The central storage can export a brick for replication to each of the N servers. Each of the N servers can have afr of a local brick and a brick exported by the server.
A sample configuration can be,
Central storage,
volume brick-1
type storage/posix
option directory /storage/1
end-volume
volume brick-2
type storage/posix
option directory /storage/2
end-volume
.
.
.
.
.
volume brick-n
type storage/posix
option directory /storage/n
end-volume
volume server
type protocol/server
option transport-type socket
subvolumes brick-1 brick-2 . . . brick-n
auth.ip.brick-1.allow <server-1 ip address>
auth.ip.brick-2.allow <server-2 ip address>
.
.
.
auth.ip.brick-n.allow <server-n ip address>
end-volume
Each of the server can have glusterfs client as,
volume local-storage
type storage/posix
option directory /storage/local/
end-volume
volume client
type protocol/client
option transport-type socket
option remote-host <ip address of Central storage>
option remote-subvolume <name of the brick exported by the central storage corresponding to this server>
end-volume
you can experiment with this configuration along with performance translators.
regards,
On Tue, Nov 11, 2008 at 6:43 AM, Serge Aleynikov <saleyn@xxxxxxxxx> wrote:
Hi,
I am new to this list, and would like to find out if GlusterFS can serve
the following need.
I have N servers running some services that log to local disk on each server and I need the content of these local logs to be replicated in "close-to-real-time" to some central storage to be accessible by other tools.
Can I use GlusterFS for this purpose? If so, could you point me to a
sample configuration setup?
Thanks.
Serge
_______________________________________________
Gluster-devel mailing list
Gluster-devel@xxxxxxxxxx
http://lists.nongnu.org/mailman/listinfo/gluster-devel
--
Raghavendra G
