I have multiple gentoo servers with root filesystem on raid1 btrfs.
This bug affects all my raid1 btrfs installations but is not present systems with non-raid1 btrfs rootfs.
Example of my fstab
LABEL=rootfs / btrfs compress=zstd,subvol=root 0 0
LABEL=rootfs /mnt/a btrfs compress=zstd,subvolid=0 0 0
LABEL=rootfs /var/lib/containers btrfs compress=zstd,subvol=containers 0 0
LABEL=rootfs /var/lib/libvirt btrfs compress=zstd,subvol=vm 0 0
LABEL=rootfs /var/tmp btrfs defaults,nodev,nosuid,noatime,compress=zstd,subvol=var_tmp 0 0
LABEL=rootfs /var/cache btrfs defaults,nodev,nosuid,noatime,compress=zstd,subvol=var_cache 0 0
When I run mount -a -v on system with non raid1 btrfs I get this output:
(mount correctly detects that filesystem/subvolumes are already mounted)
/ : ignored
/mnt/a : already mounted
/var/lib/containers : already mounted
/var/tmp : already mounted
/var/cache : already mounted
When I run mount -a -v on system with raid1 btrfs I get this output:
/ : ignored
/mnt/a : successfully mounted
/var/lib/containers : successfully mounted
/var/lib/libvirt : successfully mounted
/var/tmp : successfully mounted
/var/cache : successfully mounted
mount doesn't detect that filesystems/subvolumes are already mounted and mounts them again.
This leads to ever increasing duplications of mounted fs. Which is really annoying in lsblk output
example lsblk from one of the systems
lsblk /dev/nvme1n1
NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINTS
nvme1n1 259:0 0 476.9G 0 disk
├─nvme1n1p1 259:1 0 64M 0 part /boot/efi
├─nvme1n1p2 259:2 0 4G 0 part /boot
│ /boot
│ /boot
│ /boot
│ /boot
└─nvme1n1p3 259:3 0 472.9G 0 part /mnt/meta_snapshots
/mnt/meta
/var/cache
/var/tmp
/var/lib/libvirt
/var/lib/containers
/mnt/a
/mnt/meta_snapshots
/mnt/meta
/var/cache
/var/tmp
/var/lib/libvirt
/var/lib/containers
/mnt/a
/mnt/meta_snapshots
/mnt/meta
/var/cache
/var/tmp
/var/lib/libvirt
/var/lib/containers
/mnt/a
/mnt/a
/mnt/a
/mnt/meta
/mnt/meta
/var/cache
/mnt/a
/mnt/meta
/var/cache
/mnt/a
/mnt/meta_snapshots
/mnt/meta
/var/cache
/var/tmp
/var/lib/libvirt
/var/lib/containers
/mnt/a
/
in /proc/mounts
cat /proc/mounts | grep containers
/dev/nvme1n1p3 /var/lib/containers btrfs rw,relatime,compress=zstd:3,ssd,discard=async,space_cache=v2,subvolid=262,subvol=/containers 0 0
/dev/nvme1n1p3 /var/lib/containers btrfs rw,relatime,compress=zstd:3,ssd,discard=async,space_cache=v2,subvolid=262,subvol=/containers 0 0
/dev/nvme1n1p3 /var/lib/containers btrfs rw,relatime,compress=zstd:3,ssd,discard=async,space_cache=v2,subvolid=262,subvol=/containers 0 0
/dev/nvme1n1p3 /var/lib/containers btrfs rw,relatime,compress=zstd:3,ssd,discard=async,space_cache=v2,subvolid=262,subvol=/containers 0 0
I've retested this today on util-linux-2.41.2 version available in gentoo. I couldn't use gentoo's git version of this package to build master branch util-linux because build was failing.
I'm testing it on 6.12.47-gentoo-dist and 6.12.41-gentoo-dist kernels but I'm pretty sure it's not kernel related issue.
I don't think it's a regression, I don't remember it ever working correctly.
I have multiple gentoo servers with root filesystem on raid1 btrfs.
This bug affects all my raid1 btrfs installations but is not present systems with non-raid1 btrfs rootfs.
Example of my fstab
When I run
mount -a -von system with non raid1 btrfs I get this output:(mount correctly detects that filesystem/subvolumes are already mounted)
When I run
mount -a -von system with raid1 btrfs I get this output:mountdoesn't detect that filesystems/subvolumes are already mounted and mounts them again.This leads to ever increasing duplications of mounted fs. Which is really annoying in lsblk output
example lsblk from one of the systems
in /proc/mounts
I've retested this today on util-linux-2.41.2 version available in gentoo. I couldn't use gentoo's git version of this package to build master branch util-linux because build was failing.
I'm testing it on 6.12.47-gentoo-dist and 6.12.41-gentoo-dist kernels but I'm pretty sure it's not kernel related issue.
I don't think it's a regression, I don't remember it ever working correctly.