#2237 PR merged: Fix including of multipath disks in backup

Labels: enhancement, fixed / solved / done

petrpavlu opened issue at 2019-09-12 13:32:

Pull Request Details:

  • Type: Bug Fix / New Feature / Enhancement / Other?

    Bug Fix.

  • Impact: Low / Normal / High / Critical / Urgent

    Normal.

  • Reference to related issue (URL):

    https://github.com/rear/rear/issues/2236.

  • How was this pull request tested?

    Locally tested under KVM. Multipath configuration:

    # multipath -ll
0QEMU_QEMU_HARDDISK_0001 dm-0 QEMU,QEMU HARDDISK
size=2.0G features='1 retain_attached_hw_handler' hwhandler='0' wp=rw
|-+- policy='service-time 0' prio=1 status=active
| `- 2:0:0:1 sdb 8:16 active ready running
`-+- policy='service-time 0' prio=1 status=enabled
  `- 2:0:0:2 sda 8:0  active ready running

# parted /dev/mapper/0QEMU_QEMU_HARDDISK_0001 print
Model: Linux device-mapper (multipath) (dm)
Disk /dev/mapper/0QEMU_QEMU_HARDDISK_0001: 2147MB
Sector size (logical/physical): 512B/512B
Partition Table: gpt
Disk Flags:

Number  Start   End     Size    File system  Name     Flags
 1      1049kB  211MB   210MB   ext4         primary
 2      211MB   2147MB  1937MB  xfs          primary

# mount | grep 'on \(/ \|/mnt\)'
/dev/vda2 on / type btrfs (rw,relatime,space_cache,subvolid=259,subvol=/@/.snapshots/1/snapshot)
/dev/mapper/0QEMU_QEMU_HARDDISK_0001-part1 on /mnt/mp-1 type ext4 (rw,relatime,data=ordered)
/dev/mapper/0QEMU_QEMU_HARDDISK_0001-part2 on /mnt/mp-2 type xfs (rw,relatime,attr2,inode64,noquota)

  • Brief description of the changes in this pull request:

    The AUTOEXCLUDE_DISKS logic (320_autoexclude.sh) traverses filesystems and tries to determine the associated underlying disks that should be kept. When having a filesystem fs:/A that is present on a multipath device MP and the root filesystem fs:/ that is present on a normal disk D, the code first established that fs:/A depends on fs:/ resulting in marking D as used and then short-circuited processing of multipath devices. This resulted in MP not being considered as a used device and removed from the backup.

    The patch fixes this problem by removing the short-circuit logic in find_disk_and_multipath() so multipath devices get processed too (unless AUTOEXCLUDE_MULTIPATH is true).

    Fixes #2236.

jsmeix commented at 2019-09-13 12:03:

@schabrolles @gdha
could you have a look here and review it?

I cannot actually review it because I do not at all know about
how AUTOEXCLUDE_DISKS versus AUTOEXCLUDE_MULTIPATH
interdependencies are meant to work - i.e. I cannot imagine
if this chages may cause unexpected regressions elsewhere.

jsmeix commented at 2019-10-16 08:03:

@schabrolles
if you do not object, I would like to merge it tomorrow.

[Export of Github issue for rear/rear.]