LVM issue

circus78 · 12-09-2014, 05:06 PM

Hi,
I manage a CentOS server with soft-raid (two 80 GB sata disks).

root partition is on a LVM device.

I noticed this messages in /var/log/messages:

Code:

# dmesg |grep sda
sd 2:0:0:0: [sda] 156301488 512-byte logical blocks: (80.0 GB/74.5 GiB)
sd 2:0:0:0: [sda] Write Protect is off
sd 2:0:0:0: [sda] Mode Sense: 00 3a 00 00
sd 2:0:0:0: [sda] Write cache: disabled, read cache: enabled, doesn't support DPO or FUA
 sda:
 sda1 sda2
sd 2:0:0:0: [sda] Attached SCSI disk
dracut: Scanning devices sda2 sdb2  for LVM logical volumes vg_tokyo/lv_root vg_tokyo/lv_swap
dracut: Found duplicate PV G6N7tAm9vRcl5eiM4Voo4FBKZmQVde8Z: using /dev/sdb2 not /dev/sda2
dracut: Found duplicate PV G6N7tAm9vRcl5eiM4Voo4FBKZmQVde8Z: using /dev/sdb2 not /dev/sda2

# dmesg |grep sdb
sd 3:0:0:0: [sdb] 156301488 512-byte logical blocks: (80.0 GB/74.5 GiB)
sd 3:0:0:0: [sdb] Write Protect is off
sd 3:0:0:0: [sdb] Mode Sense: 00 3a 00 00
sd 3:0:0:0: [sdb] Write cache: disabled, read cache: enabled, doesn't support DPO or FUA
 sdb: sdb1 sdb2
sd 3:0:0:0: [sdb] Attached SCSI disk
dracut: Scanning devices sda2 sdb2  for LVM logical volumes vg_tokyo/lv_root vg_tokyo/lv_swap
dracut: Found duplicate PV G6N7tAm9vRcl5eiM4Voo4FBKZmQVde8Z: using /dev/sdb2 not /dev/sda2
dracut: Found duplicate PV G6N7tAm9vRcl5eiM4Voo4FBKZmQVde8Z: using /dev/sdb2 not /dev/sda2
EXT4-fs (sdb1): mounted filesystem with ordered data mode

What can it mean?
Thankyou!

Ser Olmy · 12-09-2014, 06:05 PM

It means exactly what it says: On your system, there are two partitions, /dev/sda2 and /dev/sdb2, with the same Physical Volume UUID.

UUIDs are supposed to be globally unique, so normally this should never happen. It could be the result of the /dev/sda2 partition being duplicated onto /dev/sdb (or the other way around) with dd or a similar tool.

propofol · 12-09-2014, 10:21 PM

Use 'blkid' to see what the current labels & UUIDs are. Modify the UUID for sdb2 or sda2 with:

Quote:

sudo tune2fs -U <new uuid> /dev/sdb2

Ser Olmy · 12-10-2014, 02:59 AM

Quote:

Originally Posted by propofol

Use 'blkid' to see what the current labels & UUIDs are. Modify the UUID for sdb2 or sda2 with:

Quote:

sudo tune2fs -U <new uuid> /dev/sdb2

That would change the file system UUID on an ext2/3/4 file system, not the UUID of an LVM Physical Volume.

And since a PV contains metadata (and usually no directly recognizable file system), running tune2fs or any other file system utility on a PV device is very likely to cause data corruption. I suspect tune2fs would detect that the device doesn't contain a valid file system, though.

rknichols · 12-10-2014, 09:56 AM

It looks to me as though the system failed to detect the RAID array and, finding two identical two identical disks, arbitrarily selected sdb2. You need to find out why the RAID detect failed. I'm no expert there -- sorry.

(I see those repeated words. Not correcting. They are just "two" apt.)

propofol · 12-12-2014, 02:32 AM

Quote:

Originally Posted by Ser Olmy

That would change the file system UUID on an ext2/3/4 file system, not the UUID of an LVM Physical Volume.
...

Thanks for the clarification - I was not aware of this.