Ubuntu Troubleshooting Guide: Find the Fix by Symptom

lsb_release -ds; uname -r
systemctl --failed
journalctl -b -p err --no-pager | tail -n 40
df -h /
sudo apt update

Find your problem

Boot and startup

Display and graphics

Hardware

Packages and updates

System, disk and permissions

Ubuntu-specific pages assume the default GNOME desktop; the same fixes apply to Kubuntu, Xubuntu and Linux Mint unless a page says otherwise, because they share Ubuntu's kernel, drivers and package archive. Not listed? Browse all error messages or the distro-neutral Linux troubleshooting guide.

Step 1: Know your version and what changed

lsb_release -a
uname -r
grep -E 'install|upgrade' /var/log/dpkg.log | tail -n 20

The last command shows the most recent package changes, which is usually where a "it worked yesterday" problem started. Older logs are in /var/log/dpkg.log.1 and compressed .gz files (zgrep reads them).

Support status matters because unsupported releases get no fixes:

Ubuntu 26.10 is due in October 2026 as a short-term release. Release lifecycles are explained in Linux release cycles.

Step 2: Read the logs

# errors in this boot
journalctl -b -p err --no-pager
# errors in the previous boot (after a crash or a black screen)
journalctl -b -1 -p err --no-pager
# kernel and hardware messages
sudo dmesg --level=err,warn | tail -n 40
# one service
journalctl -u NetworkManager -b --no-pager | tail -n 50

Not every red line is a problem; many are harmless warnings that appear on every boot. Look for messages with the time your problem happened, and for words like failed, timeout, segfault and firmware. The journalctl cheat sheet lists filters by time, unit and priority.

If the journal from earlier boots is empty, persistent logging is off. Enable it so the next crash is recorded: sudo mkdir -p /var/log/journal && sudo systemctl restart systemd-journald.

Step 3: Check services, disk and memory

systemctl --failed
df -h
df -i /
free -h

Example output of systemctl --failed:

  UNIT                 LOAD   ACTIVE SUB    DESCRIPTION
● apache2.service      loaded failed failed The Apache HTTP Server

1 loaded units listed.

Investigate failed units with systemctl status UNIT; systemd basics explains the output. A root filesystem at 100% in df -h (or inodes at 100% in df -i) breaks logins, updates and many apps; it's a common hidden cause of odd behaviour. Swap fully used in free -h points to a memory problem.

Step 4: Repair packages

A failed update leaves packages half-configured, which can break the desktop, drivers or apt itself. This sequence is safe to run at any time:

sudo dpkg --configure -a
sudo apt --fix-broken install
sudo apt update
sudo apt full-upgrade

If one of these prints an error, look it up in the packages table above; the message tells you which fix applies. For understanding what apt is doing, see the apt package manager guide.

When you can't reach a desktop: recovery mode and live USB

Text console

If the screen is black or frozen but the system is running, press Ctrl+Alt+F3 for a text login. Everything in this guide works from there.

Recovery mode

Hold Shift (BIOS) or tap Esc (UEFI) right after the firmware logo to show GRUB. Choose Advanced options for Ubuntu → the newest entry ending in (recovery mode). The recovery menu offers:

• fsck: check filesystems and remount root read-write.
• dpkg: repair broken packages (needs network).
• network: enable networking.
• root: a root shell. If the filesystem is read-only, run mount -o remount,rw /.

Also try the previous kernel in Advanced options: if it works, the problem came with a kernel or driver update. To boot with different options, such as nomodeset or systemd.unit=multi-user.target, see kernel parameters.

Live USB

Boot the Ubuntu installer USB and choose Try Ubuntu. From there you can copy your files to an external drive, check the disk (sudo fsck -f /dev/sdXN on an unmounted partition), and repair the bootloader with chroot as shown in fix grub rescue.

Roll back with a snapshot

If you set up Timeshift (or use ZFS or Btrfs snapshots), restoring the last snapshot from before the problem is often faster than debugging. Timeshift can restore from a live USB too: install it in the live session with sudo apt install timeshift, select your system disk and pick a snapshot. Snapshots don't replace backups of your personal files.

Things that commonly cause trouble

• PPAs and third-party repositories: they can replace core packages. List them with ls /etc/apt/sources.list.d/; ppa-purge reverts a PPA's packages to Ubuntu's versions. Ubuntu's own sources live in /etc/apt/sources.list.d/ubuntu.sources (deb822 format) since 23.04.
• Old tutorials using apt-key: it is removed in Ubuntu 26.04. Add repository keys to /etc/apt/keyrings/ and reference them with signed-by instead.
• Snap vs deb confusion: Firefox, Thunderbird and others are snaps by default. snap list and snap changes show snap updates, which apt doesn't see.
• sudo-rs on 26.04: Ubuntu 26.04 uses the Rust rewrite of sudo by default. Day-to-day usage is the same; rare edge-case options in old sudoers files may behave differently.
• Interrupted upgrades: if a release upgrade stopped midway, run the package repair sequence above, then sudo apt full-upgrade, before rebooting.

Asking for help effectively

When you post on Ask Ubuntu, Ubuntu Discourse or a bug tracker, include the Ubuntu version and kernel (lsb_release -ds; uname -r), the exact error text (copy, don't paraphrase), what you changed before it broke, and the relevant log lines. For hardware problems add lspci -nnk or lsusb. ubuntu-bug PACKAGE files a Launchpad bug with logs attached automatically. Ubuntu's official help pages are at help.ubuntu.com, and the distro overview is on our Ubuntu page.

Frequently asked questions