September 7, 2015

Setting up GlusterFS on CentOS 6.7 x64 on DigitalOcean

Posted on September 7, 2015  •  3 minutes  • 550 words  •  Suggest Changes

There are already allot of tutorials on the digitalocean website, including one on glusterFS on Ubuntu. I however wanted to test with CentOS before deploying GlusterFS local, while the method described by glusterfs themself refers to CentOS & RHEL variants, it actually doesn’t work. This is a fresh two node installation from start till working replication on CentOS 6.7 64bit.

Like all virtual machines, I start off with a clean slate and an up-to-date machine.

# do updates
# install yum-cron (auto updater) and nano (editor)
yum update -y && yum install yum-cron nano -y

# set timezone
 ln -sf /usr/share/zoneinfo/Europe/Brussels /etc/localtime

After that you need to add the repo data of glusterfs, as the installed repo’s don’t contain glusterfs-server (they do contain other glusterfs library’s but that one is missing.)

# add as /etc/yum.repo.d/gluster.repo

name=GlusterFS is a clustered file-system capable of scaling to several petabytes.

name=GlusterFS is a clustered file-system capable of scaling to several petabytes.

name=GlusterFS is a clustered file-system capable of scaling to several petabytes. - Source


After that you can start installing.

# you need epel, this should install the correct one
yum install epel-release

# since we need to load the data first update
# your local cache, then install gluster-server
# this will include all the other dependencies but 
# will take the latest version from the repo we just added (glusterfs repo)
yum clean all && yum update && yum install glusterfs-server

# start rpcbind & glusterfs on boot
chkconfig rpcbind on
chkconfig glusterd on

# start the services now
/etc/init.d/glusterd start
/etc/init.d/rpcbind start

notice : failing to start rpcbind, will cause NFS not being able to load. The result is that everything work until you mount NFS and start writing files in it. (they don’t get replicated) The error you would get is : (cat /var/log/glusterfs/nfs.log)

0-/usr/sbin/glusterfs: Started running 
0-epoll: Started thread with index 1
0-rpc-service: Configured rpc.outstanding-rpc-limit with value 16
0-nfs-mount: Exports auth has been disabled!
0-rpc-service: Could not register with portmap 100005 3 38465
0-nfs: Required program  MOUNT3 registration failed
0-nfs: Failed to initialize protocols
0-nfs-server: Initialization of volume 'nfs-server' failed, review your volfile again
0-nfs-server: initializing translator failed
0-graph: init failed

Next I added the both machines to each /etc/hosts (as this is a test setup and I don’t want to wait for DNS update)

[email protected]/etc/hosts first_machine first_machine second_machine second_machine

[email protected]/etc/hosts second_machine second_machine first_machine first_machine


The is needed as I used the names of the machines and otherwise you get a _gluster peer probe: failed: Probe returned with unknown errno 107 ([also possible](" target="blank)).

After that its pretty much following the tutorial : (from second_machine)

# connect them to the new trusted pool
gluster peer probe first_machine

# create local directory (if you use disks this would be the mountpoint)
mkdir -p /data/my_gluster_share

# create a replication volume of 2 bricks
# this also created /data/my_guster_share on the first_machine (make sure its empty!)
gluster volume create gv0 replica 2 first_machine:/data/my_gluster_share second_machine:/data/my_gluster_share

# start the volume
gluster volume start gv0

# check status (all columns should be Y, if not something is not working!)
gluster volume status

Hope this helps someone, up to our next adventure!


If you enjoyed this website, consider buying me a Dr. Pepper

Buy me a Dr PepperBuy me a Dr Pepper