Nothing stood between "memory is tight" and "processes die at random":
the kernel OOM killer only fires after an allocation has already failed,
so a machine under pressure thrashes in reclaim until something breaks
somewhere fatal — seen in practice as Hyprland taking SIGBUS mid-memcpy
from a client shm pool that memory pressure was tearing down.
Enable systemd-oomd, which keys on PSI stall time rather than free
pages, and mark app.slice — and only app.slice — as its kill candidate.
Hyprland runs in session.slice, so the compositor is structurally
ineligible as a victim: oomd takes the app that caused the pressure and
the session survives. Thresholds (50% stall over 20s, Fedora's desktop
defaults) live in an oomd.conf.d drop-in; candidacy ships as a vendor
drop-in under /usr/lib/systemd/user so existing users get it on package
upgrade with no per-user seeding.
New installs enable the daemon from enable-services.sh; a migration
covers existing ones, restarting an already-running oomd so it doesn't
keep stale thresholds until reboot.
Deliberately not earlyoom: it triggers only when free RAM and free swap
are both under threshold, so Omarchy's large, mostly-idle swapfiles
keep it asleep through exactly this failure.
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
graphical.target was gated on network-online.target: cups-browsed orders
itself After=network-online.target and sits in multi-user.target, so the
session waited for NetworkManager-wait-online (DHCP/Wi-Fi association)
before coming up. On the bench VM that was ~140ms, but it's the classic
multi-second boot stall on real Wi-Fi.
Mask NetworkManager-wait-online.service so network-online.target is
reached immediately and nothing in the boot waits for connectivity;
NetworkManager still comes up and cups-browsed browses once the link is
live. Mirrors the existing systemd-networkd-wait-online mask. Enabled at
install (enable-services.sh) and masked on existing installs via
migration.
Verified on a fresh boot: the graphical.target critical chain no longer
passes through network-online.target / NetworkManager-wait-online.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Remove legacy online installer entrypoints, collapse migrations for 4.0, and move setup responsibilities into target-side system, hardware, and user commands.
A pass over the install scripts, dev-tools commands, and Hyprland Lua
files that I had stuffed with explain-everything preambles. Most of
those rationales (which files ship where, why hyprctl setenv doesn't
suffice, etc.) belong in commit messages or PR descriptions, not in
code people have to read forever. Kept the few comments that document
genuinely non-obvious behaviour: the keybind-env reason for hl.env in
envs.lua, why the runtime PAM seds stay scripted in
increase-lockout-limit, the chroot/--now distinction in chroot.sh, and
the dev-pkg-test split-install reason.
The previous centralization into enable-services.sh accidentally changed
semantics for three services: docker.socket, iwd.service, and
power-profiles-daemon.service used to be enabled with bare
'sudo systemctl enable' (no --now); chrootable_systemctl_enable promoted
them to 'enable --now' outside chroot.
For iwd this is the riskiest: it's typically what's keeping the installer
online, so starting it mid-install is at best a no-op and at worst can
disrupt the active network connection. For docker.socket and
power-profiles-daemon the change is cosmetic (the socket is already
inactive, the daemon does no harm to start) but the original intent was
deferral to first boot.
New helper in install/helpers/chroot.sh:
chrootable_systemctl_enable_only <unit> # plain enable, no --now
enable-services.sh uses it for the three services that were previously
bare-enabled. The other five (bluetooth, cups, cups-browsed,
avahi-daemon, linux-modules-cleanup) keep the --now behavior they had
under chrootable_systemctl_enable.
The previous setup scattered systemctl enables across docker.sh,
printer.sh, bluetooth.sh, network.sh, powerprofilesctl-rules.sh, and
kernel-modules-hook.sh. Some used chrootable_systemctl_enable; others
used bare 'sudo systemctl enable'. Centralizing makes the install-time
service surface auditable in one place and consistent in chroot/non-chroot
behavior.
install/config/enable-services.sh enables:
- bluetooth.service
- cups.service, cups-browsed.service, avahi-daemon.service
- docker.socket
- iwd.service
- linux-modules-cleanup.service
- power-profiles-daemon.service
Each via chrootable_systemctl_enable, which uses 'enable --now' on a live
system and bare 'enable' in chroot (set OMARCHY_CHROOT_INSTALL=1).
Stays in original script (deliberately):
- sddm.sh: bare 'systemctl enable sddm.service' (no --now during install
would log the user out; lives in install/login/)
- limine-snapper.sh: limine-snapper-sync (bootloader-specific path)
- first-run/firewall.sh: ufw (deferred to first-run for UX)
- hardware/{t2,intel/lpmd,intel/thermald,apple/fix-suspend-nvme}: gated
on hardware detection
Side cleanups:
- install/config/kernel-modules-hook.sh: DELETED (sole line was the enable
for linux-modules-cleanup, now in enable-services.sh).
- install/config/hardware/printer.sh: DELETED (sole purpose was the
three cups/avahi enables; nothing else to do).
- install/config/hardware/network.sh: keep the systemd-networkd-wait-online
disable+mask; drop the iwd enable.
- install/config/hardware/bluetooth.sh: keep the AutoEnable=false sed and
the wireplumber/bt-agent user-session setup; drop the system enable.
- install/config/docker.sh: drop the docker.socket enable.
- install/config/powerprofilesctl-rules.sh: drop the
power-profiles-daemon enable.
- install/config/all.sh: register enable-services.sh; remove the two
deleted entries.
