Taking a snapshot before a risky change

This picks up from a server you can already reach over SSH.

The fastest way to take a restore point on a Hostworld VPS is to use Virtualizor’s Enduser Backup from your client area. It creates a live VM‑level backup. You will usually keep the VPS online while it runs. Expect a few minutes, then an email when it is finished. If you prefer an on‑guest snapshot you can use LVM, Btrfs or ZFS. For databases and other busy apps, quiesce writes first so the snapshot is application‑consistent.

Before you start

• This guide focuses on Hostworld VPSs managed in Virtualizor and on‑guest snapshots in AlmaLinux 9 and Ubuntu 24.04.
• Virtualizor Enduser Backup: manual backups and restores are available to you if your VPS has a Backup Plan that allows them. If you cannot see the backup option, or restore is not enabled on your plan, open a support ticket and we will check it.
• VM‑level backups are live and, by default, crash‑consistent. This is fine for the operating system and most files. Databases and mail queues may be inconsistent after a restore if you do not quiesce them. If app‑consistent safety matters, use QEMU Guest Agent freeze/thaw hooks or stop/flush the services before taking the backup.
• Virtualizor queues backups per node. During busy periods your job may wait for its turn. That affects the overall time from clicking “take backup” to getting the completion email.
• If you snapshot inside the VPS:
  • LVM snapshots must be sized correctly. If a snapshot fills to 100% it becomes invalid and you lose the ability to roll back.
  • Rolling back an LVM snapshot of the root filesystem merges on the next activation. Plan a reboot window.
  • Btrfs and ZFS snapshots live on the same storage. They are not off‑VPS backups. Keep external backups as well.
  • If /boot is on a non‑Btrfs or non‑ZFS filesystem, kernel and bootloader files are not in those snapshots. Handle /boot separately when you roll back.
• Do not reboot or power off while a backup is running. Interrupting a job risks a failed or corrupt backup. Wait for the completion email and the backup to appear in the list.

Step 1: Choose your restore point method

Pick one method based on how much you are changing and how quickly you want to revert:

• Virtualizor Enduser Backup. This is our panel route and works for whole‑VPS rollback. It keeps the VPS running while it takes a crash‑consistent snapshot of disk state. If you need app consistency, quiesce the guest first.
• LVM, Btrfs or ZFS inside the VPS. This is useful if you want a quick local revert for system packages or configs, or if you need to roll back only the root filesystem without touching the whole VM. It does not protect you from node or disk failure.

If in doubt, take a Virtualizor backup first. You can add an on‑guest snapshot as an extra layer if you want belt and braces.

Step 2: Quiesce busy apps if you need application‑consistent safety

Before any VM‑level or filesystem snapshot, decide if you need application consistency. Crash‑consistent snapshots capture whatever was on disk at that instant. Databases, caches and mail queues can look like an abrupt power loss if writes were mid‑flight. To avoid that:

• Use QEMU Guest Agent freeze/thaw. QEMU Guest Agent supports guest-fsfreeze-freeze and guest-fsfreeze-thaw, plus hook scripts for pre/post freeze, to pause filesystem writes during the snapshot then resume them. Virtualizor exposes guest‑agent controls, and the agent can be enabled per VPS.
• Or briefly stop or flush your database and mail services before you trigger the backup.

Note that freezing writes does pause applications for a few seconds. Test this during a quiet period and ensure the thaw always runs. If anything is unclear, open a support ticket and we will advise on the least disruptive approach for your VPS.

Step 3: Take a Virtualizor backup of your Hostworld VPS

1. Reach Virtualizor from your Hostworld client area.

   This opens the VPS management view where you can manage power, console and backups for your VPS.

2. Open the Enduser Backup page for your VPS and start a manual backup.

   This lists existing backups and provides a “take backup now” action. Starting it queues a live backup on the node. You will receive an email when it completes. During busy periods your job may wait for concurrency on the node.

3. Wait for completion.

   Do not reboot or power off the VPS during this time. The backup process sends an email on completion and then the new backup shows in the list. Keep the email so you can match it to the backup entry.

4. If you do not see the backup option or need restore permission enabled, open a support ticket.

   Enduser backup and restore are controlled by Backup Plans. We can check your VPS’s plan and enable the right actions for you where your service allows it.

Step 4: Know how you will roll back in Virtualizor

• Restore from the Enduser Backup list if your plan allows it. Restoring reverts the VPS disk to that point in time. Treat it as a destructive change that discards later writes. Plan a maintenance window and notify any users of the VPS before you proceed.
• After a restore, incremental backup chains may restart. That is expected behaviour for incremental systems.
• If you want us to confirm the quickest rollback path for your VPS, open a support ticket.

Step 5: Take an LVM snapshot inside the VPS

Use this only if your VPS uses LVM. You need enough free space in the volume group to hold a snapshot. If the snapshot LV fills to 100% it becomes invalid and cannot be used to roll back.

Create an LVM snapshot

This command creates a snapshot logical volume named prechange_root_snap for the existing logical volume VG/LV (replace with your volume group and logical volume names) with a size allocation for changed blocks during your maintenance window.

AlmaLinux 9

lvcreate --snapshot --size 2G --name prechange_root_snap VG/LV

Ubuntu 24.04

lvcreate --snapshot --size 2G --name prechange_root_snap VG/LV

Pick a size that can absorb the writes you expect while the snapshot exists. The more you change, the more space the snapshot uses.

Roll back an LVM snapshot

This command merges the snapshot back into the original logical volume, reverting it to the snapshot point.

AlmaLinux 9

lvconvert --merge VG/prechange_root_snap

Ubuntu 24.04

lvconvert --merge VG/prechange_root_snap

If you are rolling back the root filesystem and it is active, the merge completes at the next activation. In practice that means after a reboot. Reboot the VPS from Virtualizor to complete the rollback during your maintenance window. Do not attempt to keep working during the merge.

Step 6: Take a Btrfs snapshot inside the VPS

Use this only if your VPS uses Btrfs subvolumes for the data you want to protect. Remember that Btrfs snapshots are not backups and live on the same storage. If /boot is not on Btrfs, kernel and bootloader files are outside of the snapshot and must be handled separately during a rollback.

Create a read‑only Btrfs subvolume snapshot

This command creates a read‑only snapshot of the source subvolume at the destination path. Replace the paths to match your layout, for example snapshotting the root subvolume to a sibling directory.

AlmaLinux 9

btrfs subvolume snapshot -r <src_subvol_path> <dest_path>

Ubuntu 24.04

btrfs subvolume snapshot -r <src_subvol_path> <dest_path>

Roll back a Btrfs snapshot

• Either set the snapshot as the default subvolume and reboot into it.
• Or replace the root subvolume with the snapshot.

Both approaches are valid. Choose the one that fits your layout and comfort level. Make sure you plan for how you will handle /boot if it is on a separate non‑Btrfs filesystem.

Step 7: Take a ZFS snapshot inside the VPS

Use this if your VPS uses ZFS datasets. On Ubuntu 24.04, zsys is not installed by default, so use standard ZFS commands.

Create a ZFS snapshot

This command creates a point‑in‑time snapshot named prechange for the given ZFS dataset. Use -r to include child datasets if needed.

AlmaLinux 9

zfs snapshot pool/dataset@prechange

Ubuntu 24.04

zfs snapshot pool/dataset@prechange

Roll back a ZFS snapshot

This command reverts the dataset to the snapshot. The -R flag rolls back child datasets too.

AlmaLinux 9

zfs rollback -R pool/dataset@prechange

Ubuntu 24.04

zfs rollback -R pool/dataset@prechange

Step 8: How long it takes and whether the server stays up

• Virtualizor Enduser Backup: Virtualizor confirms “Please allow a few minutes” and sends you an email when the backup is complete. The actual time depends on the amount of data and how busy the node is. Backup concurrency is limited per node, so your job may queue. The VPS stays running during a VM‑level backup. If you enabled guest‑agent freeze/thaw for application consistency, writes pause briefly during the snapshot.
• On‑guest snapshots (LVM/Btrfs/ZFS): creation is quick and does not require a reboot. Rolling back the root filesystem with LVM completes after the next reboot. Btrfs and ZFS rollbacks are also quick but plan a reboot where appropriate for a clean switch.
• Do not interrupt a running backup. Reboots and power‑offs during a backup risk a failed or corrupt result. Wait for the completion email and see the backup listed before you proceed with your change.

Step 9: Why this five‑minute habit prevents most emergencies

Most outages after a risky change come from two things: unexpected package conflicts and configuration mistakes. A pre‑change restore point lets you put the system back exactly as it was, then try again in a controlled way. Even a crash‑consistent VM‑level backup is enough to recover the OS state. If you quiesce databases or stop services first, you protect active data as well. The time to take a restore point is almost always less than the time to rebuild, re‑configure and explain an outage.

What next

• If you are planning a larger upgrade and want a second layer of safety, take both a Virtualizor backup and an on‑guest snapshot before you start.
• If you need help enabling backups or deciding the best rollback path for your VPS, open a support ticket.
• Explore more topics in our VPS guides, or if you are comparing plans, see our UK location on the Linux VPS page.