Preparing a Debian VPS for GlusterFS

This document describes the manual steps needed to prepare a Debian VPS for GlusterFS.

Create a new VPS with Debian and select the manual installation option. During manual installation, at the disk partitioning step be sure to select "Guided - use entire disk and set up LVM". Then boot the VM in a live Linux rescue mode so that the root volume is not mounted.

Display information about logical volumes:

lvs

You should have a volume group containing your hostname (e.g. helium--vg) and a logical volume called "root". This volume contains the operating system. We will create a new logical volume for GlusterFS data.

Force fsck to check the filesystem on the "root" logical volume:

export VG=helium
fsck -f /dev/mapper/${VG}--vg-root

Reduce the filesystem:

resize2fs /dev/mapper/${VG}--vg-root 60G

Reduce the size of the logical volume:

lvreduce -L 60G /dev/mapper/${VG}--vg-root

Create a logical volume for GlusterFS:

lvcreate -n gluster -L 180G ${VG}-vg

Format the logical volume to the XFS filesystem:

mkfs.xfs -i size=512 /dev/mapper/${VG}--vg-gluster

You can test to see if it's working:

mount /dev/mapper/${VG}--vg-gluster /mnt
df -h

Reboot the system into Debian.

Some useful GlusterFS commands:

# List all the nodes in the pool
sudo gluster pool list

# List information of all volumes
sudo gluster volume info

# Display status of volumes
sudo gluster volume status

# Enable self-heal-daemon for volume
sudo gluster volume heal vol1 enable

# Displays the count of files to be healed
sudo gluster volume heal vol1 statistics heal-count

# Mount a volume
sudo mount -t glusterfs localhost:/vol1 /mnt