swap exists for hibernation, not for memory pressure - 31gib of ram on a
desktop rarely needs paging. the hibernation image must fit in swap so the
floor is "as big as ram"; 40gib leaves headroom. the 3090's vram does not
land here: with NVreg_PreserveVideoMemoryAllocations=1 the driver spools it
to NVreg_TemporaryFilePath, which nixpkgs leaves unset and so defaults to
/tmp. that is only safe while /tmp is on the real root - boot.tmp.useTmpfs
would route a full vram dump into ram and break hibernation outright.
resuming needs both boot.resumeDevice and resume_offset. the device half is
mandatory rather than optional here because boot.initrd.systemd.enable is
true, and under systemd initrd the resume= parameter is emitted only from
boot.resumeDevice - the legacy stage-1 script that scans swapDevices is not
in play. without it hibernate succeeds and the machine then cold-boots,
silently discarding the session.
resume_offset is the physical block offset of the swapfile's first extent,
since the kernel addresses the file by raw block. that value is hand-copied,
and a stale one fails silently: no swsusp signature at the given block means
`PM: Image not found (code -22)` and a cold boot, with no error at hibernate
time. so an activation check warns on drift. it is a detector, not a gate -
nixpkgs only recreates the swapfile when size changes, so on the rebuild
that resizes it the check still sees the old file and fires on the next one.
it warns rather than fails, and expects to read nothing on a fresh install
where mkswap-var-lib-swapfile.service has not run yet.
sleeping via suspend-then-hibernate (hypridle invokes it): s3 first so a
quick return is instant, then a real suspend-to-disk once the machine has
clearly been abandoned. deliberately not hybrid-sleep, which writes the disk
image up front on every single sleep - measured here at ~59s of filesystem
sync plus an 11gib write before the screens even went off, and because an s3
wake invalidates the image the next boot always reported `Image not found`.
full hibernation price paid every time, redeemed only on a power cut.
suspend-then-hibernate defers that write until it is worth doing, and unlike
hybrid-sleep it is a mode nvidia explicitly supports.
HibernateDelaySec is explicit because systemd otherwise derives the handoff
from battery discharge estimation, and this desktop has no system battery -
the sole /sys/class/power_supply entry is a logitech mouse reporting
scope=Device, which is ignored for that purpose.
also corrects a comment: a sparse swapfile does not silently corrupt.
swapon refuses files with holes outright ("skipping - it appears to have
holes"), so the failure is loud and the system simply ends up with no swap.
genuine silent corruption is the copy-on-write case, which is why btrfs
needs a NOCOW-from-creation file.
sdb (GAMES) is an ssd - samsung 870 qvo, rotational=0 - so unlike the two wd
spinners there is no spin-down to gain from automounting it, and a steam library
is better off never sitting behind a trigger. keeps nofail. dump and bak stay
lazy.
splits the old dataDrive helper into spinningDrive and eagerDrive and documents
why each drive gets which.
also records the steam gotcha found while debugging "steam library folder not
writable": it was never a permissions problem - the drives are owned by uid 1000
mode 755 and were writable throughout. steam's fhs sandbox builds its bind list
at launch by iterating /*, and /mnt does not exist on nixos until these
fileSystems entries are activated, so a steam started before that has no /mnt
for its entire lifetime. restarting steam is the fix.
importing /etc/nixos/hardware-configuration.nix by absolute path from the
shared core module had two problems: it made the flake impure (every eval
needed --impure, since the file lives outside the store) and both hosts
resolved to whichever machine's file was on disk, so evaluating t480 on shadow
silently baked in shadow's disk uuids.
shadow now carries its own generated scan at modules/hosts/shadow/_hardware.nix.
the leading underscore keeps import-tree from picking it up as a flake-parts
module - its filter drops any path containing "/_" - so it is imported
explicitly from the host's default.nix.
it is generated with --no-filesystems and all of shadow's filesystems are
declared by hand in disks.nix instead. that split is what makes regeneration
safe: nixos-generate-config emits fileSystems for whatever is mounted at the
time, which would otherwise conflict with the entries below every time the data
drives happened to be mounted. `nix run .#sync-hardware` refreshes the file.
DUMP/GAMES/BAK are mounted by label at /mnt/{dump,games,bak} with
noauto + x-systemd.automount, so the two 2.7T spinners stay parked until
something touches the path and spin back down after 10 minutes idle. nofail
keeps a missing disk from dropping the boot into emergency mode.
/ and /boot are unchanged - still the same uuids, verified identical to the
previous evaluation. t480 keeps the absolute import for now with a TODO; its
scan can only be generated on the machine itself.
ahmza