Unstick an 'omarchy update -y' that hangs after the orphan list or the kernel update
Read on 4.0.4-1: omarchy-update re-execs under script -qefc ... /tmp/omarchy-update.log, exports OMARCHY_UPDATE_UNATTENDED=1 for -y, and runs orphan-pkgs before update-restart. omarchy-update-orphan-pkgs guards only on [[ ! -t 0 || ! -t 1 ]] and uses gum confirm --default=false "Remove N orphaned package(s)?", handling a non-zero answer with Keeping orphaned packages.. omarchy-update-restart has a bare confirm_reboot and three prompts with exactly the strings the pkill patterns use, and it has no set -e, so a killed gum falls through. quattro's confirm_reboot prints Run omarchy-system-reboot when ready. under OMARCHY_UPDATE_UNATTENDED, and quattro's orphan step still has the tty-only guard. Issues #13356, #8986 and #13416 support all of this. One inaccuracy: the fix says to expect one prompt on a run that upgrades the kernel or Hyprland, but a 'no' to the kernel prompt falls through to the Hyprland check, so a run upgrading both asks twice (#13416 shows both lines). That sentence corrected. No update was run.
Removing orphans with pacman -Rns can remove build dependencies or tools you installed as dependencies and still use. Read the list before answering yes.
Symptom
An unattended omarchy update -y (from a script, an agent, ssh or a supervisor) stops producing output and never exits. The last lines are an orphan list:
Orphan system packages
asar
vulkan-headers
<nothing more>
or it got further and stops after a kernel or Hyprland upgrade. /tmp/omarchy-update.log ends in a box drawing asking Remove N orphaned package(s)? or Linux kernel has been updated. Reboot?. Keystrokes sent to the process are echoed back rather than answering.
Cause
-y sets OMARCHY_UPDATE_UNATTENDED=1, documented in omarchy-update as a promise not to ask anything, but two steps on 4.0.4-1 ignore it. omarchy-update-orphan-pkgs decides interactivity only from [[ ! -t 0 || ! -t 1 ]], and omarchy-update re-executes itself under script -qefc ... /tmp/omarchy-update.log, which supplies a pty, so the guard passes and gum confirm waits forever. omarchy-update-restart's confirm_reboot() on 4.0.4-1 is a bare gum confirm "$1" with no unattended check at all. Both confirmed by reading the scripts on this workstation. By the time either prompt appears the package transaction has already completed, so the system is fine and only the tail of the update (log analysis, status, shell restart) is waiting.
Fix
From another shell, find the waiting gum and kill it. Killing it answers no and the update carries on to the next step:
pgrep -af 'gum confirm'
pkill -f 'gum confirm --default=false Remove' # the orphan prompt
pkill -f 'gum confirm Linux kernel has been updated' # reboot prompt after a kernel upgrade
pkill -f 'gum confirm Hyprland has been updated' # reboot prompt after a Hyprland upgrade
pkill -f 'gum confirm Updates require reboot' # reboot prompt from the reboot-required marker
To avoid the orphan prompt on unattended runs, review and clear orphans beforehand in an interactive terminal:
pacman -Qtdq
omarchy-update-orphan-pkgs # asks, then runs sudo pacman -Rns on the list
There is no flag that skips the reboot prompts on 4.0.4-1. On quattro, confirm_reboot already prints Run omarchy-system-reboot when ready. instead of asking when -y is used, so that half goes away with a later release. Until then, expect to kill one reboot prompt for each condition that applies. A run that upgrades both the kernel and Hyprland asks twice, because answering no to the kernel prompt falls through to the Hyprland check. Reboot when convenient afterwards.
Verify
The update prints its final status and Restarting shell, the process exits, and pgrep -af 'omarchy-update|gum confirm' returns nothing.