The solution to the warning “mismatch_cnt is not 0 on /dev/md*”

Replaced once the junk drive in the software RAID1, added it to the raid, it successfully synchronized, installed GRUB.
After a while I received an email message:

Subject: Cron <root@server> /usr/sbin/raid-check
WARNING: mismatch_cnt is not 0 on /dev/md2

In my case, raid-check found that the mismatch_cnt counter is not equal to 0 for /dev/md2, which means that there may be broken sectors on the disk, or it simply needs to be resynchronized. Since I installed GRUB after adding the disk to the raid, this is probably the cause.

Example of viewing the counters of all arrays:

cat /sys/block/md*/md/mismatch_cnt

Or each in turn:

cat /sys/block/md0/md/mismatch_cnt
cat /sys/block/md1/md/mismatch_cnt
cat /sys/block/md2/md/mismatch_cnt

View the status of raids:

cat /sys/block/md*/md/sync_action

If mismatch_cnt is not 0 for any array, then you can try to resynchronize it:

echo 'repair' >/sys/block/md2/md/sync_action

And check:

echo 'check' >/sys/block/md2/md/sync_action

If you want to cancel the action:

echo 'idle' >/sys/block/md2/md/sync_action

Let’s see the synchronization status and other data of the array:

cat /proc/mdstat

If errors appear due to a bad disk, I recommend that you look at SMART and check it as I wrote in these articles:
Diagnostics HDD using smartmontools
Linux disk test for errors and broken sectors

See also:
How to fix the problem with mdadm disks

Join the Conversation

2 Comments

Leave a Reply

  1. I think that Ubuntu 18 stores the swap file on the raid array. According to this man page, it will always cause the mismatch_cnt
    to be > 0 manpages.ubuntu.com/manpages/cosmic/man4/md.4.html