| View previous topic :: View next topic |
| Author |
Message |
divined
Apprentice
Joined: 23 Jul 2004
Posts: 165
|
 Posted: Sat Jul 21, 2018 8:24 am Post subject: Problem mounting mdadm RAID on kernel >=4.16.6 |
 |
|
Hello everybody
I've been experiencing this problem mounting my RAID device on kernel version 4.17.6 and greater. Everything falls back to normal once I boot into the 4.17.4 kernel. Specifically, although I can see my RAID being active when executing
| Code: |
cat /proc/mdstat
Personalities : [raid6] [raid5] [raid4] [raid0] [raid1] [raid10]
md127 : active (auto-read-only) raid1 sdc1[1] sdb1[0]
976629760 blocks super 1.2 [2/2] [UU]
bitmap: 0/8 pages [0KB], 65536KB chunk
|
the system cannot mount the device.
| Code: |
mount /dev/md127 /mnt/raid
mount: /mnt/raid: mount(2) system call failed: Structure needs cleaning.
|
I don't believe the structure needs cleaning since I ran fsck on the RAID and no errors popped up. Could something have changed with the latest kernel revisions or could this be a bug? |
|
| Back to top |
|
 |
Hu
Administrator
Joined: 06 Mar 2007
Posts: 23280
|
| Posted: Sat Jul 21, 2018 4:59 pm Post subject: |
|
|
| Is the problem fixed in 4.17.7? If not, can you bisect to find the specific patch that breaks this? |
|
| Back to top |
|
|
divined
Apprentice
Joined: 23 Jul 2004
Posts: 165
|
| Posted: Sun Jul 22, 2018 6:32 am Post subject: |
|
|
| The problem seems to persist even in the 4.17.8 kernel. I wouldn't know what further info would be required in order to further clarify and pinpoint the problem. |
|
| Back to top |
|
|
Hu
Administrator
Joined: 06 Mar 2007
Posts: 23280
|
| Posted: Sun Jul 22, 2018 5:22 pm Post subject: |
|
|
| Upstream developers are more receptive to problem reports if you can identify a specific patch, not just a release, in which the problem first appeared. To do that, you need to test the individual patches that are present in the bad kernels and not present in the good kernels. This is a common task, so Git has a helper to let you find this efficiently: git bisect. Tell it the good and bad kernel commits. It will pick a commit midway between and ask you to test that one. Tell it whether the candidate commit is good or bad. It will pick a commit midway between this new point and the appropriate end point based on your answer. This repeats until you identify a specific commit as bad and its parent as good. At that point, you verify your work and report it upstream. |
|
| Back to top |
|
|
russK
l33t
Joined: 27 Jun 2006
Posts: 665
|
| Posted: Sun Jul 22, 2018 5:37 pm Post subject: |
|
|
| Have you tried looking at the RAID with 'mdadm --detail' and the devices with 'mdadm --examine'? |
|
| Back to top |
|
|
axl
Veteran
Joined: 11 Oct 2002
Posts: 1146
Location: Romania
|
| Posted: Sun Jul 22, 2018 5:49 pm Post subject: |
|
|
| the array seems to be in read-only. why? any messages in dmesg about the array? does the old kernel work? |
|
| Back to top |
|
|
|
Kernel & Hardware
