[TriLUG] Problem with LVM

T. Bryan tbryan at python.net
Sun Apr 30 23:23:40 EDT 2006


On Sunday 30 April 2006 04:12 pm, Rick DeNatale wrote:

> Well here's some more information on my setup.  I've got my root fs
> and a swap "volume" as logical volumes on a raid1 array
>
> $ cat /proc/mdstat
> Personalities : [raid1]
> md0 : active raid1 sda5[0] sdb5[1]
>       8634816 blocks [2/2] [UU]
>
> unused devices: <none>
>
> $ sudo lvmdiskscan -v
>   /dev/md0  [        8.23 GB] LVM physical volume

Thanks for the comparison.  That's pretty much what I expected to see.  I just 
wish that I had run and captured the output from these commands when it 
looked like things were working.

> I'm a little suspicious about those lines in your output:
> >   /dev/hde1 [      111.79 GB] LVM physical volume
> >   Found duplicate PV 7LOs90S2Ff8W7Tq8ZGRsT41lphbExsLh: using /dev/hdg1
> > not /dev/hde1
> >   /dev/hdg1 [      111.79 GB] LVM physical volume

Me, too.  :)

> To me this looks like, instead of building a raid array with two
> physical partitions and then making that a PV, you tried to make a
> raid array from two LVM PVs.

That's not what I attempted.  My journal says that I set this up with 

mke2fs -j /dev/hde1
mke2fs -j /dev/hdg1
mdadm --create --verbose /dev/md0 --level=1 \
      --raid-devices=2 /dev/hde1 /dev/hdg1
mdadm -S /dev/md0
mdadm -As /dev/md0

pvcreate /dev/md0
vgcreate localvg /dev/md0
lvcreate -L20G -nlv_local1 localvg
mke2fs -j /dev/localvg/lv_local1

I wasn't sure whether I needed to mke2fs both on the underlying device and on 
the logical volumes, but it seemed to work well for a while, and I was 
reading and writing data to the logical volume.  It would be nice to get it 
working again.  :)

> My setup has LVM on top of raid.  I know that way works, I'm not sure
> whether or not raid on top of LVM does.

Another question.  What is your partition type for /dev/sda5 and /dev/sdb5?

None of the HOWTOs mentioned changing it, but I notice that partition type 8e 
is Linux LVM, but mine is 83 (Linux) for ext3.

---Tom



More information about the TriLUG mailing list