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

Fix one monitor keeping the old wallpaper after a theme change

omarchyomarchy-4hyprlandwaylandnvidiaintelamddesktop
CORRECTED An auditor found the fix wrong or incomplete and rewrote it.
AUDIT NOTE

The recovery and diagnostics are fine: omarchy restart shell is correct, and the readlink/theme.name/hyprctl layers and grim+magick pixel-sampling checks are all read-only and sensible for proving the state is right while the pixels are stale. The problem is the persistence step: Omarchy 4 configures Hyprland in Lua (hypr/hyprland.lua, hypr/bindings.lua, hypr/looknfeel.lua, hypr/monitors.lua are all in omarchy-upgrade-to-quattro's always_copy_config_files), so a bind = ... line dropped into ~/.config/hypr/bindings/custom.conf is never sourced on the omarchy-4 systems this record is tagged for, and the user would conclude the whole fix failed. Also /run/user/1000 hardcodes a uid.

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

Symptom

omarchy theme set kanagawa retints everything and updates the wallpaper on the primary monitor, but the second monitor keeps showing the previous theme's wallpaper indefinitely. The bar on that same monitor updates normally (clock ticking, new colors). Only the background is stale. omarchy background refresh does nothing.

Cause

Two things. On a multi-GPU setup (the stale output is the one behind the cross-GPU blit path, and aquamarine logs GBM: Buffer is marked as multigpu, forcing linear), that output's background surface never commits a new frame after the source changes, and the compositor keeps scanning out its last buffer. A wallpaper is static, so nothing ever forces another redraw. Then the obvious fix is a guaranteed no-op: refreshBackground() in shell/plugins/background/Background.qml ends in transitionBackground(...), which early-returns on if (!path || (!force && finalPath === currentBackground)) return, and the shell already believes it is showing the new wallpaper. Compounding it, the theme reveal animation is gated on a single shared revealProgress root property, so only the first monitor whose frame loads ever runs the reveal.

Fix

The reliable recovery is a full shell restart:


omarchy restart shell

Confirm the state is right and only the pixels are wrong before doing that:


readlink ~/.local/state/omarchy/current/background
cat ~/.local/state/omarchy/current/theme.name
hyprctl layers | grep -c omarchy-background   # one per monitor, as expected

Sample what each output is actually painting:


grim -o HDMI-A-1 /tmp/a.png && magick /tmp/a.png -format '%[pixel:p{50,950}]' info:
grim -o DP-1 /tmp/b.png && magick /tmp/b.png -format '%[pixel:p{50,950}]' info:

**Keybinding correction.** Omarchy 4 configures Hyprland in Lua, not .conf: ~/.config/hypr/bindings/custom.conf is never sourced, so that bind would silently do nothing. Add it to ~/.config/hypr/bindings.lua, following the form of the entries already in that file, then:


hyprctl reload

Verify the bind actually registered:


hyprctl binds | grep -A3 -i 'restart shell'

(On Omarchy 3.x the ~/.config/hypr/bindings/custom.conf + bind = SUPER SHIFT, R, exec, omarchy restart shell form was correct.)

Verify

After omarchy restart shell, grim samples of both outputs return matching colors for the new wallpaper, and both monitors visibly show the same background.

Sources

one-monitor-keeps-old-wallpaper-after-theme-change