On 5/29/2014 10:10 PM, Shawn Edwards wrote: > Yes, I know I can try this, but I wanted to know if anyone else has > tried it or if it is known to not work because of some architectural issue. Back when I tried over a year ago, the reads were spread "fairly" over the 4 nodes (3 ro client machines, 1 rw machine) and the overall performance was much worse than having 3 clients read from 1 nfs mount. I don't think that's changed. You may also need kernel 3.16 (sic). Dima