A calmer update routine.
An update is easier when the machine already has a way back. This routine is short enough to repeat and specific enough to catch the problems that appear after a reboot.
- Maintenance
- Routine
- 5 min read
The expensive part of a failed update is not the download. It is the hour you spend discovering that the panel will not start, the network is gone, or the login screen loops. A small routine moves that discovery into a session you can afford to lose. It also gives you a written record for the next update.
Keep a spare session
A second user account is the simplest safety net. It has its own settings, so a broken panel layout in one account does not block the other. A live session from installation media is the next step up. Either way, confirm that the spare session can log in before you start an update. A safety net you have never tested is only a hope.
Read the notes before the update
Read the release note for the kernel, the desktop, and any toolkit that your theme depends on. Look for the sections on known issues, removed options, and configuration changes. If a theme relies on a widget that changed, plan to test the theme after the update. Do not change the theme on the same day as the update; one change at a time keeps the cause clear.
Make a short test list
Write the list before you reboot. A useful list covers the things you actually use, not every feature in the menu.
- Boot and login, including the spare account.
- Network, sound, and display resolution.
- Panel, dock, and the applications pinned there.
- File manager, terminal, and the browser.
- The theme, including a dialog and a menu.
- Suspend and wake, if you use them.
Update in stages
Apply the kernel, the desktop, and the applications as separate steps when your package tool allows it. Reboot and run the short list after each step. A problem is much easier to name when only one layer changed. If a step fails, stop there. Do not stack a second change on top of a broken one.
Know the way back
Find out how your bootloader selects an older kernel, and confirm that an older entry is present. If you use snapshots, take one before the update and note its name. If you keep the package cache, confirm that the older packages are still there. A rollback plan is only useful if you have read it before you need it.
Keep a short log with the date, the packages that changed, and the result of the test list. The log turns a vague memory into a fact the next time a problem appears. It also makes a better forum post when you need help.
A short checklist
- Confirm the spare session works.
- Read the release notes for known issues.
- Write the test list before the reboot.
- Update one layer at a time.
- Confirm the older kernel or snapshot is available.
- Record the result in a short log.
Further reading
Kernel.org publishes the kernel release notes. The Arch Wiki has detailed pages on bootloaders and snapshots that are useful even on other distributions. Debian and Fedora publish their own release and upgrade notes, and man pages cover the tools you use to inspect the result.