OpenZFS 2.3.1 catches up with the current Linux kernel

OpenZFS has become an integral part of modern IT, and it can be annoying if the super file system is not compatible with current versions of the Linux kernel or FreeBSD. OpenZFS 2.3.0 was technically a significant step forward, but runs at most with Linux kernel 6.12 from the end of last year. With OpenZFS 2.3.1, the current Linux kernel 6.13 from the end of January can also be used. Even Linux kernel 6.14, which is expected in around two weeks, has been partially taken into account in OpenZFS 2.3.1 – An update to OpenZFS 2.3.2 should therefore follow shortly after the Linux kernel.

Support for FreeBSD has been extended in OpenZFS 2.3.1 down to version 12.2. This is astonishing, as this version dropped out of support two years ago; FreeBSD 14.2 or 13.5 is currently available, with 15.0 due in December. The minor update 2.3.1 does not offer any new functionality, but the list of small bug fixes is quite long. For example, a problem with empty metaslabs has been fixed. Metaslabs in OpenZFS are storage blocks into which hard disk partitions are divided in order to optimize the management of free and occupied areas. They enable efficient I/O distribution and minimize fragmentation by dynamically allocating storage space during read and write operations.

OpenZFS 2.3.x: RAIDZ expandable during operation

The recently released OpenZFS 2.3.0 brought three really good innovations: RAIDZ expansion allows a RAIDZ array to be extended by additional drives. At present, only a single drive can be added at a time. In this way, simple and small NAS systems in particular can be easily expanded during operation. Previously, it was possible to replace individual drives with drives with more capacity and then "resilver" the RAIDZ until all drives had been replaced. The trick of "use sparse files for parity and offline them" was also creepy. The new RAIDZ expansion has been worked on for more than two years and is still being enthusiastically discussed. The practical function should still be used with caution.

Summarized under "Fast Dedup", OpenZFS 2.3.0 brought several improvements and optimizations for the OpenZFS Block Deduplication System. Finally, the Direct I/O support of OpenZFS 2.3.0 has significantly accelerated modern NVMe solid-state storage by optionally bypassing the ARC for read/write operations. Long names for files/directories with up to 1023 characters, JSON output support for OpenZFS tools, CPU pinning and a variety of bug fixes are very good arguments for an update from the sometimes problematic OpenZFS 2.2 branch.

(mki)