Swap Strategy: Zram, Swapfiles & Partitions

Modern guidance: zram for speed, swapfile/partition only when hibernation demands it.

Zram (recommended default)

# /etc/systemd/zram-generator.conf
[zram0]
zram-size = min(ram, 16384)        # cap 16G regardless of huge RAM
compression-algorithm = zstd
swap-priority = 100

sudo systemctl daemon-reload && sudo systemctl start systemd-zram-setup@zram0
zramctl && swapon --show           # verify

Kernel then compresses cold pages instead of hitting disk — latency drops by orders of magnitude.

Swappiness for zram world

Higher is CORRECT with zram (we WANT reclaim-to-compress):

echo 'vm.swappiness=180' | sudo tee /etc/sysctl.d/90-zram.conf
sysctl vm.page-cluster=0           # avoid oversized reads from compressed store

(Yes — opposite of the ancient "set it to 10" forum wisdom; that predates zram.)

Swapfile when hibernation required

ext4 root:

sudo fallocate -l $(grep MemTotal /proc/meminfo | awk '{print $2}')k /swapfile \
  || sudo dd if=/dev/zero of=/swapfile bs=1M count=16384 status=progress
sudo chmod 600 /swapfile && sudo mkswap /swapfile && sudo swapon /swapfile
printf '/swapfile none swap defaults 0 0\n' | sudo tee -a /etc/fstab

btrfs root extra steps: NOCOW file + no-compression + fs-uuid resume offset:

sudo btrfs filesystem mkswapfile --size 16g /swap/swapfile   # does it all
resume=UUID=<root-uuid> resume_offset=$(sudo btrfs inspect-internal map-swapfile -r /swap)

Wire resume= into cmdline + initramfs resume hook (GRUB Deep Dive, mkinitcpio & Initramfs Presets), then TEST a full hibernate cycle before relying on it.

Diagnosing memory pressure honestly

swapon --show ; free -h ; vmstat 2
psi=/proc/pressure/memory : watch some avg300 lines   # kernel's own verdict