OpenZFS 2.0.0

A major update to the file system and its maintenance tools, OpenZFS 2.0.0, has been released. The new version supports Linux kernels starting from 3.10 and FreeBSD kernels starting from version 12.2, and in addition to this, it now combines code for both operating systems in a single repository. Among the largest changes, the developers note the following:

• Added the ability to sequentially (LBA) rebuild a destroyed Mirror vDev RAID array. This mechanism is much faster than traditional “healing” recovery. However, it does not check the block checksums, which is why immediately after its completion, the next step is to start a system integrity check (scrub).

• Restoring L2ARC cache data after a system reboot. The cache itself uses a dedicated amount of RAM, without using the slower hard drive for frequent data access. Now after a reboot the L2ARC cache data will be in place.

• Support for compression in the ZStandard format, which provides a level of compression comparable to GZIP, but at the same time much higher performance. For convenience, the administrator is given the ability to select the compression level to ensure the best balance between performance and saving disk space.

• Ability to select data when transferring using send/receive commands. Now administrators can manually exclude unnecessary or private data from the transfer before copying a snapshot.

• Many other, less significant, but no less pleasant improvements have been implemented, in particular, a pam module has been written for loading folder encryption keys, man pages have been reorganized and documentation updated, added a zfs volume mount generator for systemd, expanded logging in syslog, improved compatibility with system bootloaders, and much more.

• New commands and keys have been added to existing ones, which you can read more about in brief comments on the release.

• A number of internal tools have been optimized in terms of speed and efficient use of system resources.

Full changelog.

Source: linux.org.ru