Verification

- `ping <pbs-ip>` succeeds - `pvesm status` shows pbs-main active - A new vzdump job to pbs-main completes - Other scheduled backups continue to work

Prevention

- Always test cross-subnet reachability after any network config change - Use `ifup --no-act` to dry-run an interfaces change before applying - Keep `/etc/network/interfaces` in version control - Add "verify inter-subnet connectivity" to the network change checklist

PBS unreachable after network change

Why this happens

Single-line typos in /etc/network/interfaces are the second most common cause of “everything was working, now it’s not” reports. The first is a colleague restarting the wrong service.

The diagnostic trick is to test reachability at every layer:

  1. Is the local interface up? ip link show
  2. Does the local subnet work? ping <same-subnet host>
  3. Does the default route exist? ip route show default
  4. Does cross-subnet work? ping <other-subnet host>
  5. Does DNS work? dig pbs.example.com
  6. Does the application port respond? nc -vz pbs.example.com 8007

Most network bugs fail at exactly one of these steps.