Skip to content

Releases: abbbi/virtnbdbackup

Version 2.28

16 Apr 18:26
Compare
Choose a tag to compare

Version 2.28

  • virtnbdbackup:
    • improve overlap() performance. The call should be O(L) complexity (#265) ( @dlunev)

Full Changelog: v2.27...v2.28

Version 2.27

15 Apr 15:28
Compare
Choose a tag to compare

Version 2.27

  • virtnbdbackup: Skip saving checksum information for raw images included with --raw option:
    The checksum calculated does not actually match during verify because certain
    parts of the images may be seeked instead of actually written. (#269)

Full Changelog: v2.26...v2.27

Version 2.26

03 Apr 11:03
Compare
Choose a tag to compare

Version 2.26

virtnbdbackup:

  • Add info log message before and after extent query has finished.
  • Enhance function to detect sparse/fstrimmed regions (#263).
  • Enhance CI tests

Full Changelog: v2.25...v2.26

Version 2.25

02 Apr 07:05
Compare
Choose a tag to compare

Version 2.25

Detect QCOW2 formatted direct attached RAW Volumes (#264):

Some solutions such as OpenNebula will use direct attached LVM volumes in
virtual machines. These LVM volumes were treated like a RAW device, but
allow all required operations for full/incremental backup, as they are qcow2
formatted.

Changes introduced in v2.18 would exclude these devices from backup even if
--raw option is passed. Now during disk detection, use the disk format in
the driver type setting to correctly detect a lvm volume backed by qcow2,
and handle them just like regular qcow based images.

Full Changelog: v2.24...v2.25

Version 2.24

25 Mar 20:00
Compare
Choose a tag to compare

Version 2.24

virnbdbackup:

  • Add option --no-sparse-detection (#263): The feature implemented as
    with version 2.21 requires to query the complete base:allocation
    bitmap during incremental backup.
    Depending on the Hypervisor hardware and the virtual machine disk size,
    querying the base:allocation bitmap may take longer than to simply treat
    fstrimmed blocks the same way as changed blocks and include them in the
    backup.
    The option introduced can be used to disable the new behavior and can
    enhance backup times for specific users.

See also:

https://github.com/abbbi/virtnbdbackup?tab=readme-ov-file#fstrim-and-incremental-backup-sizes

Full Changelog: v2.23...v2.24

Version 2.23

21 Mar 16:53
Compare
Choose a tag to compare

Version 2.23

virtnbdbackup:

  • Log created nbd socket information in URI format, so it can be re-used with external tools more easily.
  • Add human readable backup sizes in log output.

Full Changelog: v2.22...v2.23

Version 2.22

12 Mar 07:56
Compare
Choose a tag to compare

Version 2.22

  • Test enhancements:
    • check btrfs filesystem stats in restored VM.
    • run checksum verify after all backups.
  • virtnbdbackup:
    • If NBD connection drops, catch correct exception. (#254)
    • Provide better error message if checkpoint inconsistency is detected.

Full Changelog: v2.21...v2.22

Version 2.21

20 Feb 20:07
4b2f626
Compare
Choose a tag to compare

Version 2.21

virtnbdbackup: Detect dirty regions containing only zeroes during
incremental backup (fstrim) (#250)

Before, any dirty block being part of the incremental/differential bitmap
would be backed up, even if not required. Thanks to the input and
enhancements by

Roman Grigoriev (grigoriev177 at gmail.com) and
Denis V. Lunev (den at openvz.org)

the zeroed block regions are now compared to the base:allocation
bitmap and detected as such.

This reduces the amount of data backed up if virtual machines have disks
configured with discard=unmap option and use fstrim to free unused space.

Full Changelog: v2.20...v2.21

Version 2.20

07 Feb 11:25
Compare
Choose a tag to compare

Version 2.20

  • virtnbdmap: add option -H for better debugging, code cleanup
  • virtnbdmap: fix situations where multiple blocks in row are required to map the correct data.

Full Changelog: v2.19...v2.20

Version 2.19

05 Feb 12:13
Compare
Choose a tag to compare

Version 2.19

  • virtnbdmap sometimes fails to correctly map disk images due to
    block offsets spanning multiple frames in the stream format, resulting in
    Input/Output errors accessing the disk partition data (#249)

Full Changelog: v2.18...v2.19