Short version: I resized the partitions with fdisk, but the software RAID failed to reassemble from the new partitions. I'm looking for a diagnoosis or repair.
On Tue, 2012-08-28 at 17:15 -0700, Ross Boylan wrote: > A kvm virtual machine has two disks provided by LVM logical volumes. > They are identical; each has 3 partitions. To get more space I used > lvextend to grow the logical volumes. > > My problem is that the partition tables are unchanged; the last > partition should extend to the end of the new disk but it does not. > None of the tools I have used to change that will cooperate. > > How can I expand the partition? > > Inside the VM the disks are used like this > LVM > software-raid > partition > drive. > > Partition 3 on each virtual disk is software RAID1 with the other. The > resulting device, /dev/md1, is used for an LVM volume group. > > The virtual disks were 6G and now are 8G. > [deleted failues with cfdisk, parted, and gparted]. Despite the warnings, I used fdisk: turned off DOS compatibility switch to sectors as unit remove partition 3 create a new partition 3, accepting defaults so it starts where it did before and ends at the new end of the disk. set the mode of the partition to fd write results to disk. When I booted up there were warnings about a duplicate UUID, I think for the LVM VG. These were the result of failure of /dev/md1 to be created out of /dev/hda3 and /dev/hdb3. It brought up one of them for the volume group, and the system appears to operate normally. mdadm -E /dev/hda1 shows it is part of a raid device (it is; there should be md0 and md1 from partitions 1 and 3 respectively), but # mdadm -E /dev/hda3 mdadm: No md superblock detected on /dev/hda3. cfdisk does show the mode is fd "Linux raid autodetect". However, some tools complain that the partition does not end on a cylinder boundary. I'm guessing either something was tucked into the partition table that was overwritten, or perhaps the superblock in question is at the end of the partition. Since the end of the parition is new, that would explain the problem. Or maybe my initrd needs to be updated? /etc/mdadm/mdadm.conf still has entries for both RAIDs. I'd appreciate some help with the diagnosis and repair. Thanks. Ross -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/1346216987.26275.88.ca...@corn.betterworld.us
