| 1 |
On Wed, 07 May 2014 20:57:29 +0200 J. Roeleveld wrote: |
| 2 |
> On 7 May 2014 20:11:10 CEST, Walter Dnes <waltdnes@××××××××.org> |
| 3 |
> wrote: |
| 4 |
> >On Wed, May 07, 2014 at 08:11:02AM +0200, J. Roeleveld wrote |
| 5 |
> >> On Tuesday, May 06, 2014 05:34:52 PM Walter Dnes wrote: |
| 6 |
> >> > |
| 7 |
> >> > Unfortunately, mdev != udev. People running RAID have |
| 8 |
> >> > problems |
| 9 |
> >too. |
| 10 |
> >> |
| 11 |
> >> I know it isn't. I just find it strange that LVM can't work |
| 12 |
> >> without udev when I see options which configure the LVM-tools |
| 13 |
> >> to either double-check udevs actions or even completely bypass |
| 14 |
> >> udev: |
| 15 |
> > |
| 16 |
> > Thanks for the pointer. After turning off the udev-related |
| 17 |
> > options in |
| 18 |
> >lvm.conf, I'm getting /dev/mapper device nodes as expected. |
| 19 |
> |
| 20 |
> That is good. Now if only mdadm can be confirmed to work with |
| 21 |
> mdev. I could try it on one of my machines. |
| 22 |
|
| 23 |
And what is the problem with mdadm with mdev? I have such setup: |
| 24 |
nothing special here and works fine. |
| 25 |
|
| 26 |
Just to speed up device lookup: |
| 27 |
|
| 28 |
$ grep -v ^# /etc/mdadm.conf |
| 29 |
DEVICE /dev/sd* |
| 30 |
|
| 31 |
And here we go: |
| 32 |
|
| 33 |
$ cat /proc/mdstat |
| 34 |
Personalities : [raid10] |
| 35 |
md0 : active raid10 sdd[3] sdf[2] |
| 36 |
2930265088 blocks super 1.2 256K chunks 2 far-copies [2/2] [UU] |
| 37 |
|
| 38 |
Best regards, |
| 39 |
Andrew Savchenko |
Archives