You are not logged in.
Hi There,
I wanted to add a new logical volume to an existing volume group so I ran :-
lvcreate -n SambaShares -L 50G MainVG
It warned me that btrfs fs was detected so I stopped right there to figure things out..
# blkid /dev/MainVG/ArchRoot
/dev/MainVG/ArchRoot : UUID="0532ffd7-dc9e-4ec4-bce2-8ac21f121f12" UUID_SUB="90497162-1ffe-4a13-9c93-f32f4fd0ef98" TYPE="btrfs"
# blkid /dev/MainVG/SambaShares
/dev/MainVG/SambaShares: UUID="0532ffd7-dc9e-4ec4-bce2-8ac21f121f12" UUID_SUB="90497162-1ffe-4a13-9c93-f32f4fd0ef98" TYPE="btrfs"
The newly created logical volume points to my root!
I tried mounting the SambaShares filesystem somewhere else and verifying a new file I just wrote to the real root.
The file was present on SambaShares mount.
They are the same physical filesytem.
Any ideas what to do now? I dare not remove the new volume in case it wipes my root.
lvs output
--- Logical volume ---
LV Path /dev/MainVG/ArchRoot
LV Name ArchRoot
VG Name MainVG
LV UUID 2cVb2C-3jDT-1yGl-AM5h-BRac-BcLe-3IQoso
LV Write Access read/write
LV Creation host, time ,
LV Status available
# open 1
LV Size 50.00 GiB
Current LE 800
Segments 1
Allocation inherit
Read ahead sectors auto
- currently set to 256
Block device 253:11
--- Logical volume ---
LV Path /dev/MainVG/SambaShares
LV Name SambaShares
VG Name MainVG
LV UUID akgEB9-IsUf-Xnzw-qRnq-HrvE-buva-ObYwFV
LV Write Access read/write
LV Creation host, time storage-pc, 2015-02-11 21:56:22 +1100
LV Status available
# open 0
LV Size 50.00 GiB
Current LE 800
Segments 1
Allocation inherit
Read ahead sectors auto
- currently set to 768
Block device 253:15
/etc/lvm/backup
ArchRoot {
id = "2cVb2C-3jDT-1yGl-AM5h-BRac-BcLe-3IQoso"
status = ["READ", "WRITE", "VISIBLE"]
flags = []
segment_count = 1
segment1 {
start_extent = 0
extent_count = 800 # 50 Gigabytes
type = "striped"
stripe_count = 1 # linear
stripes = [
"pv3", 0
]
}
}
SambaShares {
id = "akgEB9-IsUf-Xnzw-qRnq-HrvE-buva-ObYwFV"
status = ["READ", "WRITE", "VISIBLE"]
flags = []
creation_host = "storage-pc"
creation_time = 1423652182 # 2015-02-11 21:56:22 +1100
segment_count = 1
segment1 {
start_extent = 0
extent_count = 800 # 50 Gigabytes
type = "striped"
stripe_count = 1 # linear
stripes = [
"pv0", 29664
]
}
Thanks,
John
Last edited by Cerynon (2015-02-12 04:21:44)
Offline
In general, it's not entirely odd for a UUID or filesystem to show up on a newly created LV.
When you delete a LV the data on it is not wiped (except maybe on SSD); you create a new LV, old data may show up.
According to your lvm backup the LVs are allocated in different places (ArchRoot on pv3:0, SambaShares on pv0:29664). This may be presented in a more readable fashion using `pvs -v --segments`, `dmsetup table` or similar. If these LV were indeed physically identical, chances are they'd show up identical in there too.
Now, btrfs has its own layer of device colemongery; so maybe something is happening there, if you [previously] joined that new LV into your btrfs collective? Unfortunately I'm not familiar enough with btrfs to help in that case.
Offline
Thanks frostschutz,
They certainly look different physically
#pvs -v --segments
PV VG Fmt Attr PSize PFree Start SSize LV Start
/dev/md3 MainVG lvm2 a-- 908.19g 358.19g 0 800 ArchRoot 0 linear /dev/md3:0-799
/dev/md0 MainVG lvm2 a-- 2.05t 109.38g 29664 800 SambaShares 0 linear /dev/md0:29664-30463
#dmsetup table
MainVG-ArchRoot: 0 104857600 linear 9:3 2048
MainVG-SambaShares: 0 104857600 linear 9:0 3888121856
When I try to mount the filesystems using /dev/mapper names 'cat /proc/mounts'
swaps between showing /dev/mapper/MainVG-ArchRoot & /dev/mapper/MainVG-SambaShares as root fs.
I think the system is getting confused somewhere, so I can't mount both of them to verify they are in fact different.
I just use btrfs as a basic filesystem within lvm volumes on top of software raid. I don't use the the raid features within btrfs.
Last edited by Cerynon (2015-02-11 13:20:44)
Offline
Okay; LVM-wise I don't see a problem here. If you can also rule out problems on the md layer (cat /proc/mdstat, no overlapping use of physical devices there?), that leaves btrfs. So you have to find a way to make btrfs spill the beans how it sees those devices. You should put btrfs in your thread title to get some btrfs-savvy people in here.
Offline
I am feeling more confident it's just a ghost image from a previous lvm pvmove operation.
Google tells me Btrfs won't let you change uuid as its referenced internally. It's also a bad idea to have 'raw'/dd copies of the btrfs filesystem on the same machine!
I will do a rsync copy of the entire root to a fresh volume as a backup.
Then will lay down a new (ext4!) filesystem on the SambaShare volume. Edit* I just removed the logical volume and recreated a fresh one. Everything is working now!
/proc/mdstat looks good, nothing got mixed up there.
Thanks for your help in verifying this.
Last edited by Cerynon (2015-02-12 04:23:00)
Offline