Gluster and backup storage on rackspace01 / rackspace02

Alexander Rydekull rydekull at gmail.com
Mon Sep 16 13:08:17 UTC 2013


So, it's up and running now.

Next step would be to configure ovirt to use the new gluster volume.

And for backups to be moved from linode to this area instead.

Please note I only allocated 1 TB to glusterlv, and 100 GB to backuplv,
This leave ~700 GB untouched for allocation where we need it.

"It's better to grow then to show" ... I just thought of that one, I know
its bad. Sorry :-D

---

## The below commands are run on each machine, until otherwise is stated

# find wwid of sdb1
multipath -l

# add wwid of sdb1 to multipath.conf to blacklist it
cat >> /etc/multipath.conf <<EOF
blacklist {
      wwid 36848f690e6d9480019ac01600496584f
}
EOF

# reload multipath
multipath -F
multipath -v2

# Create a volume group so that we can administrate storage more efficiently
vgcreate datavg /dev/sdb1

# One for gluster, one for backup
lvcreate -L 1T -n glusterlv datavg
lvcreate -L 100G -n backuplv datavg

# Create filesystems on both
mkfs.ext4 /dev/disk/by-id/dm-name-datavg-backuplv
mkfs.ext4 /dev/disk/by-id/dm-name-datavg-glusterlv

# Create directories to mount at
mkdir /srv/gluster /srv/backup

# Find UUID of new filesystems
blkid

# Fix so that they mount on boot.
cat >> /etc/fstab <<EOF
UUID="69b0c4e5-ded7-4fc9-aa6f-03f6cc4f60c2" /srv/gluster ext4 defaults 1 2
UUID="4bae10e7-0d8e-477c-aa08-15f885bc52bd" /srv/backup ext4 defaults 1 2
EOF

# Mount it!
mount -a

# Start gluster
service glusterd start

# Enable on boot
chkconfig glusterd on

# Add other node
gluster peer probe rackspace02.ovirt org

# Verify
gluster peer status


## Only execute on one node
gluster volume create vmstorage replica 2 transport tcp
rackspace01.ovirt.org:/srv/gluster rackspace02.ovirt.org:/srv/gluster
gluster volume start vmstorage


-- 
/Alexander Rydekull
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ovirt.org/pipermail/infra/attachments/20130916/ff256a31/attachment.html>


More information about the Infra mailing list