Setting up swap space: zram, swapfiles & partitions

Swap space extends physical RAM by letting the kernel page out infrequently accessed anonymous memory. This maintains buffer cache efficiency and prevents out-of-memory kernel panics. Linux supports three swap mechanisms: zram, swapfiles on standard filesystems, and dedicated swap partitions.

In-memory compressed swap: zram

zram creates an in-memory compressed block device. The kernel default compression algorithm is lzo-rle; zstd is a common high-ratio alternative. It provides high-speed paging without disk latency or SSD wear.

Configuring zram with systemd-zram-generator

Create or edit /etc/systemd/zram-generator.conf:

# /etc/systemd/zram-generator.conf
[zram0]
# Allocate zram size equal to physical RAM.
# The default is min(ram / 2, 4096). Override here for more aggressive sizing.
zram-size = ram
# Set compression algorithm
compression-algorithm = zstd
# Assign high priority so zram is used before disk swap
swap-priority = 100

Apply and verify the configuration:

# Reload and start service
sudo systemctl daemon-reload
sudo systemctl start systemd-zram-setup@zram0.service

# Verify active zram device
zramctl
swapon --show

Creating swapfiles on filesystems

Swapfiles provide flexibility by allocating swap storage directly on an existing filesystem without repartitioning.

Swapfile on Ext4 / XFS Filesystems

Use dd to create the swapfile, which writes sequentially and is supported on all kernels and filesystems. The mkswap manpage recommends dd over fallocate because preallocated files can contain unwritten extents that some kernels reject with swapon: swapfile has holes (notably ext4 on kernels 5.7–5.8, fixed in 5.9).

# 1. Allocate an 8 GB file (dd is portable; fallocate may fail on some kernels)
sudo dd if=/dev/zero of=/swapfile bs=1M count=8192 status=progress

# 2. Set strict permissions (required)
sudo chmod 600 /swapfile

# 3. Format as Linux swap
sudo mkswap -L "DISK_SWAP" /swapfile

# 4. Enable swapfile
sudo swapon /swapfile

Swapfile on Btrfs Filesystems (btrfs-progs 6.1+)

# 1. Create a dedicated swap subvolume
sudo btrfs subvolume create /swap

# 2. Create the swapfile with automatic NOCOW and preallocation
sudo btrfs filesystem mkswapfile --size 8g /swap/swapfile

# 3. Activate the swapfile
sudo swapon /swap/swapfile

Btrfs requires NOCOW for swapfiles because copy-on-write would fragment the swap block device. The Btrfs guide covers NOCOW and subvolume management in more detail.

Configuring swap priorities in /etc/fstab

Assign priority values with pri= in /etc/fstab to ensure fast swap devices are consumed before secondary storage:

# /etc/fstab swap configuration
# zram0 is automatically prioritized at pri=100 by systemd

# Secondary NVMe Swapfile (Used only when zram is saturated)
/swapfile   none   swap   defaults,pri=10   0   0

# Backup Swap on Secondary Drive
/dev/sdb2   none   swap   defaults,pri=5    0   0

If your swap device is on an SSD, enable discard with the discard mount option or fstrim.timer to maintain performance. See TRIM & discard for configuration details.