Hello! >> On Fri, Jan 14, 2011 at 3:24 PM, Amar Tumballi <amar at gluster.com >> <mailto:amar at gluster.com>> wrote: >>> With GlusterFS 3.1 and it's cool management frontend I didn't >>> found a >>> way to realize such a setup without hacking the volfile. >>> *hint-to-the-developers* :-) >> On server side in 3.1.x version you can give option like below: >> option transport-type socket,rdma >> in protocol/server volume definition in configuration file (inside >> /etc/glusterd/vols/<VOLNAME>/) On 02.03.11 14:08, Amar Tumballi wrote: > This issue was tracked > @ http://bugs.gluster.com/cgi-bin/bugzilla3/show_bug.cgi?id=2294 and is > fixed in master branch of git. This should be available in next minor > release (ie, 3.1.3). I did some checks of this feature. The gluster CLI accepts the new combined transport option: # gluster volume create space transport tcp,rdma server:/brick # gluster volume start space Afterwards I'm able to mount the volume using the defaults which results in a mount over TCP. But I was not able to mount it using RDMA. I tried the appropriate option in the mount command: # mount -t glusterfs -o transport=rdma server:/space /space This results in a mount using TCP. Forcing a mount using the "old way" didn't succeed either. I copied the generated space-fuse.vol, changed the transport to RDMA and mounted it manually: # mount -t glusterfs /etc/glusterd/vols/space/space-rdma-fuse.vol /space Once I saw an error that the client used a wrong protocol, usually it simply reports a connection and a disconnection. Looks like you have some work in the server and client area. At least NFS looks now stable, so I'm currently mounting the Gluster servers on Ethernet only nodes by using NFS. Beat -- \|/ Beat Rubischon <beat at 0x1b.ch> ( 0-0 ) http://www.0x1b.ch/~beat/ oOO--(_)--OOo--------------------------------------------------- Meine Erlebnisse, Gedanken und Traeume: http://www.0x1b.ch/blog/
