LVM Logical Volume Geometry Size Problem

Error Messages Observed

The following messages appear in dmesg after an unsuccessful activation of a volume groups logical volumes

device-mapper: table: <major>:<minor>: <block device> too small for target: start=<sector>, len=<sector>, dev_size=<size>

kernel: EXT4-fs (dm-9): bad geometry: block count <number> exceeds size of device (<number> blocks)

The following messages appear after a vgimport on a volume group

device-mapper: resume ioctl on failed: Invalid argument

Unable to resume <volume group>-<logical volume> (253:23)

Fix Procedure

1. Deactivate the volume group's logical volumes

   # vgchange -an <volume_group>

2. Rescan all SCSI Host Adapters.

   # echo '- - -' > /sys/class/scsi_host/host<#>/scan

3. Rescan all SCSI block devices associated with the volume group

   # echo 1 > /sys/block/<device>/device/rescan

4. Resize all physical volumes associated with the volume group

   # pvscan

   # pvresize /dev/<device>

5. Activate the volume group's logical volumes

   # vgchange -ay <volume_group>