Hello Chris, Thank you for your response. I could solve the problem by extending the lv as Eric suggested. Regards, Utpal Bora Ph.D. Scholar Computer Science & Engineering IIT Hyderabad http://utpalbora.com PGP key fingerprint: 2F12 635E 409F 11AC 1502 BB41 7705 9980 A062 FA70 On Sun, Dec 29, 2019 at 8:45 AM Chris Murphy <lists@xxxxxxxxxxxxxxxxx> wrote: > > On Sat, Dec 28, 2019 at 4:12 AM Utpal Bora <cs14mtech11017@xxxxxxxxxx> wrote: > > > > Hi, > > > > My XFS home drive is corrupt after trying to extend it with lvm. > > This is what I did to extend the partition. > > 1. Extend Volume group to use a new physical volume of around 1.2TB. > > This was successful without any error. > > vgextend vg-1 /dev/sdc1 > > > > 2. Extend logical volume (home-lv) to use the free space. > > lvextend -l 100%FREE /dev/mapper/vg--1-home--lv -r > > What approximate byte value is 100%FREE ? > > > 3. Resized home-lv and reduce 55 GB > > lvreduce -L 55G /dev/mapper/vg--1-home--lv -r > > If this is really a volume reduction, along with -r I would expect > this entire command to fail. XFS doesn't support shrink. Since a > successful LV shrink requires shrinking the file system first, or else > it results in truncation of the file system it contains and thus > damages it, off hand I think this is a bug in lvreduce, or possibly in > fsadm which is what -r calls to do the resize. > > My suggestion is to make no further changes at all, no further > recovery attempts. And head over to the LVM list and make the same > post as above. It's very possible there's backup LVM metadata that > will reference the PE/LE's used in the LV prior to the lvreduce. By > restoring the LV using that exact same mapping, it should be possible > to recover the file system it contains. But only if you don't make > other attempts. The more additional attempts the greater the chance of > irreparable changes. > > https://www.redhat.com/mailman/listinfo/linux-lvm > > > I assumed that -r will invoke xfs_grow internally. > > Right, but xfs_grow only grows. It doesn't shrink. So my expectation > is that the lvreduce command should fail without having made any > changes. And yet... > > -- > Chris Murphy
