[LU-11923] meta_bg conflicts with resize_inode if explicitly specified for a small filesystem

• Type: Improvement
• Resolution: Fixed
• Priority: Minor
• Fix Version/s: Lustre 2.13.0
• Affects Version/s: Lustre 2.12.0, Lustre 2.10.6
• Assignee: Andreas Dilger
• Reporter: Andreas Dilger

Description

If the meta_bg feature is explicitly specified for a filesystem smaller than 16TB, it causes a format error because the resize_inode feature is automatically enabled by /etc/mke2fs.conf for such filesystems:

mkfsoptions='-O meta_bg -E lazy_itable_init=0,lazy_journal_init=0 -J size=1024' --reformat --backfstype=ldiskfs /dev/sdc

device size = 14614528MB
formatting backing filesystem ldiskfs on /dev/sdc
        target name   testfs:OST0005
        4k blocks     3741319168
        options         -J size=1024 -I 512 -i 524288 -q -O meta_bg,extents,uninit_bg,mmp,dir_nlink,quota,huge_file,flex_bg -G 256 -E lazy_itable_init=0,lazy_journal_init=0 -F
mkfs_cmd = mke2fs -j -b 4096 -L scrlfs:OST0005   -J size=1024 -I 512 -i 524288 -q -O meta_bg,extents,uninit_bg,mmp,dir_nlink,quota,huge_file,flex_bg -G 256 -E lazy_itable_init=0,lazy_journal_init=0 -F /dev/sdc 3741319168
   The resize_inode and meta_bg features are not compatible.
   They can not be both enabled simultaneously.

Ideally this should be handled automatically in mke2fs itself, but it is also prudent to fix mkfs.lustre to pass ^resize_inode so that it also works with versions of mke2fs that do not have this fixed.

The meta_bg feature should also be enabled automatically for filesystems over 256TiB.