Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
Josh Aune <ladros at ...> writes: > > On Sun, Mar 14, 2010 at 2:48 PM, Patrick Zwahlen <paz at ...> wrote: > > As far as I can tell, even a cluster FS (OCFS2, GFS) wouldn't help. > > Not sure about your particular needs, but years ago I did build a SAN > backed GFS cluster where each node served up NFS exports of the shared > GFS mounts. It isn't a preferred solution, but it did work. > Ya, I think it might work as long as the exports are different on each nodes. In that case, it might be more efficient to have 2 DRBD ressources attached to each one of the two nodes, and a special pacemaker/hb RA that can handle add/remove from /etc/exports. The problem I can see will be to remove a mount point from /etc/exports when there are open files. If you failover the entire NFS server, you can easily 'kill -9' the server in order to stop it. Regards, - Patrick -