Skip to content
Hyperion

Update a node

The Nodes page can run the operating system upgrade and the Hyperion update on any worker node, and the system upgrade on the master. The run is a background job on the node, and the page shows its progress. Hyperion’s own update on the master is covered in the install guides.

Before you start

Section titled “Before you start”
  • Sign in as an Owner.
  • Open the node and check Operating system updates first. See Check the health of your nodes. A differs from the master's flag on the agent version means the worker needs the Hyperion update.
  • Services the upgrade touches, such as nginx, PHP and MariaDB, restart as part of it. Pick a quiet time.
  • Confirm the sites on the node have a recent backup. Open one site, open its Backups tab and check the newest entry is recent and finished. The safe mode restores programs, never site data.

Steps

Section titled “Steps”

Update a worker

Section titled “Update a worker”
  1. Open Nodes in the sidebar, then click the worker’s row.
  2. In the Update section, all three options are ticked by default. Untick what you do not want to run.
    • System packages runs apt upgrade. It never removes a package.
    • Hyperion rebuilds and restarts the agent and the web service.
    • Safe update takes a snapshot of the running binaries first and rolls back if the node is not healthy afterwards. It is ticked by default.
  3. Click Start update and confirm Start update in the dialog. The page shows “update started” and returns to the row.
  4. Watch the status line under the form. It refreshes every 3 seconds and shows a pill, the scope (apt + hyperion, apt only or hyperion only) and a log of the run.

If you untick both System packages and Hyperion, the page answers “nothing to update (tick at least one option)”.

Install system updates on the master

Section titled “Install system updates on the master”
  1. Open Nodes and click the master’s row, marked this server.
  2. In Install system updates, click Install system updates and confirm Install updates in the dialog.

This runs apt upgrade on the master and never removes a package. Anything that would need a removal is held back and listed in the log. It does not update Hyperion; the Hyperion update runs on its own schedule under Settings, Panel, Updates. Services restart as part of the upgrade, so sites and the panel can be unavailable briefly. A kernel or library update asks for a reboot afterwards, and the reboot is left to you.

From a shell

Section titled “From a shell”

On the node itself, the same job runs with:

Terminal window
sudo hctl node update --apt --hyperion --safe --follow

--follow streams the log and waits through the agent’s restart. sudo hctl node update-status prints the last run.

Check that it worked

Section titled “Check that it worked”
  • The status pill reads done. running means it is still going, failed shows the exit code, and interrupted means the run stopped before it finished, for example because of a reboot.
  • After a Hyperion update, the row’s agent version matches the master’s and the differs from the master's problem is gone.
  • Operating system updates shows no updates pending after Check now. If it shows reboot required, reboot the node when it suits you.
  • Test connection on the row shows reachable.

If it fails

Section titled “If it fails”
  • failed with an exit code: read the log under the status pill. With Safe update on, the node is back on its previous binaries and the run is reported as failed.
  • interrupted: “stopped before it could finish (reboot?) — run it again: it repairs any half-installed packages first”. Start the update again.
  • “update failed to start: …”: the node refused or did not answer. Test the connection first.
  • The row says the node runs an older Hyperion and shows no OS updates: run Update with Hyperion ticked, which brings the agent up to date.
Section titled “Related”