OPINIONATED OMARCHY
TROUBLESHOOTING CORPUS
790
RECORDS
790
AUDITED
0
UNCHECKED
2388
SOURCES
LIVE
omarchy-core high occasional DISPUTE

Release the root-owned sleep inhibitor an interrupted omarchy update left behind

omarchyomarchy-4laptopdesktopsystemd
CORRECTED An auditor found the fix wrong or incomplete and rewrote it. · checked against Omarchy omarchy 4.0.4-1 on 2026-10-05
AUDIT NOTE

Read /usr/bin/omarchy-update-stay-awake on 4.0.4-1 here: start() runs sudo (tty) or pkexec (no tty) systemd-inhibit --what=sleep:idle --who=omarchy-update ... --mode=block sleep infinity &, stores $! plus start time, and stop() does kill ... || true then prints Failed to stop the Omarchy update sleep inhibitor. and returns 1 when the process survives. start() calls stop() first. Issue #14205 (open, 2026-10-04) reports exactly this on stable 4.0.4-1 with the same inhibitor table and process chain, and states quattro fixed it via #9467 and #13361, which supports the cause. Upstream latest release is still v4.0.4, so 'newer than 4.0.4' holds. Two defects in the fix: sudo pkill -f -- 'who=omarchy-update' also matches the command line of the sudo process running pkill (pkill only excludes itself), so the pattern needs the [w] bracket form, and the 'Remove it anyway' step only lists processes with pgrep -af '^sleep infinity$', which also matches unrelated sleep infinity processes and removes nothing. Fix rewritten for both. The EPERM itself and the kill were not exercised, by brief.

The cause above was not rewritten and may still contain the error described. The fix below is the corrected version.
RISK

Do not kill the inhibitor while an update is genuinely running. Suspending mid-transaction is exactly what it exists to prevent, and a suspend or power loss during the pacman step can leave packages half-installed.

Symptom

Suspend from the system menu silently does nothing, and systemctl suspend is refused, hours after an omarchy update that was interrupted or killed. Nothing is updating. systemd-inhibit --list still shows:


WHO             WHAT        WHY                         MODE
omarchy-update  sleep:idle  Omarchy update in progress  block

Running omarchy-update-stay-awake stop prints Failed to stop the Omarchy update sleep inhibitor. It first looks like a laptop or NVIDIA suspend bug.

Cause

On Omarchy 4.0.4-1 stable, /usr/bin/omarchy-update-stay-awake start launches sudo systemd-inhibit --what=sleep:idle --who=omarchy-update ... --mode=block sleep infinity & (or pkexec when stdin is not a terminal) and records $! in $XDG_RUNTIME_DIR/omarchy-update-stay-awake/inhibit-pid. That PID is the root-owned sudo process. stop runs as the desktop user and does kill "$inhibit_pid" >/dev/null 2>&1 || true, which fails with EPERM and is swallowed. When the update is interrupted before its own cleanup, the privileged inhibitor survives with no owner, and because start begins by calling stop, later updates cannot clear it either. Confirmed by reading the 4.0.4-1 script on this workstation. Upstream quattro already replaced this with a holder that drops to the invoking user (PR #9467, refined in #13361), but that had not reached stable 4.0.4.

Fix

Make sure nothing is actually updating first:


pgrep -a omarchy-update; pgrep -ax pacman; pgrep -ax yay
systemd-inhibit --list --no-pager | grep omarchy-update
pgrep -af 'who=omarchy-update'

If the only hits are the sudo systemd-inhibit ... sleep infinity chain, kill it with privilege, then let the helper clear its state file and the stay-awake idle toggle it set:


sudo pkill -f -- '[w]ho=omarchy-update'
omarchy-update-stay-awake stop

The [w] keeps the pattern from matching the sudo pkill command line itself, which contains the same text.

A leftover sleep infinity reparented to init is harmless once systemd-inhibit is gone, because the inhibitor lock belonged to that process. Other programs also run sleep infinity, so look before removing anything:


ps -o pid,ppid,user,args -C sleep

Only a root-owned sleep infinity whose PPID is 1 is the leftover. Remove it with sudo kill on that PID, or leave it.

A reboot also clears it. The permanent fix is the upstream rewrite of omarchy-update-stay-awake (PR #9467, refined in #13361), which arrives with an Omarchy release newer than 4.0.4. Check with pacman -Q omarchy.

Verify

systemd-inhibit --list --no-pager | grep omarchy-update prints nothing, ls $XDG_RUNTIME_DIR/omarchy-update-stay-awake/ reports no such directory or an empty one, and Suspend from the system menu now suspends the machine.

Sources

update-stale-root-inhibitor-blocks-suspend