Hi Krishna,
I'll clarify a bit by saying the main purpose of having cluster/unify
plus AFR is to have an expandable grid file system but have
redundancy on it too. I don't mind having to restart the daemons or
remount the glusterfs volume.
I have the funds to build a 20 node glusterfs cluster and I was
thinking that 20 machines each with 2x750Gig SATA drives in cluster/
unify would yield around 25-30 or so TB of storage. An EMC san of
comparable size will cost us way more than this grid. If you add the
AFR portion, and say replicate *:4 , that's like saying you can take
down 3 servers and still be guaranteed to have a node in the cluster
with the file you're looking for.
Please let me know if I'm wrong, and with that in mind, can you
provide an example of how to do this on a cluster of 4 servers to
start ?
Thanks, and awesome project!
Pooya
On Mar 11, 2007, at 1:24 PM, Krishna Srinivas wrote:
Pooya,
If you change the client.vol you have to restart the glusterfs.
If you change the server.vol you have to restart the glusterfsd.
It is possible not to restart glusterfs but still grow the
glusterfs volume.
i.e by using an extra glusterfsd which does unify which communicates
with the glusterfs. you can change the configs for this glusterfsd
and restart it. However glusterfs will not work properly till you
start
this glusterfsd process, but you dont have to restart the glusterfs
process.
Let me know if I need to explain by example.
Krishna
On 3/10/07, Pooya Woodcock <pooya@xxxxxxxxxxxxxxx> wrote:
I just had a chance to experiment with GlusterFS this evening and one
question came up. I am wondering if it is possible to combine types
cluster/unify and AFR such that you have a mounted glusterfs volume
that grows in size and also has *:2 on its files, for example.
Thanks,
Pooya
_______________________________________________
Gluster-devel mailing list
Gluster-devel@xxxxxxxxxx
http://lists.nongnu.org/mailman/listinfo/gluster-devel
