I have two RAID1 drives with multiple partitions.

Each time I reboot, one of the partitions comes up degraded with only one of
the two devices. The only thing that the kernel logs is:


[ 13.327731] md/raid1:md3: active with 1 out of 2 mirrors

I can hot-add the other partition, and wait for mdadm to resync it, but on
the next reboot it's degraded again. This is getting old.


I note that both drives have md0 through md3. There are no defects on either
drive. And it's always md3 that fails to be assembled fully, when the kernel
boots.


--
users mailing list
users@lists.fedoraproject.org
To unsubscribe or change subscription options:
https://admin.fedoraproject.org/mailman/listinfo/users
Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines
Have a question? Ask away: http://ask.fedoraproject.org

> I have two RAID1 drives with multiple partitions.
>
> Each time I reboot, one of the partitions comes up degraded with only one
> of
> the two devices. The only thing that the kernel logs is:
>
> [ 13.327731] md/raid1:md3: active with 1 out of 2 mirrors
>
> I can hot-add the other partition, and wait for mdadm to resync it, but on
> the next reboot it's degraded again. This is getting old.
>
> I note that both drives have md0 through md3. There are no defects on
> either
> drive. And it's always md3 that fails to be assembled fully, when the
> kernel
> boots.

Can you check the following

1) what does /etc/mdadm.conf say for /dev/md3? it should have a line like:

ARRAY /dev/md3 level=raid1 num-devices=2
UUID=1bcb5496:c7d158e6:14c0bbc3:de3ed72e

2) what does "blkid | grep sd[ab]x" say? (x is the partition number) both
should have the same UUID number:

/dev/sdb3: UUID="1bcb5496-c7d1-58e6-14c0-bbc3de3ed72e"
TYPE="linux_raid_member"
/dev/sda3: UUID="1bcb5496-c7d1-58e6-14c0-bbc3de3ed72e"
TYPE="linux_raid_member"

3) what does /etc/fstab show for your boot (root?) partition? should be:
UUID=61a11dc6-b7b7-4fb0-893c-83ce87be9324 / ext4 defaults
1 1

4) what does "blkid | grep md3" show? should be like:
/dev/md3: UUID="61a11dc6-b7b7-4fb0-893c-83ce87be9324" TYPE="ext4"




notice the UUID's match for items 1 & 2, they also match for 3 & 4

mdadm.conf says build the array out of the partitions with the devices
with UUID 1bcb5496:c7d158e6:14c0bbc3:de3ed72e, which in my example are
/dev/sda3 and /dev/sdb3

Line 3 says that the / (root) partition is on UUID device
61a11dc6-b7b7-4fb0-893c-83ce87be9324 this UUID is the RAID device UUID,
make sure it is not pointing to the UUID of the disk.

line 4 shows the UUID for the md3 partition.

Hopefully its this simple and not something else going on.
Jeff



--
users mailing list
users@lists.fedoraproject.org
To unsubscribe or change subscription options:
https://admin.fedoraproject.org/mailman/listinfo/users
Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines
Have a question? Ask away: http://ask.fedoraproject.org

jeff@bubble.org writes:


> I have two RAID1 drives with multiple partitions.
>
> Each time I reboot, one of the partitions comes up degraded with only one
> of
> the two devices. The only thing that the kernel logs is:
>
> [ 13.327731] md/raid1:md3: active with 1 out of 2 mirrors
>
> I can hot-add the other partition, and wait for mdadm to resync it, but on
> the next reboot it's degraded again. This is getting old.
>
> I note that both drives have md0 through md3. There are no defects on
> either
> drive. And it's always md3 that fails to be assembled fully, when the
> kernel
> boots.

Can you check the following

1) what does /etc/mdadm.conf say for /dev/md3? it should have a line like:

ARRAY /dev/md3 level=raid1 num-devices=2
UUID=1bcb5496:c7d158e6:14c0bbc3:de3ed72e


Check, it does.

ARRAY /dev/md3 level=raid1 num-devices=2 UUID=ac2446aa:
661a4977:0960a296:6d0b48c3




2) what does "blkid | grep sd[ab]x" say? (x is the partition number) both
should have the same UUID number:

/dev/sdb3: UUID="1bcb5496-c7d1-58e6-14c0-bbc3de3ed72e"
TYPE="linux_raid_member"
/dev/sda3: UUID="1bcb5496-c7d1-58e6-14c0-bbc3de3ed72e"
TYPE="linux_raid_member"


Same UUID in my case:

/dev/sda5: UUID="ac2446aa-661a-4977-0960-a2966d0b48c3" UUID_SUB="e58c691d-
c8ee-d563-564c-8f9a5f69c987" LABEL="monster:3" TYPE="linux_raid_member"
/dev/sdb5: UUID="ac2446aa-661a-4977-0960-a2966d0b48c3"
UUID_SUB="59e68522-6594-779a-89ac-6ad85b603614" LABEL="monster:3"
TYPE="linux_raid_member"



3) what does /etc/fstab show for your boot (root?) partition? should be:
UUID=61a11dc6-b7b7-4fb0-893c-83ce87be9324 / ext4 defaults
1 1


It's my /home with a wonky raid, not /.

UUID=a38dc534-f0e9-4d43-8675-8052d5b6f900 /home ext3 defaults 1 3


4) what does "blkid | grep md3" show? should be like:
/dev/md3: UUID="61a11dc6-b7b7-4fb0-893c-83ce87be9324" TYPE="ext4"


/dev/md3: UUID="a38dc534-f0e9-4d43-8675-8052d5b6f900" TYPE="ext3"


notice the UUID's match for items 1 & 2, they also match for 3 & 4


Yes, everything matches.


Hopefully its this simple and not something else going on.


Nah, it's not as simple as that. Something else is going on.

I do see that my problematic md's uuid is not included in the list of uuids
passed to the kernel by grub, in /etc/default/grub. It's still getting
mounted, well half of it does.


I'm going to add it to /etc/default/grub, rebuild grub2.conf, and see what
happens.


--
users mailing list
users@lists.fedoraproject.org
To unsubscribe or change subscription options:
https://admin.fedoraproject.org/mailman/listinfo/users
Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines
Have a question? Ask away: http://ask.fedoraproject.org

Sam Varshavchik pÃ*Å¡e v St 14. 12. 2011 v 18:41 -0500:
> jeff@bubble.org writes:

False raid detection is often caused by wrong partition type.
It suppose to be

fd "Linux raid autodetect"

What do you have?
fdisk -l /dev/sda
or
fdisk -l /dev/sdb

Pavel

--
Pavel Lisy <pali@tmapy.cz>
T-MAPY spol. s r.o.

--
users mailing list
users@lists.fedoraproject.org
To unsubscribe or change subscription options:
https://admin.fedoraproject.org/mailman/listinfo/users
Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines
Have a question? Ask away: http://ask.fedoraproject.org

Pavel Lisy writes:


Sam Varshavchik pÃ*Å¡e v St 14. 12. 2011 v 18:41 -0500:
> jeff@bubble.org writes:

False raid detection is often caused by wrong partition type.
It suppose to be

fd "Linux raid autodetect"

What do you have?
fdisk -l /dev/sda
or
fdisk -l /dev/sdb


What it should be. My partition types are correct.

--
users mailing list
users@lists.fedoraproject.org
To unsubscribe or change subscription options:
https://admin.fedoraproject.org/mailman/listinfo/users
Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines
Have a question? Ask away: http://ask.fedoraproject.org