Reiser4 is the successor filesystem for ReiserFS, initially designed and developed entirely from scratch by Namesys and Hans Reiser.
It is very efficient in handling many files (often used in
/var for this purpose) and includes a plugin-based design with plugins with features such as intelligent transparent compression, both inline data and meta-data checksums through the crc32c algorithm with added optional mirrors and failover support through its own implementation of subvolumes.
Reiser4 supports different transaction models optimized for different types of storage devices. These include Write-Anywhere (or CoW), journal and a unique combination of the two called hybrid. Reiser4's main goals is performance as well as the modular nature which can expand features of the file system more easily, as well as the implementations that make it less prone to both data loss and corruption because of it's atomic approach in regards to every transaction that it performs. In the hybrid transaction model, Reiser4 heuristically alternates between journaling and Write-Anywhere/Copy-on-Write and is also the default if no mount option is specificed and this transaction model is optimized primarily for performance for rotating disks (while Write-Anywhere is aimed towards SSDs). However, in hybrid mode, it will result in more fragmentation compared to the journal transaction model over time.
However, Reiser4's journaling model (in both hybrid or journal modes) is more efficient than regular journaling file systems in that instead of traditional journaling, Reiser4 uses a more advanced technique known as wandering logs which provides journaling without having to write data to the disk twice thus reducing writes. In many cases, Reiser4 can write journal data to a disk block, then atomically swap the journal block into the file itself which avoids having to write the data twice by changing one's definition of where the log area and the committed area are instead of moving the data from the log to the committed area.
Another unique feature is the implementation of a special technique for SSD users called Precise Asynchronous Discard. This implementation means that in Reiser4, issuing discard requests is a delayed action, which is performed on a per-atom basis at transaction commit time. It allows to reduce number of discard requests issued (because the merging of extents, which needs to be discarded) and eliminates the need for issuing discards periodically as well, thus removing the need to set up an external tool like fstrim through a cron daemon which issues discard requests on a schedule.
Additionally, it also features block suballocation and is built on the dancing tree design, which was exclusively invented for the use with Reiser4 and is also one of the several unique but also fundamental differences when compared to every other file system in general. The idea behind the dancing tree approach is to speed up file system operations by delaying optimization/balancing of the tree and only writing to disk when necessary, as writing to disk is much slower than writing to memory. Because this optimization is done less often than with other tree data structures like B+ tree's or B- tree's, the optimizations and performance gains can be more extensive.
Because it is designed as an atomic file system, "your file system operations either entirely occur, or they entirely do not, and they do not corrupt due to half occurring." What this means in practice is that all operations in Reiser4 are atomic. In the case of very long writes, Reiser4 is forced to close transactions to free dirty pages in a response to memory pressure and thus, long writes in Reiser4 are split into a number of atomic writes to maintain is atomicity. However, current lead developer, mathematician and programmer PhD Edward Shishkin has suggested a change to full atomicity (atomic writes of any length) in the Write-Anywhere transaction model, where atoms can be flushed without closing a transaction.
It's core design is very modular and allows for easy integration of new modules (called "plugins"). According to the developers, because of the modular design, Reiser4 can quite easily be ported to other operating systems like FreeBSD. As of 2019, Linux is still the only operating system Reiser4 is known to be compatible with.
Synthetic benchmarks and speed comparisons between Reiser4, Btrfs and Ext4 are available, as well as benchmarks that include comparisons between the different transaction models and with/without the cryptcompress plugin.
More information about the features, plugins, design and mkfs or mount options is described in detail here.
- Reiser4 requires a patched kernel.
- It consumes a little more CPU than other filesystems (just like Btrfs). To avoid having issues on laptops using TLP for power saving, it is recommended to disable the options for SATA Link power saving in /etc/default/tlp (again, as with Btrfs).
- Even LILO as the only bootloader officially supporting Reiser4 seems to have issues with it when
/bootis formatted as Reiser4
- It is still not included in the official Linux kernel, but patches for Linux-5.x is already available.
- Access Control Lists is not implemented and as of linux-5.x.x requires that Systemd/Journal either logs to a seperate logging daemon or to Tmpfs. Another workaround is to compile systemD by source without ACL support, but is not recommended.
1. Install the AUR package which provides utilities for creating, fsck'ing and debugging Reiser4 volumes.
2. You will need a Reiser4 patched kernel. Patches can be found here or at the more recently created Git repository which is maintained by it's current lead developer, the mathematician and programmer Edward Shishkin.
3. Bootloader (Optional, only needed if you want to format your
/ (root, including /boot) as Reiser4)
a) Recommended: make a small (as mentioned above, 20-200mb) partition for
/boot with a filesystem other than Reiser4 with GParted, and then copy your
/boot folder to the partition. Update your bootloader config accordingly, eg. with Grub2 do:
# grub-mkconfig -o /boot/grub/grub.cfg.
b) If you do not use EFI and wish to put everything including
/boot on a Reiser4 partition (not recommended) you will need to use LILO. This is not advised, as you will probably get an error when trying to update
Moving to Reiser4
In the next steps we will copy the data from your current root partition to the new Reiser4 partitions. Make sure you have enough disk space on the Reiser4 partition with:
# df -h
# fdisk -l * /dev/sda1: (10 Gb, 5 Gb free); Reiserfs /mnt/reiser4 * /dev/sda2: (10 Gb, 10 Gb free); Reiser4 / * /dev/sda3: (200 Mb, 180 Mb free); ext2 /boot
Since Reiser4 supports different transaction models optimized for different types of storage media (SSDs, HDDs), the options used while formatting and mounting will differ.
Keep in mind that the defaults for mkfs.reiser4 includes enabled compression with the default algorithm being Zstd.If one wishes to use either lzo or gzip instead of Zstd, it is needed to append
If one wishes to disable compression altogether, one must append:
-o create=reg40to the
mkfs.reiser4command. Additionally, the inline checksum plugin can be enabled with
-o node=node41More information about the features, plugins and options is available here.
mkfs.reiser4 /dev/sdaX mkdir /mnt/reiser4 mount -t reiser4 -o txmod=journal,noatime,onerror=remount-ro /dev/sdaX /mnt/reiser4
It is recommended and also the default to use the Cryptcompress plugin by formatting with the following options:
mkfs.reiser4 -o create=ccreg40 /dev/sdaX
Since Reiser4 also has options specifically for SSD users as well, it is recommended to discard the partition upon creation of the filesystem, the -d switch can be applied as shown below:
mkfs.reiser4 -d -o create=ccreg40,compress=Zstd1 /dev/nvme0n1X
For drives with controllers already having hardware compression (like SandForce ones), it may be better to to disable the compression plugin.
mkfs.reiser4 -o create=reg40,node=node41 /dev/sdX
mkfs.reiser4 -d -o create=reg40,node=node41 /dev/nvme0n1pX
Once the partition is formatted, copy you current system to the new partition and create the system directories. You may either do this from Arch Linux, or to make it easier (so that you do not have to use makedev later), just boot up with the Gparted LiveCD and mount both your new Reiser4 partition and your current root partition. Then, just copy everything over (as root) like so:
cd /mnt mkdir oldroot mkdir reiser4 mount /dev/sdaX oldroot
Depending on what transaction model one wish to use which are optimized for different types of storage media, the mount option txmod=wa (for SSDs), txmod=journal (for HDDs) must be defined when mounting the partitions through the -o switch. The default is txmod=hybrid which heuristically alternates between the "wa" (write-anywhere) and "journal" models for optimized performance on rotating disks while trying to avoid excess fragmentation at the same time.
mount -t reiser4 -o txmod=hybrid,noatime,onerror=remount-ro /dev/sdaY reiser4 (the Reiser4 partition) cp -R -a /mnt/oldroot/* /mnt/reiser4/
Then, you need to mount your
/boot partition, and if you have not already, copy
/boot from your original root partition over to it.
mkdir bootpart mount /dev/sdaZ bootpart cp -R -a /mnt/oldroot/boot/* /mnt/bootpart/
Do not forget to edit your bootloader's config appropriately (see examples at the bottom of the article).
Note: If you can confirm that Reiser4 works for you, you should format the old root partition.
# # /etc/fstab: static file system information # # <file system> <dir> <type> <options> <dump> <pass> /dev/nvme0n1p1 / reiser4 noatime,txmod=wa,onerror=remount-ro,discard 0 1 /dev/sda2 /mnt/oldroot ext4 defaults 0 0 /dev/sda3 /boot ext2 defaults 0 1
# (0) Arch Linux title Arch Linux set root=(hd0,msdos3) kernel /vmlinuz-linux root=/dev/sda3 ro rootfstype=reiser4 rootflags=noatime,txmod=journal,onerror=remount-ro init=/usr/bin/bootchartd initrd /initramfs-linux.img # (1) Arch Linux title Arch Linux Fallback set root=(hd0,msdos3) kernel /vlinuz-linux root=/dev/sda3 ro rootfstype=reiser4 rootflags=noatime,txmod=journal,onerror=remount-ro initrd /initramfs-linux-fallback.img
grub-mkconfig to update your config:
# grub-mkconfig -o /boot/grub/grub.cfg
# # /etc/lilo.conf # boot=/dev/hda # This line often fixes L40 errors on bootup # disk=/dev/hda bios=0x80 default=Arch4 timeout=20 lba32 prompt compact image=/boot/vmlinuz-linux label=Arch4 root=/dev/hda5 append="video=vesafb:1024x768-24@56,ywrap,mtrr splash=verbose,theme:darch console=tty1 resume2=swap:/dev/hdb1" initrd=/boot/initramfs-linux.img read-only image=/boot/vmlinuz-linux label=Arch root=/dev/hda3 append="video=vesafb:1024x768-24@56,ywrap,mtrr splash=verbose,theme:darch console=tty1 resume2=swap:/dev/hdb1" initrd=/boot/initramfs-linux.img read-only
Run lilo to update your config:
- Permissions: chown -R username.group <userdir>
- If you have problem with su command after the change of fs, you should reinstall coreutils package.