Gluster users - Feb 2015 - Using glusterfs to create distributed disk on web server

Hi there.

I had a web server instance, hosted in Digital Ocean cloud now. CentOS 6
x86_64. There is a low disk space on this instance and I want to move
static files to glusterfs pool. I plan to dynamically add new instances
when will be a need for it. For example, buying new instance, add it into
glusterfs pool, thats it, we automatically had more disk space for static
files.

A new schema first consists of two servers (bricks), one of them is current
web server.
Assume that current static files directory is /var/www/static. What i want
to do:

On current server stop all http and database services.. then
mv /var/www/static /var/www/static-brick

Then combine current_server:/var/www/static-brick with
new_server:/glusterfs-brick-on-root-partition into one glusterfs vol0 (in
distributed mode). Pay attention that I want to create one brick from
current static files directory, that already filled with files.

Then on current server: mount current_server:/vol0 /var/www/static

A new glusterfs volume that filled up with current static files and allow
distribution to new server after old server will be fill.

Some questions in that scheme:
Because my instances are on Digital Ocean, I cannot allocate separate
volumes to bricks. Just create directory and point glusterfs to it as a
brick. How I must limit glusterfs with writing new files into brick to
avoid overfill root partition on the instance? For example, I want to stop
writing on old server when 10 Gb free space will be achieved? After that
all new files must be written to new server. And again, when new server
reaches some limit, I buying new server, include it into gluster pool and
continue work with more disk space?
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://www.gluster.org/pipermail/gluster-users/attachments/20150206/95b7e662/attachment.html>