OPINIONATED OMARCHY
TROUBLESHOOTING CORPUS
790
RECORDS
790
AUDITED
0
UNCHECKED
2388
SOURCES
LIVE
power-suspend medium very-common DISPUTE

Diagnose 'systemctl suspend does nothing' when an inhibitor is holding it

omarchyarchendeavouroscachyosmanjarohyprlandlaptopdesktop
CORRECTED An auditor found the fix wrong or incomplete and rewrote it.
AUDIT NOTE

Every systemd claim held and every Omarchy claim was Omarchy 3. Checked on this workstation (omarchy 4.0.2-1, omarchy-settings 4.0.2-1, systemd 261.2-1, Hyprland 0.56.2-1): systemd-inhibit --list and its --what/--mode filters print the columns the record describes, -i is --check-inhibitors=no, InhibitDelayMaxSec= defaults to 5, LidSwitchIgnoreInhibited= defaults to yes and the three key options to no, all read from the installed logind.conf(5) and systemctl(1) and systemd-inhibit(1) man pages. Four defects. First, hypridle is not installed (pacman -Q hypridle fails), there is no ~/.config/hypr/hypridle.conf and none in /usr/share/omarchy/config/hypr/. A gh api query on omacom/omarchy tree quattro returns zero hypridle paths, while the two cited basecamp/omarchy/master raw URLs still serve the Omarchy 3 hypridle.conf and the old pkill-based toggle, which is where the error came from. Omarchy 4 does idle in /usr/share/omarchy/shell/plugins/services/idle/Service.qml and locks before sleep via the user unit omarchy-sleep-lock.service running omarchy-system-sleep-monitor. Second, omarchy-toggle-idle on 4.0.2-1 takes toggle|stay-awake|allow-idle|status and only touches or removes ~/.local/state/omarchy/indicators/stay-awake, so the record's double invocation is a no-op round trip and the stay-awake state, which disables idle with no inhibitor to see, was never mentioned. This machine is currently in that state (omarchy-shell idle status returns "stayAwake":true,"enabled":false). Third, step 3's /etc/systemd/logind.conf.d/10-inhibit.conf is dead on Omarchy, because omarchy-settings owns 20-inhibit-delay.conf with InhibitDelayMaxSec=15, drop-ins sort lexicographically with the last writer winning, and the live busctl get-property ... InhibitDelayMaxUSec reads t 15000000. Fourth, step 6's logind idle branch is inert here because IdleAction reads s "ignore" and Omarchy's idle monitor sets respectInhibitors: true on a Wayland idle-inhibit monitor, so an app blocking the lock shows as .inhibitingIdle == true in hyprctl clients -j (field confirmed present on 0.56.2) and never in systemd-inhibit --list --what=idle, which was empty here. Rewrote the fix with plain-Arch and Omarchy 4 branches labelled, added omarchy-debug-idle, the stay-awake trap and the omarchy-hyprland-session-locked stale-lock check, and switched logind restart to reload (CanReload=yes). Not exercised: nothing was suspended, no drop-in was written and logind was not reloaded, so the new InhibitDelayMaxSec override and the lid setting are read from the man page and the shipped drop-in rather than from a live change. The record was not GNOME or KDE mis-specialised.

The cause above was rewritten on 2026-09-12 to match this note; the fix was corrected by the audit itself.
RISK

systemctl suspend -i overrides the lock rather than resolving it. If the inhibitor was a package upgrade, a disk write, a dd/rsync or a VM snapshot, suspending through it can corrupt that work. Read the Why string before you override. On Omarchy 4 it also skips nothing of the lock path, because that inhibitor is delay rather than block, but killing the Omarchy sleep inhibitor by PID does defeat it and leaves the machine suspending with the session unlocked. Use sudo systemctl reload systemd-logind rather than restart after editing logind drop-ins, because a restart can tear down running graphical sessions. Leaving omarchy-toggle-idle stay-awake set means the machine never locks or blanks at all, which is a security regression that no inhibitor listing will reveal. Omarchy 4's lock screen has no unlock() IPC and survives its own client dying, so do not experiment with locking on a machine you can only reach over ssh.

Symptom

systemctl suspend returns instantly and the machine just sits there. Closing the lid does nothing either. Occasionally I get a message naming the culprit:


Operation inhibited by "Steam" (PID 4711 "steam", user me), reason is "Downloading".
Please retry operation after closing inhibitors and logging out other users.

Adding -i makes it suspend fine, so something is blocking it. I just can't see what.

Cause

logind honours inhibitor locks. A block inhibitor prohibits sleep indefinitely (Steam while downloading, a browser playing video, mpv, a package manager, an SSH session holding systemd-inhibit). A delay inhibitor only postpones sleep up to InhibitDelayMaxSec, which is 5 seconds on stock systemd and 15 seconds on Omarchy 4 because omarchy-settings ships a drop-in raising it. Delay inhibitors are normal: NetworkManager, UPower and, on Omarchy 4, the user unit omarchy-sleep-lock.service all hold one, the last so omarchy-system-sleep-lock can lock the session before the machine sleeps. Separately, an idle-only inhibitor stops logind's IdleAction from firing without blocking a manual systemctl suspend, which is why the two symptoms look different. On Omarchy 4 that second half is the part most advice gets wrong: IdleAction is set to ignore, idle is detected by a Quickshell plugin using the Wayland idle-inhibit protocol rather than by logind, and the whole idle path can be switched off by a state file that no inhibitor listing will ever show.

Fix

**1. See who is holding a lock, the first-line diagnostic**


systemd-inhibit --list
systemd-inhibit --list --mode=block          # only the hard blockers
systemd-inhibit --list --what=sleep
systemd-inhibit --list --what=idle

The columns are WHO UID USER PID COMM WHAT WHY MODE. Quit that app, or kill the PID.

On a healthy Omarchy 4 machine the sleep list already has three delay inhibitors and none of them is a fault:


WHO            UID  USER        PID  COMM            WHAT  WHY                                       MODE
NetworkManager 0    root        1470 NetworkManager  sleep NetworkManager needs to turn off networks delay
UPower         0    root        2330 upowerd         sleep Pause device polling                      delay
Omarchy        1000 you         2531 systemd-inhibit sleep Lock screen before suspend                delay

The third is the user unit omarchy-sleep-lock.service running /usr/bin/omarchy-system-sleep-monitor. Do not kill it. It is what locks the screen before the machine sleeps, and without it the machine suspends with the session exposed.

**2. Override once, deliberately**


systemctl suspend -i        # -i is shorthand for --check-inhibitors=no

**3. Give delay-mode inhibitors more room**

Plain Arch: InhibitDelayMaxSec= defaults to 5 seconds.

Omarchy 4: omarchy-settings already ships /etc/systemd/logind.conf.d/20-inhibit-delay.conf setting InhibitDelayMaxSec=15, because Quickshell needs longer than 5 seconds to secure the session when closing the lid also reconfigures displays. Read the effective value rather than assuming a default:


busctl get-property org.freedesktop.login1 /org/freedesktop/login1 \
  org.freedesktop.login1.Manager InhibitDelayMaxUSec     # t 15000000 on Omarchy 4

Drop-ins are read in lexicographic filename order and the last file to set an option wins, so a file named 10-inhibit.conf is silently overridden by Omarchy's 20-inhibit-delay.conf and changes nothing. Name yours so it sorts after. systemd recommends the 60 to 90 range for local drop-ins:

/etc/systemd/logind.conf.d/90-inhibit.conf:


[Login]
InhibitDelayMaxSec=20

sudo systemctl reload systemd-logind

Prefer reload over restart. systemctl show systemd-logind -p CanReload prints CanReload=yes, and a restart can tear down the graphical session.

**4. Make lid close respect block inhibitors**

By default LidSwitchIgnoreInhibited=yes, so closing the lid suspends regardless of who is blocking. To make it obey, and again using a filename that sorts after Omarchy's own drop-ins:

/etc/systemd/logind.conf.d/90-lid.conf:


[Login]
LidSwitchIgnoreInhibited=no

(SuspendKeyIgnoreInhibited=, HibernateKeyIgnoreInhibited=, PowerKeyIgnoreInhibited= and RebootKeyIgnoreInhibited= already default to no.)

**5. Omarchy 4 has no hypridle, so ignore every hypridle recipe**

pacman -Q hypridle returns "package not found" and there is no ~/.config/hypr/hypridle.conf. Idle detection, the screensaver and the lock are a Quickshell plugin at /usr/share/omarchy/shell/plugins/services/idle/. The hypridle options inhibit_sleep, ignore_dbus_inhibit and ignore_systemd_inhibit do not exist here. Anything telling you to edit hypridle.conf is Omarchy 3 advice.

The one command that dumps the whole idle, screensaver and lock picture:


omarchy-debug-idle

The two IPC calls it wraps, if you want just those:


omarchy-shell idle status | jq .
omarchy-shell lock status | jq .

**6. The Omarchy trap: stay-awake**

omarchy-toggle-idle does not start or stop a daemon. It creates or removes ~/.local/state/omarchy/indicators/stay-awake, and while that file exists the Quickshell idle monitor is disabled outright. Nothing dims, screensaves or locks, and systemd-inhibit --list shows nothing at all, because no inhibitor is involved.


omarchy-toggle-idle status     # {"enabled":true,...} means stay-awake is ON, idle is OFF
ls -l ~/.local/state/omarchy/indicators/stay-awake
omarchy-toggle-idle allow-idle # idle back on
omarchy-toggle-idle stay-awake # idle off, on purpose

The accepted arguments are toggle, stay-awake, allow-idle and status. Running it twice is a round trip back to where you started, not a restart.

**7. An app blocking idle on Omarchy 4 is a Wayland inhibitor, not a logind one**

The Quickshell idle monitor runs with respectInhibitors: true, so a browser playing video takes an idle-inhibit-unstable-v1 lock that the compositor honours and logind never sees. systemd-inhibit --list --what=idle will be empty while the screen refuses to lock. Ask Hyprland instead:


hyprctl clients -j | jq -r '.[] | select(.inhibitingIdle == true) | [.pid,.class,.title] | @tsv'

**8. logind IdleAction is inert on Omarchy 4**

loginctl idle hints and logind's IdleAction= only matter on a plain Arch or systemd desktop. Omarchy leaves IdleAction=ignore, so this branch will never explain an Omarchy symptom:


busctl get-property org.freedesktop.login1 /org/freedesktop/login1 \
  org.freedesktop.login1.Manager IdleAction      # s "ignore" on Omarchy 4
loginctl list-sessions
loginctl show-session "$XDG_SESSION_ID" -p IdleHint -p IdleSinceHint

**9. If the screen locked and will not come back**

Omarchy 4's lock is an ext-session-lock surface drawn by omarchy-shell. hyprlock is not installed, the lock IPC exposes no unlock(), and the compositor stays locked even if the lock client dies. There is no headless way out: type the password at the console, or power cycle. Detect the stale case with:


omarchy-hyprland-session-locked; echo $?    # 0 locked, 1 unlocked, 2 undetermined

**10. Take a deliberate lock yourself** when you want to protect a long job:


systemd-inhibit --what=sleep --why="backup running" -- restic backup /home

--mode= defaults to block and --what= defaults to idle:sleep:shutdown, so the explicit --what=sleep above narrows it.

Verify

systemd-inhibit --list --mode=block is empty (or lists only things you expect), then systemctl suspend without -i actually suspends. On Omarchy 4, systemd-inhibit --list --what=sleep should still show three delay entries, including Omarchy with Lock screen before suspend, and systemctl --user is-active omarchy-sleep-lock.service prints active. For an idle or lock complaint on Omarchy 4, omarchy-toggle-idle status prints {"enabled":false,...} once stay-awake is off, omarchy-shell idle status reports "enabled":true, and hyprctl clients -j | jq '.[] | select(.inhibitingIdle == true)' returns nothing once the offending app is closed.

Sources

suspend-blocked-by-inhibitor-lock