mjhjmtu

I'm running Fedora Server Edition 26 and I have two external USB drives with a partition on each that I have combined in RAID1. My /etc/fstab file has this line for automounting the array:

UUID=B0C4-A677 /mnt/backup-raid exfat uid=strwrsdbz,gid=strwrsdbz,umask=022,windows_names,locale=en.utf8,nobootwait,nofail 0 2

However, when booting is finished the array at /mnt/backup-raid is not mounted. If I check the journal logs I see

Oct 28 21:32:07 hostname systemd[1]: Started File System Check on /dev/disk/by-uuid/B0C4-A677.
Oct 28 21:32:07 hostname audit[1]: SERVICE_START pid=1 uid=0 auid=4294967295 ses=4294967295 subj=system_u:system_r:init_t:s0 msg='unit=systemd-fsck@dev-disk-byx2duuid-B0C4x2dA677 comm="systemd" exe="/usr/lib/systemd/systemd" hostname=? addr=? terminal=? res=success'
Oct 28 21:32:07 hostname kernel: audit: type=1130 audit(1509240727.851:75): pid=1 uid=0 auid=4294967295 ses=4294967295 subj=system_u:system_r:init_t:s0 msg='unit=systemd-fsck@dev-disk-byx2duuid-B0C4x2dA677 comm="systemd" exe="/usr/lib/systemd/systemd" hostname=? addr=? terminal=? res=success'
Oct 28 21:32:08 hostname systemd[1]: Mounting /mnt/backup-raid...
Oct 28 21:32:08 hostname systemd[1]: Mounted /mnt/c.
Oct 28 21:32:08 hostname ntfs-3g[702]: Version 2017.3.23 integrated FUSE 28
Oct 28 21:32:08 hostname ntfs-3g[702]: Mounted /dev/sda1 (Read-Write, label "", NTFS 3.1)
Oct 28 21:32:08 hostname ntfs-3g[702]: Cmdline options: rw,uid=1000,gid=1000,umask=022,windows_names,locale=en.utf8
Oct 28 21:32:08 hostname ntfs-3g[702]: Mount options: rw,allow_other,nonempty,relatime,default_permissions,fsname=/dev/sda1,blkdev,blksize=4096
Oct 28 21:32:08 hostname ntfs-3g[702]: Global ownership and permissions enforced, configuration type 7
Oct 28 21:32:08 hostname lvm[599]: 3 logical volume(s) in volume group "fedora" now active
Oct 28 21:32:08 hostname systemd[1]: Started LVM2 PV scan on device 8:5.
Oct 28 21:32:08 hostname audit[1]: SERVICE_START pid=1 uid=0 auid=4294967295 ses=4294967295 subj=system_u:system_r:init_t:s0 msg='unit=lvm2-pvscan@8:5 comm="systemd" exe="/usr/lib/systemd/systemd" hostname=? addr=? terminal=? res=success'
Oct 28 21:32:08 hostname kernel: audit: type=1130 audit(1509240728.594:76): pid=1 uid=0 auid=4294967295 ses=4294967295 subj=system_u:system_r:init_t:s0 msg='unit=lvm2-pvscan@8:5 comm="systemd" exe="/usr/lib/systemd/systemd" hostname=? addr=? terminal=? res=success'
Oct 28 21:32:08 hostname systemd[1]: Found device /dev/mapper/fedora-home.
Oct 28 21:32:08 hostname systemd[1]: Mounting /home...
Oct 28 21:32:08 hostname kernel: XFS (dm-2): Mounting V4 Filesystem
Oct 28 21:32:08 hostname systemd[1]: Mounted /mnt/igel1.
Oct 28 21:32:08 hostname systemd-fsck[666]: /dev/sda3: clean, 376/128016 files, 291819/512000 blocks
Oct 28 21:32:08 hostname systemd[1]: Mounted /mnt/igel2.
Oct 28 21:32:08 hostname systemd[1]: Mounted /mnt/backup-raid.

*snip*

Oct 28 21:32:33 hostname systemd[1]: Created slice system-mdadmx2dlastx2dresort.slice.
Oct 28 21:32:33 hostname systemd[1]: Starting Activate md array even though degraded...
Oct 28 21:32:33 hostname systemd[1]: Unmounting /mnt/backup-raid...
Oct 28 21:32:34 hostname systemd[1]: Started Activate md array even though degraded.
Oct 28 21:32:34 hostname audit[1]: SERVICE_START pid=1 uid=0 auid=4294967295 ses=4294967295 subj=system_u:system_r:init_t:s0 msg='unit=mdadm-last-resort@md0 comm="systemd" exe="/usr/lib/systemd/systemd" hostname=? addr=? terminal=? res=success'
Oct 28 21:32:34 hostname audit[1]: SERVICE_STOP pid=1 uid=0 auid=4294967295 ses=4294967295 subj=system_u:system_r:init_t:s0 msg='unit=mdadm-last-resort@md0 comm="systemd" exe="/usr/lib/systemd/systemd" hostname=? addr=? terminal=? res=success'
Oct 28 21:32:34 hostname kernel: md0:
Oct 28 21:32:34 hostname systemd[1]: Unmounted /mnt/backup-raid.

So it looks like it gets mounted in that first log block but then later it gets unmounted because it is appearing as degraded. But once it's finished booting I can run sudo mount -a and the array mounts without issue. The contents appear correctly in /mnt/backup-raid and checking /proc/mdstat shows

Personalities : [raid1]
md0 : active raid1 sdc2[0] sdb2[2]
 485345344 blocks super 1.2 [2/2] [UU]
 bitmap: 0/4 pages [0KB], 65536KB chunk

unused devices: <none>

so everything looks healthy. In case it helps, my /etc/mdadm.conf contains

ARRAY /dev/md0 level=raid1 num-devices=2 metadata=1.2 name=hostname:backup-raid UUID=6c8bf3df:c4147eb1:4c3f88d8:e94d1dbc devices=/dev/sdb2,/dev/sdc2

I found this email thread which appeared to be dealing with a similar situation but it looks to me like it just went silent. I'm sorry if the answer is in that email thread and I missed it but it just gets a bit too dense for me to follow.

The 'Errorneous detection of degraded array' systemd-devel thread is about a race condition between udev and the mdadm-last-resort timer/service. The Conflicts=sys-devices-virtual-block-%i.device line then triggers the umount of the previously mounted filesystem.

The thread also mentions a workaround that should also fix your issue: replace the Conflicts=... line with a ConditionPathExissts=... line:

# cp /usr/lib/systemd/system/mdadm-last-resort@.* /etc/systemd/system/
# sed -i 's@^Conflicts=sys-devices-virtual-block-%i.device@ConditionPathExists=/sys/devices/virtual/block/%i@' 
 /etc/systemd/system/mdadm-last-resort@.*
# shutdown -r now

Note that a drop-in replacement via /etc/systemd/system/.../override.conf doesn't work for removing the Conflicts= line.

You can subscribe to the related upstream systemd issue Need a uni-directional version of "Conflicts" to get notified about changes regarding the underlying issue.

See also my Fedora 27 bug report where this issue manifests itself in /boot/efi not being mounted when placed on a RAID-1 mirror.