systemd snapshot/revert: does it restore network devices after a failed upgrade?

No, systemctl revert does not restore network devices

systemctl snapshot and systemctl revert manage the state of systemd units (whether they are active, inactive, or failed), not the state of the Linux kernel or the filesystem. Because network device naming, MAC addresses, and DNS configuration files exist outside the scope of unit state, they are not restored by a revert operation.

The gap between unit state and system state

It is a common misconception that systemctl snapshot creates a system-wide restore point. In reality, it only records which services were running. The following elements are not captured or restored:

• Kernel Networking Stack: Interface names (e.g., eth0 vs enp0s3) and MAC addresses are handled by the kernel and udev.
• Filesystem Configuration: Changes to /etc/resolv.conf, /etc/network/interfaces, or Netplan YAML files are permanent unless you are using a filesystem-level snapshot (like Btrfs or LVM).
• Hardware State: Any driver changes or firmware updates applied during the upgrade remain in effect.

Reconciling network devices after a revert

If a failed upgrade changed your interface names or broke connectivity, follow these steps to manually reconcile the state:

1. Identify current naming: Run ip link show to see the current device names assigned by the kernel.
2. Restore udev rules: If names changed, check /etc/udev/rules.d/ for any modified rules that map MAC addresses to specific names.
3. Manual renaming: If you need to temporarily rename an interface to match old configuration files, use:

   ip link set dev [new_name] name [old_name]

4. Restart network units: Once the names are corrected, trigger the network units that systemctl revert attempted to restore:

   systemctl restart systemd-networkd # or NetworkManager

Verifying DNS and Connectivity

Since /etc/resolv.conf is often a symlink managed by systemd-resolved or NetworkManager, a unit revert will not roll back the content of the file. To verify the state:

• Compare configuration: Use cat /etc/resolv.conf to check if the nameservers match your pre-upgrade requirements.
• Test resolution: Run resolvectl status (if using systemd-resolved) to see which DNS servers are actually being queried.
• End-to-end check: Use ping -c 3 google.com to confirm the full stack (Interface → Gateway → DNS) is functional.

Diagnostic Requirement: To provide a more specific recovery path, please specify if you are using systemd-networkd, NetworkManager, or static ifupdown configurations.