[LU-3167] The MDT volume label tokenizer is not consistent

• Type: Bug
• Resolution: Not a Bug
• Priority: Minor
• Affects Version/s: Lustre 2.3.0
• Environment: 2.6.32-279.5.1.el6_lustre.gb16fe80.x86_64
• Severity: 3
• Assignee: WC Triage
• Reporter: Zhiqi Tao (Inactive)

Description

When specifying --index in the mkfs.lustre command, an indexed volume label is automatically assigned, for example "regal:MDT0000". The colon sign, ":", is the tokenizer.

[root@regal-mds00 ~]# mkfs.lustre --mdt --mgsnode=10.10.187.20@o2ib --mgsnode=10.10.187.30@o2ib --failnode=10.10.187.30@o2ib --fsname=regal --index=0 --verbose --reformat /dev/mapper/mpathb

Permanent disk data:
Target: regal:MDT0000
Index: 0
Lustre FS: regal
Mount type: ldiskfs
Flags: 0x61
(MDT first_time update )
Persistent mount opts: user_xattr,errors=remount-ro
Parameters: mgsnode=10.10.187.20@o2ib mgsnode=10.10.187.30@o2ib failover.node=10.10.187.30@o2ib

device size = 2855900MB
formatting backing filesystem ldiskfs on /dev/mapper/mpathb
target name regal:MDT0000
4k blocks 731110400
options -J size=400 -I 512 -i 2048 -O dirdata,uninit_bg,^extents,mmp,dir_nlink,huge_file,flex_bg -E lazy_journal_init -F
mkfs_cmd = mke2fs -j -b 4096 -L regal:MDT0000 -J size=400 -I 512 -i 2048 -O dirdata,uninit_bg,^extents,mmp,dir_nlink,huge_file,flex_bg -E lazy_journal_init -F /dev/mapper/mpathb 731110400
cmd: mke2fs -j -b 4096 -L regal:MDT0000 -J size=400 -I 512 -i 2048 -O dirdata,uninit_bg,^extents,mmp,dir_nlink,huge_file,flex_bg -E lazy_journal_init -F /dev/mapper/mpathb 731110400
mke2fs 1.42.6.wc2 (10-Dec-2012)
Filesystem label=regal:MDT0000
OS type: Linux
Block size=4096 (log=2)
Fragment size=4096 (log=2)
Stride=0 blocks, Stripe width=0 blocks
1462239232 inodes, 731110400 blocks
36555520 blocks (5.00%) reserved for the super user
First data block=0
Maximum filesystem blocks=2879389696
44624 block groups
16384 blocks per group, 16384 fragments per group
32768 inodes per group
Superblock backups stored on blocks:
16384, 49152, 81920, 114688, 147456, 409600, 442368, 802816, 1327104,
2048000, 3981312, 5619712, 10240000, 11943936, 35831808, 39337984,
51200000, 107495424, 256000000, 275365888, 322486272

Allocating group tables: done
Writing inode tables: done
Creating journal (102400 blocks): done
Multiple mount protection is enabled with update interval 5 seconds.
Writing superblocks and filesystem accounting information: done

Writing CONFIGS/mountdata

[root@regal-mds00 ~]# blkid /dev/mapper/mpathb
/dev/mapper/mpathb: LABEL="regal:MDT0000" UUID="41f5335a-aa76-46ba-b168-94db8aadbc0f" TYPE="ext4"

However, the tokenizer of the MDT volume label is changed to dash "-" after the first mount.

[root@regal-mds00 ~]# mount -t lustre /dev/mapper/mpathb /mnt/mdt0

[root@regal-mds00 ~]# blkid /dev/mapper/mpathb
/dev/mapper/mpathb: LABEL="regal-MDT0000" UUID="41f5335a-aa76-46ba-b168-94db8aadbc0f" TYPE="ext4"

It would be nice if the tokenizer could be consistent. The volume label is often used when mounting multipath devices.

Issue Links

• is related to: LU-3006 failover nids added after format don't propogate to clients

Comments

[Andreas Dilger]

This is related to LU-3006, which is potentially adding further overloading of the filesystem label.

Zhiqi, the long-term label will be of the form

{fsname}

-MDTnnnn, and given that the initial formatting of the filesystem will require the actual device name, it doesn't seem unreasonable that the first mount should use the device name as well...